1 //====- X86InstrSSE.td - Describe the X86 Instruction Set --*- tablegen -*-===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This file describes the X86 SSE instruction set, defining the instructions,
11 // and properties of the instructions which are needed for code generation,
12 // machine code emission, and analysis.
14 //===----------------------------------------------------------------------===//
17 //===----------------------------------------------------------------------===//
18 // SSE scalar FP Instructions
19 //===----------------------------------------------------------------------===//
21 // CMOV* - Used to implement the SSE SELECT DAG operation. Expanded after
22 // instruction selection into a branch sequence.
23 let Uses = [EFLAGS], usesCustomInserter = 1 in {
24 def CMOV_FR32 : I<0, Pseudo,
25 (outs FR32:$dst), (ins FR32:$t, FR32:$f, i8imm:$cond),
27 [(set FR32:$dst, (X86cmov FR32:$t, FR32:$f, imm:$cond,
29 def CMOV_FR64 : I<0, Pseudo,
30 (outs FR64:$dst), (ins FR64:$t, FR64:$f, i8imm:$cond),
32 [(set FR64:$dst, (X86cmov FR64:$t, FR64:$f, imm:$cond,
34 def CMOV_V4F32 : I<0, Pseudo,
35 (outs VR128:$dst), (ins VR128:$t, VR128:$f, i8imm:$cond),
36 "#CMOV_V4F32 PSEUDO!",
38 (v4f32 (X86cmov VR128:$t, VR128:$f, imm:$cond,
40 def CMOV_V2F64 : I<0, Pseudo,
41 (outs VR128:$dst), (ins VR128:$t, VR128:$f, i8imm:$cond),
42 "#CMOV_V2F64 PSEUDO!",
44 (v2f64 (X86cmov VR128:$t, VR128:$f, imm:$cond,
46 def CMOV_V2I64 : I<0, Pseudo,
47 (outs VR128:$dst), (ins VR128:$t, VR128:$f, i8imm:$cond),
48 "#CMOV_V2I64 PSEUDO!",
50 (v2i64 (X86cmov VR128:$t, VR128:$f, imm:$cond,
54 //===----------------------------------------------------------------------===//
55 // SSE 1 & 2 Instructions Classes
56 //===----------------------------------------------------------------------===//
58 /// sse12_fp_scalar - SSE 1 & 2 scalar instructions class
59 multiclass sse12_fp_scalar<bits<8> opc, string OpcodeStr, SDNode OpNode,
60 RegisterClass RC, X86MemOperand x86memop,
62 let isCommutable = 1 in {
63 def rr : SI<opc, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, RC:$src2),
65 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
66 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
67 [(set RC:$dst, (OpNode RC:$src1, RC:$src2))]>;
69 def rm : SI<opc, MRMSrcMem, (outs RC:$dst), (ins RC:$src1, x86memop:$src2),
71 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
72 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
73 [(set RC:$dst, (OpNode RC:$src1, (load addr:$src2)))]>;
76 /// sse12_fp_scalar_int - SSE 1 & 2 scalar instructions intrinsics class
77 multiclass sse12_fp_scalar_int<bits<8> opc, string OpcodeStr, RegisterClass RC,
78 string asm, string SSEVer, string FPSizeStr,
79 Operand memopr, ComplexPattern mem_cpat,
81 def rr_Int : SI<opc, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, RC:$src2),
83 !strconcat(asm, "\t{$src2, $dst|$dst, $src2}"),
84 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
85 [(set RC:$dst, (!nameconcat<Intrinsic>("int_x86_sse",
86 !strconcat(SSEVer, !strconcat("_",
87 !strconcat(OpcodeStr, FPSizeStr))))
88 RC:$src1, RC:$src2))]>;
89 def rm_Int : SI<opc, MRMSrcMem, (outs RC:$dst), (ins RC:$src1, memopr:$src2),
91 !strconcat(asm, "\t{$src2, $dst|$dst, $src2}"),
92 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
93 [(set RC:$dst, (!nameconcat<Intrinsic>("int_x86_sse",
94 !strconcat(SSEVer, !strconcat("_",
95 !strconcat(OpcodeStr, FPSizeStr))))
96 RC:$src1, mem_cpat:$src2))]>;
99 /// sse12_fp_packed - SSE 1 & 2 packed instructions class
100 multiclass sse12_fp_packed<bits<8> opc, string OpcodeStr, SDNode OpNode,
101 RegisterClass RC, ValueType vt,
102 X86MemOperand x86memop, PatFrag mem_frag,
103 Domain d, bit Is2Addr = 1> {
104 let isCommutable = 1 in
105 def rr : PI<opc, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, RC:$src2),
107 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
108 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
109 [(set RC:$dst, (vt (OpNode RC:$src1, RC:$src2)))], d>;
111 def rm : PI<opc, MRMSrcMem, (outs RC:$dst), (ins RC:$src1, x86memop:$src2),
113 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
114 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
115 [(set RC:$dst, (OpNode RC:$src1, (mem_frag addr:$src2)))], d>;
118 /// sse12_fp_packed_logical_rm - SSE 1 & 2 packed instructions class
119 multiclass sse12_fp_packed_logical_rm<bits<8> opc, RegisterClass RC, Domain d,
120 string OpcodeStr, X86MemOperand x86memop,
121 list<dag> pat_rr, list<dag> pat_rm,
123 let isCommutable = 1 in
124 def rr : PI<opc, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, RC:$src2),
126 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
127 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
129 def rm : PI<opc, MRMSrcMem, (outs RC:$dst), (ins RC:$src1, x86memop:$src2),
131 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
132 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
136 /// sse12_fp_packed_int - SSE 1 & 2 packed instructions intrinsics class
137 multiclass sse12_fp_packed_int<bits<8> opc, string OpcodeStr, RegisterClass RC,
138 string asm, string SSEVer, string FPSizeStr,
139 X86MemOperand x86memop, PatFrag mem_frag,
140 Domain d, bit Is2Addr = 1> {
141 def rr_Int : PI<opc, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, RC:$src2),
143 !strconcat(asm, "\t{$src2, $dst|$dst, $src2}"),
144 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
145 [(set RC:$dst, (!nameconcat<Intrinsic>("int_x86_",
146 !strconcat(SSEVer, !strconcat("_",
147 !strconcat(OpcodeStr, FPSizeStr))))
148 RC:$src1, RC:$src2))], d>;
149 def rm_Int : PI<opc, MRMSrcMem, (outs RC:$dst), (ins RC:$src1,x86memop:$src2),
151 !strconcat(asm, "\t{$src2, $dst|$dst, $src2}"),
152 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
153 [(set RC:$dst, (!nameconcat<Intrinsic>("int_x86_",
154 !strconcat(SSEVer, !strconcat("_",
155 !strconcat(OpcodeStr, FPSizeStr))))
156 RC:$src1, (mem_frag addr:$src2)))], d>;
159 //===----------------------------------------------------------------------===//
160 // SSE 1 & 2 - Move Instructions
161 //===----------------------------------------------------------------------===//
163 class sse12_move_rr<RegisterClass RC, ValueType vt, string asm> :
164 SI<0x10, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src1, RC:$src2), asm,
165 [(set (vt VR128:$dst), (movl VR128:$src1, (scalar_to_vector RC:$src2)))]>;
167 // Loading from memory automatically zeroing upper bits.
168 class sse12_move_rm<RegisterClass RC, X86MemOperand x86memop,
169 PatFrag mem_pat, string OpcodeStr> :
170 SI<0x10, MRMSrcMem, (outs RC:$dst), (ins x86memop:$src),
171 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
172 [(set RC:$dst, (mem_pat addr:$src))]>;
174 // Move Instructions. Register-to-register movss/movsd is not used for FR32/64
175 // register copies because it's a partial register update; FsMOVAPSrr/FsMOVAPDrr
176 // is used instead. Register-to-register movss/movsd is not modeled as an
177 // INSERT_SUBREG because INSERT_SUBREG requires that the insert be implementable
178 // in terms of a copy, and just mentioned, we don't use movss/movsd for copies.
179 let isAsmParserOnly = 1 in {
180 def VMOVSSrr : sse12_move_rr<FR32, v4f32,
181 "movss\t{$src2, $src1, $dst|$dst, $src1, $src2}">, XS, VEX_4V;
182 def VMOVSDrr : sse12_move_rr<FR64, v2f64,
183 "movsd\t{$src2, $src1, $dst|$dst, $src1, $src2}">, XD, VEX_4V;
185 let canFoldAsLoad = 1, isReMaterializable = 1 in {
186 def VMOVSSrm : sse12_move_rm<FR32, f32mem, loadf32, "movss">, XS, VEX;
188 let AddedComplexity = 20 in
189 def VMOVSDrm : sse12_move_rm<FR64, f64mem, loadf64, "movsd">, XD, VEX;
193 let Constraints = "$src1 = $dst" in {
194 def MOVSSrr : sse12_move_rr<FR32, v4f32,
195 "movss\t{$src2, $dst|$dst, $src2}">, XS;
196 def MOVSDrr : sse12_move_rr<FR64, v2f64,
197 "movsd\t{$src2, $dst|$dst, $src2}">, XD;
200 let canFoldAsLoad = 1, isReMaterializable = 1 in {
201 def MOVSSrm : sse12_move_rm<FR32, f32mem, loadf32, "movss">, XS;
203 let AddedComplexity = 20 in
204 def MOVSDrm : sse12_move_rm<FR64, f64mem, loadf64, "movsd">, XD;
207 let AddedComplexity = 15 in {
208 // Extract the low 32-bit value from one vector and insert it into another.
209 def : Pat<(v4f32 (movl VR128:$src1, VR128:$src2)),
210 (MOVSSrr (v4f32 VR128:$src1),
211 (EXTRACT_SUBREG (v4f32 VR128:$src2), sub_ss))>;
212 // Extract the low 64-bit value from one vector and insert it into another.
213 def : Pat<(v2f64 (movl VR128:$src1, VR128:$src2)),
214 (MOVSDrr (v2f64 VR128:$src1),
215 (EXTRACT_SUBREG (v2f64 VR128:$src2), sub_sd))>;
218 // Implicitly promote a 32-bit scalar to a vector.
219 def : Pat<(v4f32 (scalar_to_vector FR32:$src)),
220 (INSERT_SUBREG (v4f32 (IMPLICIT_DEF)), FR32:$src, sub_ss)>;
221 // Implicitly promote a 64-bit scalar to a vector.
222 def : Pat<(v2f64 (scalar_to_vector FR64:$src)),
223 (INSERT_SUBREG (v2f64 (IMPLICIT_DEF)), FR64:$src, sub_sd)>;
225 let AddedComplexity = 20 in {
226 // MOVSSrm zeros the high parts of the register; represent this
227 // with SUBREG_TO_REG.
228 def : Pat<(v4f32 (X86vzmovl (v4f32 (scalar_to_vector (loadf32 addr:$src))))),
229 (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), sub_ss)>;
230 def : Pat<(v4f32 (scalar_to_vector (loadf32 addr:$src))),
231 (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), sub_ss)>;
232 def : Pat<(v4f32 (X86vzmovl (loadv4f32 addr:$src))),
233 (SUBREG_TO_REG (i32 0), (MOVSSrm addr:$src), sub_ss)>;
234 // MOVSDrm zeros the high parts of the register; represent this
235 // with SUBREG_TO_REG.
236 def : Pat<(v2f64 (X86vzmovl (v2f64 (scalar_to_vector (loadf64 addr:$src))))),
237 (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), sub_sd)>;
238 def : Pat<(v2f64 (scalar_to_vector (loadf64 addr:$src))),
239 (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), sub_sd)>;
240 def : Pat<(v2f64 (X86vzmovl (loadv2f64 addr:$src))),
241 (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), sub_sd)>;
242 def : Pat<(v2f64 (X86vzmovl (bc_v2f64 (loadv4f32 addr:$src)))),
243 (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), sub_sd)>;
244 def : Pat<(v2f64 (X86vzload addr:$src)),
245 (SUBREG_TO_REG (i64 0), (MOVSDrm addr:$src), sub_sd)>;
248 // Store scalar value to memory.
249 def MOVSSmr : SSI<0x11, MRMDestMem, (outs), (ins f32mem:$dst, FR32:$src),
250 "movss\t{$src, $dst|$dst, $src}",
251 [(store FR32:$src, addr:$dst)]>;
252 def MOVSDmr : SDI<0x11, MRMDestMem, (outs), (ins f64mem:$dst, FR64:$src),
253 "movsd\t{$src, $dst|$dst, $src}",
254 [(store FR64:$src, addr:$dst)]>;
256 let isAsmParserOnly = 1 in {
257 def VMOVSSmr : SI<0x11, MRMDestMem, (outs), (ins f32mem:$dst, FR32:$src),
258 "movss\t{$src, $dst|$dst, $src}",
259 [(store FR32:$src, addr:$dst)]>, XS, VEX;
260 def VMOVSDmr : SI<0x11, MRMDestMem, (outs), (ins f64mem:$dst, FR64:$src),
261 "movsd\t{$src, $dst|$dst, $src}",
262 [(store FR64:$src, addr:$dst)]>, XD, VEX;
265 // Extract and store.
266 def : Pat<(store (f32 (vector_extract (v4f32 VR128:$src), (iPTR 0))),
269 (EXTRACT_SUBREG (v4f32 VR128:$src), sub_ss))>;
270 def : Pat<(store (f64 (vector_extract (v2f64 VR128:$src), (iPTR 0))),
273 (EXTRACT_SUBREG (v2f64 VR128:$src), sub_sd))>;
275 // Move Aligned/Unaligned floating point values
276 multiclass sse12_mov_packed<bits<8> opc, RegisterClass RC,
277 X86MemOperand x86memop, PatFrag ld_frag,
278 string asm, Domain d,
279 bit IsReMaterializable = 1> {
280 let neverHasSideEffects = 1 in
281 def rr : PI<opc, MRMSrcReg, (outs RC:$dst), (ins RC:$src),
282 !strconcat(asm, "\t{$src, $dst|$dst, $src}"), [], d>;
283 let canFoldAsLoad = 1, isReMaterializable = IsReMaterializable in
284 def rm : PI<opc, MRMSrcMem, (outs RC:$dst), (ins x86memop:$src),
285 !strconcat(asm, "\t{$src, $dst|$dst, $src}"),
286 [(set RC:$dst, (ld_frag addr:$src))], d>;
289 let isAsmParserOnly = 1 in {
290 defm VMOVAPS : sse12_mov_packed<0x28, VR128, f128mem, alignedloadv4f32,
291 "movaps", SSEPackedSingle>, VEX;
292 defm VMOVAPD : sse12_mov_packed<0x28, VR128, f128mem, alignedloadv2f64,
293 "movapd", SSEPackedDouble>, OpSize, VEX;
294 defm VMOVUPS : sse12_mov_packed<0x10, VR128, f128mem, loadv4f32,
295 "movups", SSEPackedSingle>, VEX;
296 defm VMOVUPD : sse12_mov_packed<0x10, VR128, f128mem, loadv2f64,
297 "movupd", SSEPackedDouble, 0>, OpSize, VEX;
299 defm VMOVAPSY : sse12_mov_packed<0x28, VR256, f256mem, alignedloadv8f32,
300 "movaps", SSEPackedSingle>, VEX;
301 defm VMOVAPDY : sse12_mov_packed<0x28, VR256, f256mem, alignedloadv4f64,
302 "movapd", SSEPackedDouble>, OpSize, VEX;
303 defm VMOVUPSY : sse12_mov_packed<0x10, VR256, f256mem, loadv8f32,
304 "movups", SSEPackedSingle>, VEX;
305 defm VMOVUPDY : sse12_mov_packed<0x10, VR256, f256mem, loadv4f64,
306 "movupd", SSEPackedDouble, 0>, OpSize, VEX;
308 defm MOVAPS : sse12_mov_packed<0x28, VR128, f128mem, alignedloadv4f32,
309 "movaps", SSEPackedSingle>, TB;
310 defm MOVAPD : sse12_mov_packed<0x28, VR128, f128mem, alignedloadv2f64,
311 "movapd", SSEPackedDouble>, TB, OpSize;
312 defm MOVUPS : sse12_mov_packed<0x10, VR128, f128mem, loadv4f32,
313 "movups", SSEPackedSingle>, TB;
314 defm MOVUPD : sse12_mov_packed<0x10, VR128, f128mem, loadv2f64,
315 "movupd", SSEPackedDouble, 0>, TB, OpSize;
317 let isAsmParserOnly = 1 in {
318 def VMOVAPSmr : VPSI<0x29, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
319 "movaps\t{$src, $dst|$dst, $src}",
320 [(alignedstore (v4f32 VR128:$src), addr:$dst)]>, VEX;
321 def VMOVAPDmr : VPDI<0x29, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
322 "movapd\t{$src, $dst|$dst, $src}",
323 [(alignedstore (v2f64 VR128:$src), addr:$dst)]>, VEX;
324 def VMOVUPSmr : VPSI<0x11, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
325 "movups\t{$src, $dst|$dst, $src}",
326 [(store (v4f32 VR128:$src), addr:$dst)]>, VEX;
327 def VMOVUPDmr : VPDI<0x11, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
328 "movupd\t{$src, $dst|$dst, $src}",
329 [(store (v2f64 VR128:$src), addr:$dst)]>, VEX;
330 def VMOVAPSYmr : VPSI<0x29, MRMDestMem, (outs), (ins f256mem:$dst, VR256:$src),
331 "movaps\t{$src, $dst|$dst, $src}",
332 [(alignedstore (v8f32 VR256:$src), addr:$dst)]>, VEX;
333 def VMOVAPDYmr : VPDI<0x29, MRMDestMem, (outs), (ins f256mem:$dst, VR256:$src),
334 "movapd\t{$src, $dst|$dst, $src}",
335 [(alignedstore (v4f64 VR256:$src), addr:$dst)]>, VEX;
336 def VMOVUPSYmr : VPSI<0x11, MRMDestMem, (outs), (ins f256mem:$dst, VR256:$src),
337 "movups\t{$src, $dst|$dst, $src}",
338 [(store (v8f32 VR256:$src), addr:$dst)]>, VEX;
339 def VMOVUPDYmr : VPDI<0x11, MRMDestMem, (outs), (ins f256mem:$dst, VR256:$src),
340 "movupd\t{$src, $dst|$dst, $src}",
341 [(store (v4f64 VR256:$src), addr:$dst)]>, VEX;
344 def : Pat<(int_x86_avx_loadu_ps_256 addr:$src), (VMOVUPSYrm addr:$src)>;
345 def : Pat<(int_x86_avx_storeu_ps_256 addr:$dst, VR256:$src),
346 (VMOVUPSYmr addr:$dst, VR256:$src)>;
348 def : Pat<(int_x86_avx_loadu_pd_256 addr:$src), (VMOVUPDYrm addr:$src)>;
349 def : Pat<(int_x86_avx_storeu_pd_256 addr:$dst, VR256:$src),
350 (VMOVUPDYmr addr:$dst, VR256:$src)>;
352 def MOVAPSmr : PSI<0x29, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
353 "movaps\t{$src, $dst|$dst, $src}",
354 [(alignedstore (v4f32 VR128:$src), addr:$dst)]>;
355 def MOVAPDmr : PDI<0x29, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
356 "movapd\t{$src, $dst|$dst, $src}",
357 [(alignedstore (v2f64 VR128:$src), addr:$dst)]>;
358 def MOVUPSmr : PSI<0x11, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
359 "movups\t{$src, $dst|$dst, $src}",
360 [(store (v4f32 VR128:$src), addr:$dst)]>;
361 def MOVUPDmr : PDI<0x11, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
362 "movupd\t{$src, $dst|$dst, $src}",
363 [(store (v2f64 VR128:$src), addr:$dst)]>;
365 // Intrinsic forms of MOVUPS/D load and store
366 let isAsmParserOnly = 1 in {
367 let canFoldAsLoad = 1, isReMaterializable = 1 in
368 def VMOVUPSrm_Int : VPSI<0x10, MRMSrcMem, (outs VR128:$dst),
370 "movups\t{$src, $dst|$dst, $src}",
371 [(set VR128:$dst, (int_x86_sse_loadu_ps addr:$src))]>, VEX;
372 def VMOVUPDrm_Int : VPDI<0x10, MRMSrcMem, (outs VR128:$dst),
374 "movupd\t{$src, $dst|$dst, $src}",
375 [(set VR128:$dst, (int_x86_sse2_loadu_pd addr:$src))]>, VEX;
376 def VMOVUPSmr_Int : VPSI<0x11, MRMDestMem, (outs),
377 (ins f128mem:$dst, VR128:$src),
378 "movups\t{$src, $dst|$dst, $src}",
379 [(int_x86_sse_storeu_ps addr:$dst, VR128:$src)]>, VEX;
380 def VMOVUPDmr_Int : VPDI<0x11, MRMDestMem, (outs),
381 (ins f128mem:$dst, VR128:$src),
382 "movupd\t{$src, $dst|$dst, $src}",
383 [(int_x86_sse2_storeu_pd addr:$dst, VR128:$src)]>, VEX;
385 let canFoldAsLoad = 1, isReMaterializable = 1 in
386 def MOVUPSrm_Int : PSI<0x10, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
387 "movups\t{$src, $dst|$dst, $src}",
388 [(set VR128:$dst, (int_x86_sse_loadu_ps addr:$src))]>;
389 def MOVUPDrm_Int : PDI<0x10, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
390 "movupd\t{$src, $dst|$dst, $src}",
391 [(set VR128:$dst, (int_x86_sse2_loadu_pd addr:$src))]>;
393 def MOVUPSmr_Int : PSI<0x11, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
394 "movups\t{$src, $dst|$dst, $src}",
395 [(int_x86_sse_storeu_ps addr:$dst, VR128:$src)]>;
396 def MOVUPDmr_Int : PDI<0x11, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
397 "movupd\t{$src, $dst|$dst, $src}",
398 [(int_x86_sse2_storeu_pd addr:$dst, VR128:$src)]>;
400 // Move Low/High packed floating point values
401 multiclass sse12_mov_hilo_packed<bits<8>opc, RegisterClass RC,
402 PatFrag mov_frag, string base_opc,
404 def PSrm : PI<opc, MRMSrcMem,
405 (outs VR128:$dst), (ins VR128:$src1, f64mem:$src2),
406 !strconcat(!strconcat(base_opc,"s"), asm_opr),
409 (bc_v4f32 (v2f64 (scalar_to_vector (loadf64 addr:$src2))))))],
410 SSEPackedSingle>, TB;
412 def PDrm : PI<opc, MRMSrcMem,
413 (outs RC:$dst), (ins RC:$src1, f64mem:$src2),
414 !strconcat(!strconcat(base_opc,"d"), asm_opr),
415 [(set RC:$dst, (v2f64 (mov_frag RC:$src1,
416 (scalar_to_vector (loadf64 addr:$src2)))))],
417 SSEPackedDouble>, TB, OpSize;
420 let isAsmParserOnly = 1, AddedComplexity = 20 in {
421 defm VMOVL : sse12_mov_hilo_packed<0x12, VR128, movlp, "movlp",
422 "\t{$src2, $src1, $dst|$dst, $src1, $src2}">, VEX_4V;
423 defm VMOVH : sse12_mov_hilo_packed<0x16, VR128, movlhps, "movhp",
424 "\t{$src2, $src1, $dst|$dst, $src1, $src2}">, VEX_4V;
426 let Constraints = "$src1 = $dst", AddedComplexity = 20 in {
427 defm MOVL : sse12_mov_hilo_packed<0x12, VR128, movlp, "movlp",
428 "\t{$src2, $dst|$dst, $src2}">;
429 defm MOVH : sse12_mov_hilo_packed<0x16, VR128, movlhps, "movhp",
430 "\t{$src2, $dst|$dst, $src2}">;
433 let isAsmParserOnly = 1 in {
434 def VMOVLPSmr : VPSI<0x13, MRMDestMem, (outs), (ins f64mem:$dst, VR128:$src),
435 "movlps\t{$src, $dst|$dst, $src}",
436 [(store (f64 (vector_extract (bc_v2f64 (v4f32 VR128:$src)),
437 (iPTR 0))), addr:$dst)]>, VEX;
438 def VMOVLPDmr : VPDI<0x13, MRMDestMem, (outs), (ins f64mem:$dst, VR128:$src),
439 "movlpd\t{$src, $dst|$dst, $src}",
440 [(store (f64 (vector_extract (v2f64 VR128:$src),
441 (iPTR 0))), addr:$dst)]>, VEX;
443 def MOVLPSmr : PSI<0x13, MRMDestMem, (outs), (ins f64mem:$dst, VR128:$src),
444 "movlps\t{$src, $dst|$dst, $src}",
445 [(store (f64 (vector_extract (bc_v2f64 (v4f32 VR128:$src)),
446 (iPTR 0))), addr:$dst)]>;
447 def MOVLPDmr : PDI<0x13, MRMDestMem, (outs), (ins f64mem:$dst, VR128:$src),
448 "movlpd\t{$src, $dst|$dst, $src}",
449 [(store (f64 (vector_extract (v2f64 VR128:$src),
450 (iPTR 0))), addr:$dst)]>;
452 // v2f64 extract element 1 is always custom lowered to unpack high to low
453 // and extract element 0 so the non-store version isn't too horrible.
454 let isAsmParserOnly = 1 in {
455 def VMOVHPSmr : VPSI<0x17, MRMDestMem, (outs), (ins f64mem:$dst, VR128:$src),
456 "movhps\t{$src, $dst|$dst, $src}",
457 [(store (f64 (vector_extract
458 (unpckh (bc_v2f64 (v4f32 VR128:$src)),
459 (undef)), (iPTR 0))), addr:$dst)]>,
461 def VMOVHPDmr : VPDI<0x17, MRMDestMem, (outs), (ins f64mem:$dst, VR128:$src),
462 "movhpd\t{$src, $dst|$dst, $src}",
463 [(store (f64 (vector_extract
464 (v2f64 (unpckh VR128:$src, (undef))),
465 (iPTR 0))), addr:$dst)]>,
468 def MOVHPSmr : PSI<0x17, MRMDestMem, (outs), (ins f64mem:$dst, VR128:$src),
469 "movhps\t{$src, $dst|$dst, $src}",
470 [(store (f64 (vector_extract
471 (unpckh (bc_v2f64 (v4f32 VR128:$src)),
472 (undef)), (iPTR 0))), addr:$dst)]>;
473 def MOVHPDmr : PDI<0x17, MRMDestMem, (outs), (ins f64mem:$dst, VR128:$src),
474 "movhpd\t{$src, $dst|$dst, $src}",
475 [(store (f64 (vector_extract
476 (v2f64 (unpckh VR128:$src, (undef))),
477 (iPTR 0))), addr:$dst)]>;
479 let isAsmParserOnly = 1, AddedComplexity = 20 in {
480 def VMOVLHPSrr : VPSI<0x16, MRMSrcReg, (outs VR128:$dst),
481 (ins VR128:$src1, VR128:$src2),
482 "movlhps\t{$src2, $src1, $dst|$dst, $src1, $src2}",
484 (v4f32 (movlhps VR128:$src1, VR128:$src2)))]>,
486 def VMOVHLPSrr : VPSI<0x12, MRMSrcReg, (outs VR128:$dst),
487 (ins VR128:$src1, VR128:$src2),
488 "movhlps\t{$src2, $src1, $dst|$dst, $src1, $src2}",
490 (v4f32 (movhlps VR128:$src1, VR128:$src2)))]>,
493 let Constraints = "$src1 = $dst", AddedComplexity = 20 in {
494 def MOVLHPSrr : PSI<0x16, MRMSrcReg, (outs VR128:$dst),
495 (ins VR128:$src1, VR128:$src2),
496 "movlhps\t{$src2, $dst|$dst, $src2}",
498 (v4f32 (movlhps VR128:$src1, VR128:$src2)))]>;
499 def MOVHLPSrr : PSI<0x12, MRMSrcReg, (outs VR128:$dst),
500 (ins VR128:$src1, VR128:$src2),
501 "movhlps\t{$src2, $dst|$dst, $src2}",
503 (v4f32 (movhlps VR128:$src1, VR128:$src2)))]>;
506 def : Pat<(movlhps VR128:$src1, (bc_v4i32 (v2i64 (X86vzload addr:$src2)))),
507 (MOVHPSrm (v4i32 VR128:$src1), addr:$src2)>;
508 let AddedComplexity = 20 in {
509 def : Pat<(v4f32 (movddup VR128:$src, (undef))),
510 (MOVLHPSrr (v4f32 VR128:$src), (v4f32 VR128:$src))>;
511 def : Pat<(v2i64 (movddup VR128:$src, (undef))),
512 (MOVLHPSrr (v2i64 VR128:$src), (v2i64 VR128:$src))>;
515 //===----------------------------------------------------------------------===//
516 // SSE 1 & 2 - Conversion Instructions
517 //===----------------------------------------------------------------------===//
519 multiclass sse12_cvt_s<bits<8> opc, RegisterClass SrcRC, RegisterClass DstRC,
520 SDNode OpNode, X86MemOperand x86memop, PatFrag ld_frag,
522 def rr : SI<opc, MRMSrcReg, (outs DstRC:$dst), (ins SrcRC:$src), asm,
523 [(set DstRC:$dst, (OpNode SrcRC:$src))]>;
524 def rm : SI<opc, MRMSrcMem, (outs DstRC:$dst), (ins x86memop:$src), asm,
525 [(set DstRC:$dst, (OpNode (ld_frag addr:$src)))]>;
528 multiclass sse12_cvt_s_np<bits<8> opc, RegisterClass SrcRC, RegisterClass DstRC,
529 X86MemOperand x86memop, string asm> {
530 def rr : SI<opc, MRMSrcReg, (outs DstRC:$dst), (ins SrcRC:$src), asm,
532 def rm : SI<opc, MRMSrcMem, (outs DstRC:$dst), (ins x86memop:$src), asm,
536 multiclass sse12_cvt_p<bits<8> opc, RegisterClass SrcRC, RegisterClass DstRC,
537 SDNode OpNode, X86MemOperand x86memop, PatFrag ld_frag,
538 string asm, Domain d> {
539 def rr : PI<opc, MRMSrcReg, (outs DstRC:$dst), (ins SrcRC:$src), asm,
540 [(set DstRC:$dst, (OpNode SrcRC:$src))], d>;
541 def rm : PI<opc, MRMSrcMem, (outs DstRC:$dst), (ins x86memop:$src), asm,
542 [(set DstRC:$dst, (OpNode (ld_frag addr:$src)))], d>;
545 multiclass sse12_vcvt_avx<bits<8> opc, RegisterClass SrcRC, RegisterClass DstRC,
546 X86MemOperand x86memop, string asm> {
547 def rr : SI<opc, MRMSrcReg, (outs DstRC:$dst), (ins DstRC:$src1, SrcRC:$src),
548 !strconcat(asm,"\t{$src, $src1, $dst|$dst, $src1, $src}"), []>;
549 def rm : SI<opc, MRMSrcMem, (outs DstRC:$dst),
550 (ins DstRC:$src1, x86memop:$src),
551 !strconcat(asm,"\t{$src, $src1, $dst|$dst, $src1, $src}"), []>;
554 let isAsmParserOnly = 1 in {
555 defm VCVTTSS2SI : sse12_cvt_s<0x2C, FR32, GR32, fp_to_sint, f32mem, loadf32,
556 "cvttss2si\t{$src, $dst|$dst, $src}">, XS, VEX;
557 defm VCVTTSS2SI64 : sse12_cvt_s<0x2C, FR32, GR64, fp_to_sint, f32mem, loadf32,
558 "cvttss2si\t{$src, $dst|$dst, $src}">, XS, VEX,
560 defm VCVTTSD2SI : sse12_cvt_s<0x2C, FR64, GR32, fp_to_sint, f64mem, loadf64,
561 "cvttsd2si\t{$src, $dst|$dst, $src}">, XD, VEX;
562 defm VCVTTSD2SI64 : sse12_cvt_s<0x2C, FR64, GR64, fp_to_sint, f64mem, loadf64,
563 "cvttsd2si\t{$src, $dst|$dst, $src}">, XD,
566 // The assembler can recognize rr 64-bit instructions by seeing a rxx
567 // register, but the same isn't true when only using memory operands,
568 // provide other assembly "l" and "q" forms to address this explicitly
569 // where appropriate to do so.
570 defm VCVTSI2SS : sse12_vcvt_avx<0x2A, GR32, FR32, i32mem, "cvtsi2ss">, XS,
572 defm VCVTSI2SS64 : sse12_vcvt_avx<0x2A, GR64, FR32, i64mem, "cvtsi2ss{q}">, XS,
574 defm VCVTSI2SD : sse12_vcvt_avx<0x2A, GR32, FR64, i32mem, "cvtsi2sd">, XD,
576 defm VCVTSI2SDL : sse12_vcvt_avx<0x2A, GR32, FR64, i32mem, "cvtsi2sd{l}">, XD,
578 defm VCVTSI2SD64 : sse12_vcvt_avx<0x2A, GR64, FR64, i64mem, "cvtsi2sd{q}">, XD,
582 defm CVTTSS2SI : sse12_cvt_s<0x2C, FR32, GR32, fp_to_sint, f32mem, loadf32,
583 "cvttss2si\t{$src, $dst|$dst, $src}">, XS;
584 defm CVTTSS2SI64 : sse12_cvt_s<0x2C, FR32, GR64, fp_to_sint, f32mem, loadf32,
585 "cvttss2si{q}\t{$src, $dst|$dst, $src}">, XS, REX_W;
586 defm CVTTSD2SI : sse12_cvt_s<0x2C, FR64, GR32, fp_to_sint, f64mem, loadf64,
587 "cvttsd2si\t{$src, $dst|$dst, $src}">, XD;
588 defm CVTTSD2SI64 : sse12_cvt_s<0x2C, FR64, GR64, fp_to_sint, f64mem, loadf64,
589 "cvttsd2si{q}\t{$src, $dst|$dst, $src}">, XD, REX_W;
590 defm CVTSI2SS : sse12_cvt_s<0x2A, GR32, FR32, sint_to_fp, i32mem, loadi32,
591 "cvtsi2ss\t{$src, $dst|$dst, $src}">, XS;
592 defm CVTSI2SS64 : sse12_cvt_s<0x2A, GR64, FR32, sint_to_fp, i64mem, loadi64,
593 "cvtsi2ss{q}\t{$src, $dst|$dst, $src}">, XS, REX_W;
594 defm CVTSI2SD : sse12_cvt_s<0x2A, GR32, FR64, sint_to_fp, i32mem, loadi32,
595 "cvtsi2sd\t{$src, $dst|$dst, $src}">, XD;
596 defm CVTSI2SD64 : sse12_cvt_s<0x2A, GR64, FR64, sint_to_fp, i64mem, loadi64,
597 "cvtsi2sd{q}\t{$src, $dst|$dst, $src}">, XD, REX_W;
599 // Conversion Instructions Intrinsics - Match intrinsics which expect MM
600 // and/or XMM operand(s).
601 multiclass sse12_cvt_pint<bits<8> opc, RegisterClass SrcRC, RegisterClass DstRC,
602 Intrinsic Int, X86MemOperand x86memop, PatFrag ld_frag,
603 string asm, Domain d> {
604 def rr : PI<opc, MRMSrcReg, (outs DstRC:$dst), (ins SrcRC:$src), asm,
605 [(set DstRC:$dst, (Int SrcRC:$src))], d>;
606 def rm : PI<opc, MRMSrcMem, (outs DstRC:$dst), (ins x86memop:$src), asm,
607 [(set DstRC:$dst, (Int (ld_frag addr:$src)))], d>;
610 multiclass sse12_cvt_sint<bits<8> opc, RegisterClass SrcRC, RegisterClass DstRC,
611 Intrinsic Int, X86MemOperand x86memop, PatFrag ld_frag,
613 def rr : SI<opc, MRMSrcReg, (outs DstRC:$dst), (ins SrcRC:$src),
614 !strconcat(asm, "\t{$src, $dst|$dst, $src}"),
615 [(set DstRC:$dst, (Int SrcRC:$src))]>;
616 def rm : SI<opc, MRMSrcMem, (outs DstRC:$dst), (ins x86memop:$src),
617 !strconcat(asm, "\t{$src, $dst|$dst, $src}"),
618 [(set DstRC:$dst, (Int (ld_frag addr:$src)))]>;
621 multiclass sse12_cvt_pint_3addr<bits<8> opc, RegisterClass SrcRC,
622 RegisterClass DstRC, Intrinsic Int, X86MemOperand x86memop,
623 PatFrag ld_frag, string asm, Domain d> {
624 def rr : PI<opc, MRMSrcReg, (outs DstRC:$dst), (ins DstRC:$src1, SrcRC:$src2),
625 asm, [(set DstRC:$dst, (Int DstRC:$src1, SrcRC:$src2))], d>;
626 def rm : PI<opc, MRMSrcMem, (outs DstRC:$dst),
627 (ins DstRC:$src1, x86memop:$src2), asm,
628 [(set DstRC:$dst, (Int DstRC:$src1, (ld_frag addr:$src2)))], d>;
631 multiclass sse12_cvt_sint_3addr<bits<8> opc, RegisterClass SrcRC,
632 RegisterClass DstRC, Intrinsic Int, X86MemOperand x86memop,
633 PatFrag ld_frag, string asm, bit Is2Addr = 1> {
634 def rr : SI<opc, MRMSrcReg, (outs DstRC:$dst), (ins DstRC:$src1, SrcRC:$src2),
636 !strconcat(asm, "\t{$src2, $dst|$dst, $src2}"),
637 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
638 [(set DstRC:$dst, (Int DstRC:$src1, SrcRC:$src2))]>;
639 def rm : SI<opc, MRMSrcMem, (outs DstRC:$dst),
640 (ins DstRC:$src1, x86memop:$src2),
642 !strconcat(asm, "\t{$src2, $dst|$dst, $src2}"),
643 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
644 [(set DstRC:$dst, (Int DstRC:$src1, (ld_frag addr:$src2)))]>;
647 let isAsmParserOnly = 1 in {
648 defm Int_VCVTSS2SI : sse12_cvt_sint<0x2D, VR128, GR32, int_x86_sse_cvtss2si,
649 f32mem, load, "cvtss2si">, XS, VEX;
650 defm Int_VCVTSS2SI64 : sse12_cvt_sint<0x2D, VR128, GR64,
651 int_x86_sse_cvtss2si64, f32mem, load, "cvtss2si">,
653 defm Int_VCVTSD2SI : sse12_cvt_sint<0x2D, VR128, GR32, int_x86_sse2_cvtsd2si,
654 f128mem, load, "cvtsd2si">, XD, VEX;
655 defm Int_VCVTSD2SI64 : sse12_cvt_sint<0x2D, VR128, GR64,
656 int_x86_sse2_cvtsd2si64, f128mem, load, "cvtsd2si">,
659 // FIXME: The asm matcher has a hack to ignore instructions with _Int and Int_
660 // Get rid of this hack or rename the intrinsics, there are several
661 // intructions that only match with the intrinsic form, why create duplicates
662 // to let them be recognized by the assembler?
663 defm VCVTSD2SI_alt : sse12_cvt_s_np<0x2D, FR64, GR32, f64mem,
664 "cvtsd2si\t{$src, $dst|$dst, $src}">, XD, VEX;
665 defm VCVTSD2SI64 : sse12_cvt_s_np<0x2D, FR64, GR64, f64mem,
666 "cvtsd2si\t{$src, $dst|$dst, $src}">, XD, VEX, VEX_W;
668 defm Int_CVTSS2SI : sse12_cvt_sint<0x2D, VR128, GR32, int_x86_sse_cvtss2si,
669 f32mem, load, "cvtss2si">, XS;
670 defm Int_CVTSS2SI64 : sse12_cvt_sint<0x2D, VR128, GR64, int_x86_sse_cvtss2si64,
671 f32mem, load, "cvtss2si{q}">, XS, REX_W;
672 defm Int_CVTSD2SI : sse12_cvt_sint<0x2D, VR128, GR32, int_x86_sse2_cvtsd2si,
673 f128mem, load, "cvtsd2si">, XD;
674 defm Int_CVTSD2SI64 : sse12_cvt_sint<0x2D, VR128, GR64, int_x86_sse2_cvtsd2si64,
675 f128mem, load, "cvtsd2si">, XD, REX_W;
677 defm CVTSD2SI64 : sse12_cvt_s_np<0x2D, VR128, GR64, f64mem, "cvtsd2si{q}">, XD,
680 let isAsmParserOnly = 1 in {
681 defm Int_VCVTSI2SS : sse12_cvt_sint_3addr<0x2A, GR32, VR128,
682 int_x86_sse_cvtsi2ss, i32mem, loadi32, "cvtsi2ss", 0>, XS, VEX_4V;
683 defm Int_VCVTSI2SS64 : sse12_cvt_sint_3addr<0x2A, GR64, VR128,
684 int_x86_sse_cvtsi642ss, i64mem, loadi64, "cvtsi2ss", 0>, XS, VEX_4V,
686 defm Int_VCVTSI2SD : sse12_cvt_sint_3addr<0x2A, GR32, VR128,
687 int_x86_sse2_cvtsi2sd, i32mem, loadi32, "cvtsi2sd", 0>, XD, VEX_4V;
688 defm Int_VCVTSI2SD64 : sse12_cvt_sint_3addr<0x2A, GR64, VR128,
689 int_x86_sse2_cvtsi642sd, i64mem, loadi64, "cvtsi2sd", 0>, XD,
693 let Constraints = "$src1 = $dst" in {
694 defm Int_CVTSI2SS : sse12_cvt_sint_3addr<0x2A, GR32, VR128,
695 int_x86_sse_cvtsi2ss, i32mem, loadi32,
697 defm Int_CVTSI2SS64 : sse12_cvt_sint_3addr<0x2A, GR64, VR128,
698 int_x86_sse_cvtsi642ss, i64mem, loadi64,
699 "cvtsi2ss{q}">, XS, REX_W;
700 defm Int_CVTSI2SD : sse12_cvt_sint_3addr<0x2A, GR32, VR128,
701 int_x86_sse2_cvtsi2sd, i32mem, loadi32,
703 defm Int_CVTSI2SD64 : sse12_cvt_sint_3addr<0x2A, GR64, VR128,
704 int_x86_sse2_cvtsi642sd, i64mem, loadi64,
705 "cvtsi2sd">, XD, REX_W;
708 // Instructions below don't have an AVX form.
709 defm Int_CVTPS2PI : sse12_cvt_pint<0x2D, VR128, VR64, int_x86_sse_cvtps2pi,
710 f64mem, load, "cvtps2pi\t{$src, $dst|$dst, $src}",
711 SSEPackedSingle>, TB;
712 defm Int_CVTPD2PI : sse12_cvt_pint<0x2D, VR128, VR64, int_x86_sse_cvtpd2pi,
713 f128mem, memop, "cvtpd2pi\t{$src, $dst|$dst, $src}",
714 SSEPackedDouble>, TB, OpSize;
715 defm Int_CVTTPS2PI : sse12_cvt_pint<0x2C, VR128, VR64, int_x86_sse_cvttps2pi,
716 f64mem, load, "cvttps2pi\t{$src, $dst|$dst, $src}",
717 SSEPackedSingle>, TB;
718 defm Int_CVTTPD2PI : sse12_cvt_pint<0x2C, VR128, VR64, int_x86_sse_cvttpd2pi,
719 f128mem, memop, "cvttpd2pi\t{$src, $dst|$dst, $src}",
720 SSEPackedDouble>, TB, OpSize;
721 defm Int_CVTPI2PD : sse12_cvt_pint<0x2A, VR64, VR128, int_x86_sse_cvtpi2pd,
722 i64mem, load, "cvtpi2pd\t{$src, $dst|$dst, $src}",
723 SSEPackedDouble>, TB, OpSize;
724 let Constraints = "$src1 = $dst" in {
725 defm Int_CVTPI2PS : sse12_cvt_pint_3addr<0x2A, VR64, VR128,
726 int_x86_sse_cvtpi2ps,
727 i64mem, load, "cvtpi2ps\t{$src2, $dst|$dst, $src2}",
728 SSEPackedSingle>, TB;
733 // Aliases for intrinsics
734 let isAsmParserOnly = 1 in {
735 defm Int_VCVTTSS2SI : sse12_cvt_sint<0x2C, VR128, GR32, int_x86_sse_cvttss2si,
736 f32mem, load, "cvttss2si">, XS, VEX;
737 defm Int_VCVTTSS2SI64 : sse12_cvt_sint<0x2C, VR128, GR64,
738 int_x86_sse_cvttss2si64, f32mem, load,
739 "cvttss2si">, XS, VEX, VEX_W;
740 defm Int_VCVTTSD2SI : sse12_cvt_sint<0x2C, VR128, GR32, int_x86_sse2_cvttsd2si,
741 f128mem, load, "cvttss2si">, XD, VEX;
742 defm Int_VCVTTSD2SI64 : sse12_cvt_sint<0x2C, VR128, GR64,
743 int_x86_sse2_cvttsd2si64, f128mem, load,
744 "cvttss2si">, XD, VEX, VEX_W;
746 defm Int_CVTTSS2SI : sse12_cvt_sint<0x2C, VR128, GR32, int_x86_sse_cvttss2si,
747 f32mem, load, "cvttss2si">, XS;
748 defm Int_CVTTSS2SI64 : sse12_cvt_sint<0x2C, VR128, GR64,
749 int_x86_sse_cvttss2si64, f32mem, load,
750 "cvttss2si{q}">, XS, REX_W;
751 defm Int_CVTTSD2SI : sse12_cvt_sint<0x2C, VR128, GR32, int_x86_sse2_cvttsd2si,
752 f128mem, load, "cvttss2si">, XD;
753 defm Int_CVTTSD2SI64 : sse12_cvt_sint<0x2C, VR128, GR64,
754 int_x86_sse2_cvttsd2si64, f128mem, load,
755 "cvttss2si{q}">, XD, REX_W;
757 let isAsmParserOnly = 1, Pattern = []<dag> in {
758 defm VCVTSS2SI : sse12_cvt_s<0x2D, FR32, GR32, undef, f32mem, load,
759 "cvtss2si{l}\t{$src, $dst|$dst, $src}">, XS, VEX;
760 defm VCVTSS2SI64 : sse12_cvt_s<0x2D, FR32, GR64, undef, f32mem, load,
761 "cvtss2si\t{$src, $dst|$dst, $src}">, XS, VEX,
763 defm VCVTDQ2PS : sse12_cvt_p<0x5B, VR128, VR128, undef, i128mem, load,
764 "cvtdq2ps\t{$src, $dst|$dst, $src}",
765 SSEPackedSingle>, TB, VEX;
766 defm VCVTDQ2PSY : sse12_cvt_p<0x5B, VR256, VR256, undef, i256mem, load,
767 "cvtdq2ps\t{$src, $dst|$dst, $src}",
768 SSEPackedSingle>, TB, VEX;
770 let Pattern = []<dag> in {
771 defm CVTSS2SI : sse12_cvt_s<0x2D, FR32, GR32, undef, f32mem, load /*dummy*/,
772 "cvtss2si{l}\t{$src, $dst|$dst, $src}">, XS;
773 defm CVTSS2SI64 : sse12_cvt_s<0x2D, FR32, GR64, undef, f32mem, load /*dummy*/,
774 "cvtss2si{q}\t{$src, $dst|$dst, $src}">, XS, REX_W;
775 defm CVTDQ2PS : sse12_cvt_p<0x5B, VR128, VR128, undef, i128mem, load /*dummy*/,
776 "cvtdq2ps\t{$src, $dst|$dst, $src}",
777 SSEPackedSingle>, TB; /* PD SSE3 form is avaiable */
782 // Convert scalar double to scalar single
783 let isAsmParserOnly = 1 in {
784 def VCVTSD2SSrr : VSDI<0x5A, MRMSrcReg, (outs FR32:$dst),
785 (ins FR64:$src1, FR64:$src2),
786 "cvtsd2ss\t{$src2, $src1, $dst|$dst, $src1, $src2}", []>,
788 def VCVTSD2SSrm : I<0x5A, MRMSrcMem, (outs FR32:$dst),
789 (ins FR64:$src1, f64mem:$src2),
790 "vcvtsd2ss\t{$src2, $src1, $dst|$dst, $src1, $src2}",
791 []>, XD, Requires<[HasAVX, OptForSize]>, VEX_4V;
793 def CVTSD2SSrr : SDI<0x5A, MRMSrcReg, (outs FR32:$dst), (ins FR64:$src),
794 "cvtsd2ss\t{$src, $dst|$dst, $src}",
795 [(set FR32:$dst, (fround FR64:$src))]>;
796 def CVTSD2SSrm : I<0x5A, MRMSrcMem, (outs FR32:$dst), (ins f64mem:$src),
797 "cvtsd2ss\t{$src, $dst|$dst, $src}",
798 [(set FR32:$dst, (fround (loadf64 addr:$src)))]>, XD,
799 Requires<[HasSSE2, OptForSize]>;
801 let isAsmParserOnly = 1 in
802 defm Int_VCVTSD2SS: sse12_cvt_sint_3addr<0x5A, VR128, VR128,
803 int_x86_sse2_cvtsd2ss, f64mem, load, "cvtsd2ss", 0>,
805 let Constraints = "$src1 = $dst" in
806 defm Int_CVTSD2SS: sse12_cvt_sint_3addr<0x5A, VR128, VR128,
807 int_x86_sse2_cvtsd2ss, f64mem, load, "cvtsd2ss">, XS;
809 // Convert scalar single to scalar double
810 let isAsmParserOnly = 1 in { // SSE2 instructions with XS prefix
811 def VCVTSS2SDrr : I<0x5A, MRMSrcReg, (outs FR64:$dst),
812 (ins FR32:$src1, FR32:$src2),
813 "vcvtss2sd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
814 []>, XS, Requires<[HasAVX]>, VEX_4V;
815 def VCVTSS2SDrm : I<0x5A, MRMSrcMem, (outs FR64:$dst),
816 (ins FR32:$src1, f32mem:$src2),
817 "vcvtss2sd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
818 []>, XS, VEX_4V, Requires<[HasAVX, OptForSize]>;
820 def CVTSS2SDrr : I<0x5A, MRMSrcReg, (outs FR64:$dst), (ins FR32:$src),
821 "cvtss2sd\t{$src, $dst|$dst, $src}",
822 [(set FR64:$dst, (fextend FR32:$src))]>, XS,
824 def CVTSS2SDrm : I<0x5A, MRMSrcMem, (outs FR64:$dst), (ins f32mem:$src),
825 "cvtss2sd\t{$src, $dst|$dst, $src}",
826 [(set FR64:$dst, (extloadf32 addr:$src))]>, XS,
827 Requires<[HasSSE2, OptForSize]>;
829 let isAsmParserOnly = 1 in {
830 def Int_VCVTSS2SDrr: I<0x5A, MRMSrcReg,
831 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
832 "vcvtss2sd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
833 [(set VR128:$dst, (int_x86_sse2_cvtss2sd VR128:$src1,
834 VR128:$src2))]>, XS, VEX_4V,
836 def Int_VCVTSS2SDrm: I<0x5A, MRMSrcMem,
837 (outs VR128:$dst), (ins VR128:$src1, f32mem:$src2),
838 "vcvtss2sd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
839 [(set VR128:$dst, (int_x86_sse2_cvtss2sd VR128:$src1,
840 (load addr:$src2)))]>, XS, VEX_4V,
843 let Constraints = "$src1 = $dst" in { // SSE2 instructions with XS prefix
844 def Int_CVTSS2SDrr: I<0x5A, MRMSrcReg,
845 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
846 "cvtss2sd\t{$src2, $dst|$dst, $src2}",
847 [(set VR128:$dst, (int_x86_sse2_cvtss2sd VR128:$src1,
850 def Int_CVTSS2SDrm: I<0x5A, MRMSrcMem,
851 (outs VR128:$dst), (ins VR128:$src1, f32mem:$src2),
852 "cvtss2sd\t{$src2, $dst|$dst, $src2}",
853 [(set VR128:$dst, (int_x86_sse2_cvtss2sd VR128:$src1,
854 (load addr:$src2)))]>, XS,
858 def : Pat<(extloadf32 addr:$src),
859 (CVTSS2SDrr (MOVSSrm addr:$src))>,
860 Requires<[HasSSE2, OptForSpeed]>;
862 // Convert doubleword to packed single/double fp
863 let isAsmParserOnly = 1 in { // SSE2 instructions without OpSize prefix
864 def Int_VCVTDQ2PSrr : I<0x5B, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
865 "vcvtdq2ps\t{$src, $dst|$dst, $src}",
866 [(set VR128:$dst, (int_x86_sse2_cvtdq2ps VR128:$src))]>,
867 TB, VEX, Requires<[HasAVX]>;
868 def Int_VCVTDQ2PSrm : I<0x5B, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
869 "vcvtdq2ps\t{$src, $dst|$dst, $src}",
870 [(set VR128:$dst, (int_x86_sse2_cvtdq2ps
871 (bitconvert (memopv2i64 addr:$src))))]>,
872 TB, VEX, Requires<[HasAVX]>;
874 def Int_CVTDQ2PSrr : I<0x5B, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
875 "cvtdq2ps\t{$src, $dst|$dst, $src}",
876 [(set VR128:$dst, (int_x86_sse2_cvtdq2ps VR128:$src))]>,
877 TB, Requires<[HasSSE2]>;
878 def Int_CVTDQ2PSrm : I<0x5B, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
879 "cvtdq2ps\t{$src, $dst|$dst, $src}",
880 [(set VR128:$dst, (int_x86_sse2_cvtdq2ps
881 (bitconvert (memopv2i64 addr:$src))))]>,
882 TB, Requires<[HasSSE2]>;
884 // FIXME: why the non-intrinsic version is described as SSE3?
885 let isAsmParserOnly = 1 in { // SSE2 instructions with XS prefix
886 def Int_VCVTDQ2PDrr : I<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
887 "vcvtdq2pd\t{$src, $dst|$dst, $src}",
888 [(set VR128:$dst, (int_x86_sse2_cvtdq2pd VR128:$src))]>,
889 XS, VEX, Requires<[HasAVX]>;
890 def Int_VCVTDQ2PDrm : I<0xE6, MRMSrcMem, (outs VR128:$dst), (ins i64mem:$src),
891 "vcvtdq2pd\t{$src, $dst|$dst, $src}",
892 [(set VR128:$dst, (int_x86_sse2_cvtdq2pd
893 (bitconvert (memopv2i64 addr:$src))))]>,
894 XS, VEX, Requires<[HasAVX]>;
896 def Int_CVTDQ2PDrr : I<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
897 "cvtdq2pd\t{$src, $dst|$dst, $src}",
898 [(set VR128:$dst, (int_x86_sse2_cvtdq2pd VR128:$src))]>,
899 XS, Requires<[HasSSE2]>;
900 def Int_CVTDQ2PDrm : I<0xE6, MRMSrcMem, (outs VR128:$dst), (ins i64mem:$src),
901 "cvtdq2pd\t{$src, $dst|$dst, $src}",
902 [(set VR128:$dst, (int_x86_sse2_cvtdq2pd
903 (bitconvert (memopv2i64 addr:$src))))]>,
904 XS, Requires<[HasSSE2]>;
907 // Convert packed single/double fp to doubleword
908 let isAsmParserOnly = 1 in {
909 def VCVTPS2DQrr : VPDI<0x5B, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
910 "cvtps2dq\t{$src, $dst|$dst, $src}", []>, VEX;
911 def VCVTPS2DQrm : VPDI<0x5B, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
912 "cvtps2dq\t{$src, $dst|$dst, $src}", []>, VEX;
913 def VCVTPS2DQYrr : VPDI<0x5B, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src),
914 "cvtps2dq\t{$src, $dst|$dst, $src}", []>, VEX;
915 def VCVTPS2DQYrm : VPDI<0x5B, MRMSrcMem, (outs VR256:$dst), (ins f256mem:$src),
916 "cvtps2dq\t{$src, $dst|$dst, $src}", []>, VEX;
918 def CVTPS2DQrr : PDI<0x5B, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
919 "cvtps2dq\t{$src, $dst|$dst, $src}", []>;
920 def CVTPS2DQrm : PDI<0x5B, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
921 "cvtps2dq\t{$src, $dst|$dst, $src}", []>;
923 let isAsmParserOnly = 1 in {
924 def Int_VCVTPS2DQrr : VPDI<0x5B, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
925 "cvtps2dq\t{$src, $dst|$dst, $src}",
926 [(set VR128:$dst, (int_x86_sse2_cvtps2dq VR128:$src))]>,
928 def Int_VCVTPS2DQrm : VPDI<0x5B, MRMSrcMem, (outs VR128:$dst),
930 "cvtps2dq\t{$src, $dst|$dst, $src}",
931 [(set VR128:$dst, (int_x86_sse2_cvtps2dq
932 (memop addr:$src)))]>, VEX;
934 def Int_CVTPS2DQrr : PDI<0x5B, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
935 "cvtps2dq\t{$src, $dst|$dst, $src}",
936 [(set VR128:$dst, (int_x86_sse2_cvtps2dq VR128:$src))]>;
937 def Int_CVTPS2DQrm : PDI<0x5B, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
938 "cvtps2dq\t{$src, $dst|$dst, $src}",
939 [(set VR128:$dst, (int_x86_sse2_cvtps2dq
940 (memop addr:$src)))]>;
942 let isAsmParserOnly = 1 in { // SSE2 packed instructions with XD prefix
943 def Int_VCVTPD2DQrr : I<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
944 "vcvtpd2dq\t{$src, $dst|$dst, $src}",
945 [(set VR128:$dst, (int_x86_sse2_cvtpd2dq VR128:$src))]>,
946 XD, VEX, Requires<[HasAVX]>;
947 def Int_VCVTPD2DQrm : I<0xE6, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
948 "vcvtpd2dq\t{$src, $dst|$dst, $src}",
949 [(set VR128:$dst, (int_x86_sse2_cvtpd2dq
950 (memop addr:$src)))]>,
951 XD, VEX, Requires<[HasAVX]>;
953 def Int_CVTPD2DQrr : I<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
954 "cvtpd2dq\t{$src, $dst|$dst, $src}",
955 [(set VR128:$dst, (int_x86_sse2_cvtpd2dq VR128:$src))]>,
956 XD, Requires<[HasSSE2]>;
957 def Int_CVTPD2DQrm : I<0xE6, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
958 "cvtpd2dq\t{$src, $dst|$dst, $src}",
959 [(set VR128:$dst, (int_x86_sse2_cvtpd2dq
960 (memop addr:$src)))]>,
961 XD, Requires<[HasSSE2]>;
964 // Convert with truncation packed single/double fp to doubleword
965 let isAsmParserOnly = 1 in { // SSE2 packed instructions with XS prefix
966 def VCVTTPS2DQrr : VSSI<0x5B, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
967 "cvttps2dq\t{$src, $dst|$dst, $src}", []>, VEX;
968 def VCVTTPS2DQrm : VSSI<0x5B, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
969 "cvttps2dq\t{$src, $dst|$dst, $src}", []>, VEX;
970 def VCVTTPS2DQYrr : VSSI<0x5B, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src),
971 "cvttps2dq\t{$src, $dst|$dst, $src}", []>, VEX;
972 def VCVTTPS2DQYrm : VSSI<0x5B, MRMSrcMem, (outs VR256:$dst), (ins f256mem:$src),
973 "cvttps2dq\t{$src, $dst|$dst, $src}", []>, VEX;
975 def CVTTPS2DQrr : SSI<0x5B, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
976 "cvttps2dq\t{$src, $dst|$dst, $src}", []>;
977 def CVTTPS2DQrm : SSI<0x5B, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
978 "cvttps2dq\t{$src, $dst|$dst, $src}", []>;
981 let isAsmParserOnly = 1 in {
982 def Int_VCVTTPS2DQrr : I<0x5B, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
983 "vcvttps2dq\t{$src, $dst|$dst, $src}",
985 (int_x86_sse2_cvttps2dq VR128:$src))]>,
986 XS, VEX, Requires<[HasAVX]>;
987 def Int_VCVTTPS2DQrm : I<0x5B, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
988 "vcvttps2dq\t{$src, $dst|$dst, $src}",
989 [(set VR128:$dst, (int_x86_sse2_cvttps2dq
990 (memop addr:$src)))]>,
991 XS, VEX, Requires<[HasAVX]>;
993 def Int_CVTTPS2DQrr : I<0x5B, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
994 "cvttps2dq\t{$src, $dst|$dst, $src}",
996 (int_x86_sse2_cvttps2dq VR128:$src))]>,
997 XS, Requires<[HasSSE2]>;
998 def Int_CVTTPS2DQrm : I<0x5B, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
999 "cvttps2dq\t{$src, $dst|$dst, $src}",
1000 [(set VR128:$dst, (int_x86_sse2_cvttps2dq
1001 (memop addr:$src)))]>,
1002 XS, Requires<[HasSSE2]>;
1004 let isAsmParserOnly = 1 in {
1005 def Int_VCVTTPD2DQrr : VPDI<0xE6, MRMSrcReg, (outs VR128:$dst),
1007 "cvttpd2dq\t{$src, $dst|$dst, $src}",
1008 [(set VR128:$dst, (int_x86_sse2_cvttpd2dq VR128:$src))]>,
1010 def Int_VCVTTPD2DQrm : VPDI<0xE6, MRMSrcMem, (outs VR128:$dst),
1012 "cvttpd2dq\t{$src, $dst|$dst, $src}",
1013 [(set VR128:$dst, (int_x86_sse2_cvttpd2dq
1014 (memop addr:$src)))]>, VEX;
1016 def Int_CVTTPD2DQrr : PDI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1017 "cvttpd2dq\t{$src, $dst|$dst, $src}",
1018 [(set VR128:$dst, (int_x86_sse2_cvttpd2dq VR128:$src))]>;
1019 def Int_CVTTPD2DQrm : PDI<0xE6, MRMSrcMem, (outs VR128:$dst),(ins f128mem:$src),
1020 "cvttpd2dq\t{$src, $dst|$dst, $src}",
1021 [(set VR128:$dst, (int_x86_sse2_cvttpd2dq
1022 (memop addr:$src)))]>;
1024 let isAsmParserOnly = 1 in {
1025 // The assembler can recognize rr 256-bit instructions by seeing a ymm
1026 // register, but the same isn't true when using memory operands instead.
1027 // Provide other assembly rr and rm forms to address this explicitly.
1028 def VCVTTPD2DQrr : VPDI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1029 "cvttpd2dq\t{$src, $dst|$dst, $src}", []>, VEX;
1030 def VCVTTPD2DQXrYr : VPDI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR256:$src),
1031 "cvttpd2dq\t{$src, $dst|$dst, $src}", []>, VEX;
1034 def VCVTTPD2DQXrr : VPDI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1035 "cvttpd2dqx\t{$src, $dst|$dst, $src}", []>, VEX;
1036 def VCVTTPD2DQXrm : VPDI<0xE6, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
1037 "cvttpd2dqx\t{$src, $dst|$dst, $src}", []>, VEX;
1040 def VCVTTPD2DQYrr : VPDI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR256:$src),
1041 "cvttpd2dqy\t{$src, $dst|$dst, $src}", []>, VEX;
1042 def VCVTTPD2DQYrm : VPDI<0xE6, MRMSrcMem, (outs VR128:$dst), (ins f256mem:$src),
1043 "cvttpd2dqy\t{$src, $dst|$dst, $src}", []>, VEX, VEX_L;
1046 // Convert packed single to packed double
1047 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
1048 // SSE2 instructions without OpSize prefix
1049 def VCVTPS2PDrr : I<0x5A, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1050 "vcvtps2pd\t{$src, $dst|$dst, $src}", []>, VEX;
1051 def VCVTPS2PDrm : I<0x5A, MRMSrcMem, (outs VR128:$dst), (ins f64mem:$src),
1052 "vcvtps2pd\t{$src, $dst|$dst, $src}", []>, VEX;
1053 def VCVTPS2PDYrr : I<0x5A, MRMSrcReg, (outs VR256:$dst), (ins VR128:$src),
1054 "vcvtps2pd\t{$src, $dst|$dst, $src}", []>, VEX;
1055 def VCVTPS2PDYrm : I<0x5A, MRMSrcMem, (outs VR256:$dst), (ins f128mem:$src),
1056 "vcvtps2pd\t{$src, $dst|$dst, $src}", []>, VEX;
1058 def CVTPS2PDrr : I<0x5A, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1059 "cvtps2pd\t{$src, $dst|$dst, $src}", []>, TB;
1060 def CVTPS2PDrm : I<0x5A, MRMSrcMem, (outs VR128:$dst), (ins f64mem:$src),
1061 "cvtps2pd\t{$src, $dst|$dst, $src}", []>, TB;
1063 let isAsmParserOnly = 1 in {
1064 def Int_VCVTPS2PDrr : I<0x5A, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1065 "vcvtps2pd\t{$src, $dst|$dst, $src}",
1066 [(set VR128:$dst, (int_x86_sse2_cvtps2pd VR128:$src))]>,
1067 VEX, Requires<[HasAVX]>;
1068 def Int_VCVTPS2PDrm : I<0x5A, MRMSrcMem, (outs VR128:$dst), (ins f64mem:$src),
1069 "vcvtps2pd\t{$src, $dst|$dst, $src}",
1070 [(set VR128:$dst, (int_x86_sse2_cvtps2pd
1071 (load addr:$src)))]>,
1072 VEX, Requires<[HasAVX]>;
1074 def Int_CVTPS2PDrr : I<0x5A, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1075 "cvtps2pd\t{$src, $dst|$dst, $src}",
1076 [(set VR128:$dst, (int_x86_sse2_cvtps2pd VR128:$src))]>,
1077 TB, Requires<[HasSSE2]>;
1078 def Int_CVTPS2PDrm : I<0x5A, MRMSrcMem, (outs VR128:$dst), (ins f64mem:$src),
1079 "cvtps2pd\t{$src, $dst|$dst, $src}",
1080 [(set VR128:$dst, (int_x86_sse2_cvtps2pd
1081 (load addr:$src)))]>,
1082 TB, Requires<[HasSSE2]>;
1084 // Convert packed double to packed single
1085 let isAsmParserOnly = 1 in {
1086 // The assembler can recognize rr 256-bit instructions by seeing a ymm
1087 // register, but the same isn't true when using memory operands instead.
1088 // Provide other assembly rr and rm forms to address this explicitly.
1089 def VCVTPD2PSrr : VPDI<0x5A, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1090 "cvtpd2ps\t{$src, $dst|$dst, $src}", []>, VEX;
1091 def VCVTPD2PSXrYr : VPDI<0x5A, MRMSrcReg, (outs VR128:$dst), (ins VR256:$src),
1092 "cvtpd2ps\t{$src, $dst|$dst, $src}", []>, VEX;
1095 def VCVTPD2PSXrr : VPDI<0x5A, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1096 "cvtpd2psx\t{$src, $dst|$dst, $src}", []>, VEX;
1097 def VCVTPD2PSXrm : VPDI<0x5A, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
1098 "cvtpd2psx\t{$src, $dst|$dst, $src}", []>, VEX;
1101 def VCVTPD2PSYrr : VPDI<0x5A, MRMSrcReg, (outs VR128:$dst), (ins VR256:$src),
1102 "cvtpd2psy\t{$src, $dst|$dst, $src}", []>, VEX;
1103 def VCVTPD2PSYrm : VPDI<0x5A, MRMSrcMem, (outs VR128:$dst), (ins f256mem:$src),
1104 "cvtpd2psy\t{$src, $dst|$dst, $src}", []>, VEX, VEX_L;
1106 def CVTPD2PSrr : PDI<0x5A, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1107 "cvtpd2ps\t{$src, $dst|$dst, $src}", []>;
1108 def CVTPD2PSrm : PDI<0x5A, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
1109 "cvtpd2ps\t{$src, $dst|$dst, $src}", []>;
1112 let isAsmParserOnly = 1 in {
1113 def Int_VCVTPD2PSrr : VPDI<0x5A, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1114 "cvtpd2ps\t{$src, $dst|$dst, $src}",
1115 [(set VR128:$dst, (int_x86_sse2_cvtpd2ps VR128:$src))]>;
1116 def Int_VCVTPD2PSrm : VPDI<0x5A, MRMSrcMem, (outs VR128:$dst),
1118 "cvtpd2ps\t{$src, $dst|$dst, $src}",
1119 [(set VR128:$dst, (int_x86_sse2_cvtpd2ps
1120 (memop addr:$src)))]>;
1122 def Int_CVTPD2PSrr : PDI<0x5A, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1123 "cvtpd2ps\t{$src, $dst|$dst, $src}",
1124 [(set VR128:$dst, (int_x86_sse2_cvtpd2ps VR128:$src))]>;
1125 def Int_CVTPD2PSrm : PDI<0x5A, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
1126 "cvtpd2ps\t{$src, $dst|$dst, $src}",
1127 [(set VR128:$dst, (int_x86_sse2_cvtpd2ps
1128 (memop addr:$src)))]>;
1130 // AVX 256-bit register conversion intrinsics
1131 // FIXME: Migrate SSE conversion intrinsics matching to use patterns as below
1132 // whenever possible to avoid declaring two versions of each one.
1133 def : Pat<(int_x86_avx_cvtdq2_ps_256 VR256:$src),
1134 (VCVTDQ2PSYrr VR256:$src)>;
1135 def : Pat<(int_x86_avx_cvtdq2_ps_256 (memopv8i32 addr:$src)),
1136 (VCVTDQ2PSYrm addr:$src)>;
1138 def : Pat<(int_x86_avx_cvt_pd2_ps_256 VR256:$src),
1139 (VCVTPD2PSYrr VR256:$src)>;
1140 def : Pat<(int_x86_avx_cvt_pd2_ps_256 (memopv4f64 addr:$src)),
1141 (VCVTPD2PSYrm addr:$src)>;
1143 def : Pat<(int_x86_avx_cvt_ps2dq_256 VR256:$src),
1144 (VCVTPS2DQYrr VR256:$src)>;
1145 def : Pat<(int_x86_avx_cvt_ps2dq_256 (memopv8f32 addr:$src)),
1146 (VCVTPS2DQYrm addr:$src)>;
1148 def : Pat<(int_x86_avx_cvt_ps2_pd_256 VR128:$src),
1149 (VCVTPS2PDYrr VR128:$src)>;
1150 def : Pat<(int_x86_avx_cvt_ps2_pd_256 (memopv4f32 addr:$src)),
1151 (VCVTPS2PDYrm addr:$src)>;
1153 def : Pat<(int_x86_avx_cvtt_pd2dq_256 VR256:$src),
1154 (VCVTTPD2DQYrr VR256:$src)>;
1155 def : Pat<(int_x86_avx_cvtt_pd2dq_256 (memopv4f64 addr:$src)),
1156 (VCVTTPD2DQYrm addr:$src)>;
1158 def : Pat<(int_x86_avx_cvtt_ps2dq_256 VR256:$src),
1159 (VCVTTPS2DQYrr VR256:$src)>;
1160 def : Pat<(int_x86_avx_cvtt_ps2dq_256 (memopv8f32 addr:$src)),
1161 (VCVTTPS2DQYrm addr:$src)>;
1163 //===----------------------------------------------------------------------===//
1164 // SSE 1 & 2 - Compare Instructions
1165 //===----------------------------------------------------------------------===//
1167 // sse12_cmp_scalar - sse 1 & 2 compare scalar instructions
1168 multiclass sse12_cmp_scalar<RegisterClass RC, X86MemOperand x86memop,
1169 string asm, string asm_alt> {
1170 def rr : SIi8<0xC2, MRMSrcReg,
1171 (outs RC:$dst), (ins RC:$src1, RC:$src, SSECC:$cc),
1174 def rm : SIi8<0xC2, MRMSrcMem,
1175 (outs RC:$dst), (ins RC:$src1, x86memop:$src, SSECC:$cc),
1177 // Accept explicit immediate argument form instead of comparison code.
1178 let isAsmParserOnly = 1 in {
1179 def rr_alt : SIi8<0xC2, MRMSrcReg,
1180 (outs RC:$dst), (ins RC:$src1, RC:$src, i8imm:$src2),
1183 def rm_alt : SIi8<0xC2, MRMSrcMem,
1184 (outs RC:$dst), (ins RC:$src1, x86memop:$src, i8imm:$src2),
1189 let neverHasSideEffects = 1, isAsmParserOnly = 1 in {
1190 defm VCMPSS : sse12_cmp_scalar<FR32, f32mem,
1191 "cmp${cc}ss\t{$src, $src1, $dst|$dst, $src1, $src}",
1192 "cmpss\t{$src2, $src, $src1, $dst|$dst, $src1, $src, $src2}">,
1194 defm VCMPSD : sse12_cmp_scalar<FR64, f64mem,
1195 "cmp${cc}sd\t{$src, $src1, $dst|$dst, $src1, $src}",
1196 "cmpsd\t{$src2, $src, $src1, $dst|$dst, $src1, $src, $src2}">,
1200 let Constraints = "$src1 = $dst", neverHasSideEffects = 1 in {
1201 defm CMPSS : sse12_cmp_scalar<FR32, f32mem,
1202 "cmp${cc}ss\t{$src, $dst|$dst, $src}",
1203 "cmpss\t{$src2, $src, $dst|$dst, $src, $src2}">, XS;
1204 defm CMPSD : sse12_cmp_scalar<FR64, f64mem,
1205 "cmp${cc}sd\t{$src, $dst|$dst, $src}",
1206 "cmpsd\t{$src2, $src, $dst|$dst, $src, $src2}">, XD;
1209 multiclass sse12_cmp_scalar_int<RegisterClass RC, Operand memopr,
1210 ComplexPattern mem_cpat, Intrinsic Int, string asm> {
1211 def rr : SIi8<0xC2, MRMSrcReg, (outs VR128:$dst),
1212 (ins VR128:$src1, VR128:$src, SSECC:$cc), asm,
1213 [(set VR128:$dst, (Int VR128:$src1,
1214 VR128:$src, imm:$cc))]>;
1215 def rm : SIi8<0xC2, MRMSrcMem, (outs VR128:$dst),
1216 (ins VR128:$src1, memopr:$src, SSECC:$cc), asm,
1217 [(set VR128:$dst, (Int VR128:$src1,
1218 mem_cpat:$src, imm:$cc))]>;
1221 // Aliases to match intrinsics which expect XMM operand(s).
1223 let isAsmParserOnly = 1 in {
1224 defm Int_VCMPSS : sse12_cmp_scalar_int<VR128, ssmem, sse_load_f32,
1226 "cmp${cc}ss\t{$src, $src1, $dst|$dst, $src1, $src}">,
1228 defm Int_VCMPSD : sse12_cmp_scalar_int<VR128, sdmem, sse_load_f64,
1229 int_x86_sse2_cmp_sd,
1230 "cmp${cc}sd\t{$src, $src1, $dst|$dst, $src1, $src}">,
1233 let Constraints = "$src1 = $dst" in {
1234 defm Int_CMPSS : sse12_cmp_scalar_int<VR128, ssmem, sse_load_f32,
1236 "cmp${cc}ss\t{$src, $dst|$dst, $src}">, XS;
1237 defm Int_CMPSD : sse12_cmp_scalar_int<VR128, sdmem, sse_load_f64,
1238 int_x86_sse2_cmp_sd,
1239 "cmp${cc}sd\t{$src, $dst|$dst, $src}">, XD;
1243 // sse12_ord_cmp - Unordered/Ordered scalar fp compare and set EFLAGS
1244 multiclass sse12_ord_cmp<bits<8> opc, RegisterClass RC, SDNode OpNode,
1245 ValueType vt, X86MemOperand x86memop,
1246 PatFrag ld_frag, string OpcodeStr, Domain d> {
1247 def rr: PI<opc, MRMSrcReg, (outs), (ins RC:$src1, RC:$src2),
1248 !strconcat(OpcodeStr, "\t{$src2, $src1|$src1, $src2}"),
1249 [(set EFLAGS, (OpNode (vt RC:$src1), RC:$src2))], d>;
1250 def rm: PI<opc, MRMSrcMem, (outs), (ins RC:$src1, x86memop:$src2),
1251 !strconcat(OpcodeStr, "\t{$src2, $src1|$src1, $src2}"),
1252 [(set EFLAGS, (OpNode (vt RC:$src1),
1253 (ld_frag addr:$src2)))], d>;
1256 let Defs = [EFLAGS] in {
1257 let isAsmParserOnly = 1 in {
1258 defm VUCOMISS : sse12_ord_cmp<0x2E, FR32, X86cmp, f32, f32mem, loadf32,
1259 "ucomiss", SSEPackedSingle>, VEX;
1260 defm VUCOMISD : sse12_ord_cmp<0x2E, FR64, X86cmp, f64, f64mem, loadf64,
1261 "ucomisd", SSEPackedDouble>, OpSize, VEX;
1262 let Pattern = []<dag> in {
1263 defm VCOMISS : sse12_ord_cmp<0x2F, VR128, undef, v4f32, f128mem, load,
1264 "comiss", SSEPackedSingle>, VEX;
1265 defm VCOMISD : sse12_ord_cmp<0x2F, VR128, undef, v2f64, f128mem, load,
1266 "comisd", SSEPackedDouble>, OpSize, VEX;
1269 defm Int_VUCOMISS : sse12_ord_cmp<0x2E, VR128, X86ucomi, v4f32, f128mem,
1270 load, "ucomiss", SSEPackedSingle>, VEX;
1271 defm Int_VUCOMISD : sse12_ord_cmp<0x2E, VR128, X86ucomi, v2f64, f128mem,
1272 load, "ucomisd", SSEPackedDouble>, OpSize, VEX;
1274 defm Int_VCOMISS : sse12_ord_cmp<0x2F, VR128, X86comi, v4f32, f128mem,
1275 load, "comiss", SSEPackedSingle>, VEX;
1276 defm Int_VCOMISD : sse12_ord_cmp<0x2F, VR128, X86comi, v2f64, f128mem,
1277 load, "comisd", SSEPackedDouble>, OpSize, VEX;
1279 defm UCOMISS : sse12_ord_cmp<0x2E, FR32, X86cmp, f32, f32mem, loadf32,
1280 "ucomiss", SSEPackedSingle>, TB;
1281 defm UCOMISD : sse12_ord_cmp<0x2E, FR64, X86cmp, f64, f64mem, loadf64,
1282 "ucomisd", SSEPackedDouble>, TB, OpSize;
1284 let Pattern = []<dag> in {
1285 defm COMISS : sse12_ord_cmp<0x2F, VR128, undef, v4f32, f128mem, load,
1286 "comiss", SSEPackedSingle>, TB;
1287 defm COMISD : sse12_ord_cmp<0x2F, VR128, undef, v2f64, f128mem, load,
1288 "comisd", SSEPackedDouble>, TB, OpSize;
1291 defm Int_UCOMISS : sse12_ord_cmp<0x2E, VR128, X86ucomi, v4f32, f128mem,
1292 load, "ucomiss", SSEPackedSingle>, TB;
1293 defm Int_UCOMISD : sse12_ord_cmp<0x2E, VR128, X86ucomi, v2f64, f128mem,
1294 load, "ucomisd", SSEPackedDouble>, TB, OpSize;
1296 defm Int_COMISS : sse12_ord_cmp<0x2F, VR128, X86comi, v4f32, f128mem, load,
1297 "comiss", SSEPackedSingle>, TB;
1298 defm Int_COMISD : sse12_ord_cmp<0x2F, VR128, X86comi, v2f64, f128mem, load,
1299 "comisd", SSEPackedDouble>, TB, OpSize;
1300 } // Defs = [EFLAGS]
1302 // sse12_cmp_packed - sse 1 & 2 compared packed instructions
1303 multiclass sse12_cmp_packed<RegisterClass RC, X86MemOperand x86memop,
1304 Intrinsic Int, string asm, string asm_alt,
1306 def rri : PIi8<0xC2, MRMSrcReg,
1307 (outs RC:$dst), (ins RC:$src1, RC:$src, SSECC:$cc), asm,
1308 [(set RC:$dst, (Int RC:$src1, RC:$src, imm:$cc))], d>;
1309 def rmi : PIi8<0xC2, MRMSrcMem,
1310 (outs RC:$dst), (ins RC:$src1, f128mem:$src, SSECC:$cc), asm,
1311 [(set RC:$dst, (Int RC:$src1, (memop addr:$src), imm:$cc))], d>;
1312 // Accept explicit immediate argument form instead of comparison code.
1313 let isAsmParserOnly = 1 in {
1314 def rri_alt : PIi8<0xC2, MRMSrcReg,
1315 (outs RC:$dst), (ins RC:$src1, RC:$src, i8imm:$src2),
1317 def rmi_alt : PIi8<0xC2, MRMSrcMem,
1318 (outs RC:$dst), (ins RC:$src1, f128mem:$src, i8imm:$src2),
1323 let isAsmParserOnly = 1 in {
1324 defm VCMPPS : sse12_cmp_packed<VR128, f128mem, int_x86_sse_cmp_ps,
1325 "cmp${cc}ps\t{$src, $src1, $dst|$dst, $src1, $src}",
1326 "cmpps\t{$src2, $src, $src1, $dst|$dst, $src1, $src, $src2}",
1327 SSEPackedSingle>, VEX_4V;
1328 defm VCMPPD : sse12_cmp_packed<VR128, f128mem, int_x86_sse2_cmp_pd,
1329 "cmp${cc}pd\t{$src, $src1, $dst|$dst, $src1, $src}",
1330 "cmppd\t{$src2, $src, $src1, $dst|$dst, $src1, $src, $src2}",
1331 SSEPackedDouble>, OpSize, VEX_4V;
1332 defm VCMPPSY : sse12_cmp_packed<VR256, f256mem, int_x86_avx_cmp_ps_256,
1333 "cmp${cc}ps\t{$src, $src1, $dst|$dst, $src1, $src}",
1334 "cmpps\t{$src2, $src, $src1, $dst|$dst, $src1, $src, $src2}",
1335 SSEPackedSingle>, VEX_4V;
1336 defm VCMPPDY : sse12_cmp_packed<VR256, f256mem, int_x86_avx_cmp_pd_256,
1337 "cmp${cc}pd\t{$src, $src1, $dst|$dst, $src1, $src}",
1338 "cmppd\t{$src2, $src, $src1, $dst|$dst, $src1, $src, $src2}",
1339 SSEPackedDouble>, OpSize, VEX_4V;
1341 let Constraints = "$src1 = $dst" in {
1342 defm CMPPS : sse12_cmp_packed<VR128, f128mem, int_x86_sse_cmp_ps,
1343 "cmp${cc}ps\t{$src, $dst|$dst, $src}",
1344 "cmpps\t{$src2, $src, $dst|$dst, $src, $src2}",
1345 SSEPackedSingle>, TB;
1346 defm CMPPD : sse12_cmp_packed<VR128, f128mem, int_x86_sse2_cmp_pd,
1347 "cmp${cc}pd\t{$src, $dst|$dst, $src}",
1348 "cmppd\t{$src2, $src, $dst|$dst, $src, $src2}",
1349 SSEPackedDouble>, TB, OpSize;
1352 def : Pat<(v4i32 (X86cmpps (v4f32 VR128:$src1), VR128:$src2, imm:$cc)),
1353 (CMPPSrri (v4f32 VR128:$src1), (v4f32 VR128:$src2), imm:$cc)>;
1354 def : Pat<(v4i32 (X86cmpps (v4f32 VR128:$src1), (memop addr:$src2), imm:$cc)),
1355 (CMPPSrmi (v4f32 VR128:$src1), addr:$src2, imm:$cc)>;
1356 def : Pat<(v2i64 (X86cmppd (v2f64 VR128:$src1), VR128:$src2, imm:$cc)),
1357 (CMPPDrri VR128:$src1, VR128:$src2, imm:$cc)>;
1358 def : Pat<(v2i64 (X86cmppd (v2f64 VR128:$src1), (memop addr:$src2), imm:$cc)),
1359 (CMPPDrmi VR128:$src1, addr:$src2, imm:$cc)>;
1361 //===----------------------------------------------------------------------===//
1362 // SSE 1 & 2 - Shuffle Instructions
1363 //===----------------------------------------------------------------------===//
1365 /// sse12_shuffle - sse 1 & 2 shuffle instructions
1366 multiclass sse12_shuffle<RegisterClass RC, X86MemOperand x86memop,
1367 ValueType vt, string asm, PatFrag mem_frag,
1368 Domain d, bit IsConvertibleToThreeAddress = 0> {
1369 def rmi : PIi8<0xC6, MRMSrcMem, (outs RC:$dst),
1370 (ins RC:$src1, f128mem:$src2, i8imm:$src3), asm,
1371 [(set RC:$dst, (vt (shufp:$src3
1372 RC:$src1, (mem_frag addr:$src2))))], d>;
1373 let isConvertibleToThreeAddress = IsConvertibleToThreeAddress in
1374 def rri : PIi8<0xC6, MRMSrcReg, (outs RC:$dst),
1375 (ins RC:$src1, RC:$src2, i8imm:$src3), asm,
1377 (vt (shufp:$src3 RC:$src1, RC:$src2)))], d>;
1380 let isAsmParserOnly = 1 in {
1381 defm VSHUFPS : sse12_shuffle<VR128, f128mem, v4f32,
1382 "shufps\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}",
1383 memopv4f32, SSEPackedSingle>, VEX_4V;
1384 defm VSHUFPSY : sse12_shuffle<VR256, f256mem, v8f32,
1385 "shufps\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}",
1386 memopv8f32, SSEPackedSingle>, VEX_4V;
1387 defm VSHUFPD : sse12_shuffle<VR128, f128mem, v2f64,
1388 "shufpd\t{$src3, $src2, $src1, $dst|$dst, $src2, $src2, $src3}",
1389 memopv2f64, SSEPackedDouble>, OpSize, VEX_4V;
1390 defm VSHUFPDY : sse12_shuffle<VR256, f256mem, v4f64,
1391 "shufpd\t{$src3, $src2, $src1, $dst|$dst, $src2, $src2, $src3}",
1392 memopv4f64, SSEPackedDouble>, OpSize, VEX_4V;
1395 let Constraints = "$src1 = $dst" in {
1396 defm SHUFPS : sse12_shuffle<VR128, f128mem, v4f32,
1397 "shufps\t{$src3, $src2, $dst|$dst, $src2, $src3}",
1398 memopv4f32, SSEPackedSingle, 1 /* cvt to pshufd */>,
1400 defm SHUFPD : sse12_shuffle<VR128, f128mem, v2f64,
1401 "shufpd\t{$src3, $src2, $dst|$dst, $src2, $src3}",
1402 memopv2f64, SSEPackedDouble>, TB, OpSize;
1405 //===----------------------------------------------------------------------===//
1406 // SSE 1 & 2 - Unpack Instructions
1407 //===----------------------------------------------------------------------===//
1409 /// sse12_unpack_interleave - sse 1 & 2 unpack and interleave
1410 multiclass sse12_unpack_interleave<bits<8> opc, PatFrag OpNode, ValueType vt,
1411 PatFrag mem_frag, RegisterClass RC,
1412 X86MemOperand x86memop, string asm,
1414 def rr : PI<opc, MRMSrcReg,
1415 (outs RC:$dst), (ins RC:$src1, RC:$src2),
1417 (vt (OpNode RC:$src1, RC:$src2)))], d>;
1418 def rm : PI<opc, MRMSrcMem,
1419 (outs RC:$dst), (ins RC:$src1, x86memop:$src2),
1421 (vt (OpNode RC:$src1,
1422 (mem_frag addr:$src2))))], d>;
1425 let AddedComplexity = 10 in {
1426 let isAsmParserOnly = 1 in {
1427 defm VUNPCKHPS: sse12_unpack_interleave<0x15, unpckh, v4f32, memopv4f32,
1428 VR128, f128mem, "unpckhps\t{$src2, $src1, $dst|$dst, $src1, $src2}",
1429 SSEPackedSingle>, VEX_4V;
1430 defm VUNPCKHPD: sse12_unpack_interleave<0x15, unpckh, v2f64, memopv2f64,
1431 VR128, f128mem, "unpckhpd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
1432 SSEPackedDouble>, OpSize, VEX_4V;
1433 defm VUNPCKLPS: sse12_unpack_interleave<0x14, unpckl, v4f32, memopv4f32,
1434 VR128, f128mem, "unpcklps\t{$src2, $src1, $dst|$dst, $src1, $src2}",
1435 SSEPackedSingle>, VEX_4V;
1436 defm VUNPCKLPD: sse12_unpack_interleave<0x14, unpckl, v2f64, memopv2f64,
1437 VR128, f128mem, "unpcklpd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
1438 SSEPackedDouble>, OpSize, VEX_4V;
1440 defm VUNPCKHPSY: sse12_unpack_interleave<0x15, unpckh, v8f32, memopv8f32,
1441 VR256, f256mem, "unpckhps\t{$src2, $src1, $dst|$dst, $src1, $src2}",
1442 SSEPackedSingle>, VEX_4V;
1443 defm VUNPCKHPDY: sse12_unpack_interleave<0x15, unpckh, v4f64, memopv4f64,
1444 VR256, f256mem, "unpckhpd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
1445 SSEPackedDouble>, OpSize, VEX_4V;
1446 defm VUNPCKLPSY: sse12_unpack_interleave<0x14, unpckl, v8f32, memopv8f32,
1447 VR256, f256mem, "unpcklps\t{$src2, $src1, $dst|$dst, $src1, $src2}",
1448 SSEPackedSingle>, VEX_4V;
1449 defm VUNPCKLPDY: sse12_unpack_interleave<0x14, unpckl, v4f64, memopv4f64,
1450 VR256, f256mem, "unpcklpd\t{$src2, $src1, $dst|$dst, $src1, $src2}",
1451 SSEPackedDouble>, OpSize, VEX_4V;
1454 let Constraints = "$src1 = $dst" in {
1455 defm UNPCKHPS: sse12_unpack_interleave<0x15, unpckh, v4f32, memopv4f32,
1456 VR128, f128mem, "unpckhps\t{$src2, $dst|$dst, $src2}",
1457 SSEPackedSingle>, TB;
1458 defm UNPCKHPD: sse12_unpack_interleave<0x15, unpckh, v2f64, memopv2f64,
1459 VR128, f128mem, "unpckhpd\t{$src2, $dst|$dst, $src2}",
1460 SSEPackedDouble>, TB, OpSize;
1461 defm UNPCKLPS: sse12_unpack_interleave<0x14, unpckl, v4f32, memopv4f32,
1462 VR128, f128mem, "unpcklps\t{$src2, $dst|$dst, $src2}",
1463 SSEPackedSingle>, TB;
1464 defm UNPCKLPD: sse12_unpack_interleave<0x14, unpckl, v2f64, memopv2f64,
1465 VR128, f128mem, "unpcklpd\t{$src2, $dst|$dst, $src2}",
1466 SSEPackedDouble>, TB, OpSize;
1467 } // Constraints = "$src1 = $dst"
1468 } // AddedComplexity
1470 //===----------------------------------------------------------------------===//
1471 // SSE 1 & 2 - Extract Floating-Point Sign mask
1472 //===----------------------------------------------------------------------===//
1474 /// sse12_extr_sign_mask - sse 1 & 2 unpack and interleave
1475 multiclass sse12_extr_sign_mask<RegisterClass RC, Intrinsic Int, string asm,
1477 def rr : PI<0x50, MRMSrcReg, (outs GR32:$dst), (ins RC:$src),
1478 !strconcat(asm, "\t{$src, $dst|$dst, $src}"),
1479 [(set GR32:$dst, (Int RC:$src))], d>;
1483 defm MOVMSKPS : sse12_extr_sign_mask<VR128, int_x86_sse_movmsk_ps, "movmskps",
1484 SSEPackedSingle>, TB;
1485 defm MOVMSKPD : sse12_extr_sign_mask<VR128, int_x86_sse2_movmsk_pd, "movmskpd",
1486 SSEPackedDouble>, TB, OpSize;
1488 let isAsmParserOnly = 1 in {
1489 defm VMOVMSKPS : sse12_extr_sign_mask<VR128, int_x86_sse_movmsk_ps,
1490 "movmskps", SSEPackedSingle>, VEX;
1491 defm VMOVMSKPD : sse12_extr_sign_mask<VR128, int_x86_sse2_movmsk_pd,
1492 "movmskpd", SSEPackedDouble>, OpSize,
1494 defm VMOVMSKPSY : sse12_extr_sign_mask<VR256, int_x86_avx_movmsk_ps_256,
1495 "movmskps", SSEPackedSingle>, VEX;
1496 defm VMOVMSKPDY : sse12_extr_sign_mask<VR256, int_x86_avx_movmsk_pd_256,
1497 "movmskpd", SSEPackedDouble>, OpSize,
1501 def VMOVMSKPSr64r : PI<0x50, MRMSrcReg, (outs GR64:$dst), (ins VR128:$src),
1502 "movmskps\t{$src, $dst|$dst, $src}", [], SSEPackedSingle>, VEX;
1503 def VMOVMSKPDr64r : PI<0x50, MRMSrcReg, (outs GR64:$dst), (ins VR128:$src),
1504 "movmskpd\t{$src, $dst|$dst, $src}", [], SSEPackedDouble>, OpSize,
1506 def VMOVMSKPSYr64r : PI<0x50, MRMSrcReg, (outs GR64:$dst), (ins VR256:$src),
1507 "movmskps\t{$src, $dst|$dst, $src}", [], SSEPackedSingle>, VEX;
1508 def VMOVMSKPDYr64r : PI<0x50, MRMSrcReg, (outs GR64:$dst), (ins VR256:$src),
1509 "movmskpd\t{$src, $dst|$dst, $src}", [], SSEPackedDouble>, OpSize,
1513 //===----------------------------------------------------------------------===//
1514 // SSE 1 & 2 - Misc aliasing of packed SSE 1 & 2 instructions
1515 //===----------------------------------------------------------------------===//
1517 // Aliases of packed SSE1 & SSE2 instructions for scalar use. These all have
1518 // names that start with 'Fs'.
1520 // Alias instructions that map fld0 to pxor for sse.
1521 let isReMaterializable = 1, isAsCheapAsAMove = 1, isCodeGenOnly = 1,
1522 canFoldAsLoad = 1 in {
1523 // FIXME: Set encoding to pseudo!
1524 def FsFLD0SS : I<0xEF, MRMInitReg, (outs FR32:$dst), (ins), "",
1525 [(set FR32:$dst, fp32imm0)]>,
1526 Requires<[HasSSE1]>, TB, OpSize;
1527 def FsFLD0SD : I<0xEF, MRMInitReg, (outs FR64:$dst), (ins), "",
1528 [(set FR64:$dst, fpimm0)]>,
1529 Requires<[HasSSE2]>, TB, OpSize;
1532 // Alias instruction to do FR32 or FR64 reg-to-reg copy using movaps. Upper
1533 // bits are disregarded.
1534 let neverHasSideEffects = 1 in {
1535 def FsMOVAPSrr : PSI<0x28, MRMSrcReg, (outs FR32:$dst), (ins FR32:$src),
1536 "movaps\t{$src, $dst|$dst, $src}", []>;
1537 def FsMOVAPDrr : PDI<0x28, MRMSrcReg, (outs FR64:$dst), (ins FR64:$src),
1538 "movapd\t{$src, $dst|$dst, $src}", []>;
1541 // Alias instruction to load FR32 or FR64 from f128mem using movaps. Upper
1542 // bits are disregarded.
1543 let canFoldAsLoad = 1, isReMaterializable = 1 in {
1544 def FsMOVAPSrm : PSI<0x28, MRMSrcMem, (outs FR32:$dst), (ins f128mem:$src),
1545 "movaps\t{$src, $dst|$dst, $src}",
1546 [(set FR32:$dst, (alignedloadfsf32 addr:$src))]>;
1547 def FsMOVAPDrm : PDI<0x28, MRMSrcMem, (outs FR64:$dst), (ins f128mem:$src),
1548 "movapd\t{$src, $dst|$dst, $src}",
1549 [(set FR64:$dst, (alignedloadfsf64 addr:$src))]>;
1552 //===----------------------------------------------------------------------===//
1553 // SSE 1 & 2 - Logical Instructions
1554 //===----------------------------------------------------------------------===//
1556 /// sse12_fp_alias_pack_logical - SSE 1 & 2 aliased packed FP logical ops
1558 multiclass sse12_fp_alias_pack_logical<bits<8> opc, string OpcodeStr,
1560 let isAsmParserOnly = 1 in {
1561 defm V#NAME#PS : sse12_fp_packed<opc, !strconcat(OpcodeStr, "ps"), OpNode,
1562 FR32, f32, f128mem, memopfsf32, SSEPackedSingle, 0>, VEX_4V;
1564 defm V#NAME#PD : sse12_fp_packed<opc, !strconcat(OpcodeStr, "pd"), OpNode,
1565 FR64, f64, f128mem, memopfsf64, SSEPackedDouble, 0>, OpSize, VEX_4V;
1568 let Constraints = "$src1 = $dst" in {
1569 defm PS : sse12_fp_packed<opc, !strconcat(OpcodeStr, "ps"), OpNode, FR32,
1570 f32, f128mem, memopfsf32, SSEPackedSingle>, TB;
1572 defm PD : sse12_fp_packed<opc, !strconcat(OpcodeStr, "pd"), OpNode, FR64,
1573 f64, f128mem, memopfsf64, SSEPackedDouble>, TB, OpSize;
1577 // Alias bitwise logical operations using SSE logical ops on packed FP values.
1578 let mayLoad = 0 in {
1579 defm FsAND : sse12_fp_alias_pack_logical<0x54, "and", X86fand>;
1580 defm FsOR : sse12_fp_alias_pack_logical<0x56, "or", X86for>;
1581 defm FsXOR : sse12_fp_alias_pack_logical<0x57, "xor", X86fxor>;
1584 let neverHasSideEffects = 1, Pattern = []<dag>, isCommutable = 0 in
1585 defm FsANDN : sse12_fp_alias_pack_logical<0x55, "andn", undef>;
1587 /// sse12_fp_packed_logical - SSE 1 & 2 packed FP logical ops
1589 multiclass sse12_fp_packed_logical<bits<8> opc, string OpcodeStr,
1590 SDNode OpNode, int HasPat = 0,
1591 list<list<dag>> Pattern = []> {
1592 let isAsmParserOnly = 1, Pattern = []<dag> in {
1593 defm V#NAME#PS : sse12_fp_packed_logical_rm<opc, VR128, SSEPackedSingle,
1594 !strconcat(OpcodeStr, "ps"), f128mem,
1595 !if(HasPat, Pattern[0], // rr
1596 [(set VR128:$dst, (v2i64 (OpNode VR128:$src1,
1598 !if(HasPat, Pattern[2], // rm
1599 [(set VR128:$dst, (OpNode (bc_v2i64 (v4f32 VR128:$src1)),
1600 (memopv2i64 addr:$src2)))]), 0>,
1603 defm V#NAME#PD : sse12_fp_packed_logical_rm<opc, VR128, SSEPackedDouble,
1604 !strconcat(OpcodeStr, "pd"), f128mem,
1605 !if(HasPat, Pattern[1], // rr
1606 [(set VR128:$dst, (OpNode (bc_v2i64 (v2f64 VR128:$src1)),
1609 !if(HasPat, Pattern[3], // rm
1610 [(set VR128:$dst, (OpNode (bc_v2i64 (v2f64 VR128:$src1)),
1611 (memopv2i64 addr:$src2)))]), 0>,
1614 let Constraints = "$src1 = $dst" in {
1615 defm PS : sse12_fp_packed_logical_rm<opc, VR128, SSEPackedSingle,
1616 !strconcat(OpcodeStr, "ps"), f128mem,
1617 !if(HasPat, Pattern[0], // rr
1618 [(set VR128:$dst, (v2i64 (OpNode VR128:$src1,
1620 !if(HasPat, Pattern[2], // rm
1621 [(set VR128:$dst, (OpNode (bc_v2i64 (v4f32 VR128:$src1)),
1622 (memopv2i64 addr:$src2)))])>, TB;
1624 defm PD : sse12_fp_packed_logical_rm<opc, VR128, SSEPackedDouble,
1625 !strconcat(OpcodeStr, "pd"), f128mem,
1626 !if(HasPat, Pattern[1], // rr
1627 [(set VR128:$dst, (OpNode (bc_v2i64 (v2f64 VR128:$src1)),
1630 !if(HasPat, Pattern[3], // rm
1631 [(set VR128:$dst, (OpNode (bc_v2i64 (v2f64 VR128:$src1)),
1632 (memopv2i64 addr:$src2)))])>,
1637 /// sse12_fp_packed_logical_y - AVX 256-bit SSE 1 & 2 logical ops forms
1639 let isAsmParserOnly = 1 in {
1640 multiclass sse12_fp_packed_logical_y<bits<8> opc, string OpcodeStr> {
1641 defm PSY : sse12_fp_packed_logical_rm<opc, VR256, SSEPackedSingle,
1642 !strconcat(OpcodeStr, "ps"), f256mem, [], [], 0>, VEX_4V;
1644 defm PDY : sse12_fp_packed_logical_rm<opc, VR256, SSEPackedDouble,
1645 !strconcat(OpcodeStr, "pd"), f256mem, [], [], 0>, OpSize, VEX_4V;
1649 // AVX 256-bit packed logical ops forms
1650 defm VAND : sse12_fp_packed_logical_y<0x54, "and">;
1651 defm VOR : sse12_fp_packed_logical_y<0x56, "or">;
1652 defm VXOR : sse12_fp_packed_logical_y<0x57, "xor">;
1653 let isCommutable = 0 in
1654 defm VANDN : sse12_fp_packed_logical_y<0x55, "andn">;
1656 defm AND : sse12_fp_packed_logical<0x54, "and", and>;
1657 defm OR : sse12_fp_packed_logical<0x56, "or", or>;
1658 defm XOR : sse12_fp_packed_logical<0x57, "xor", xor>;
1659 let isCommutable = 0 in
1660 defm ANDN : sse12_fp_packed_logical<0x55, "andn", undef /* dummy */, 1, [
1662 [(set VR128:$dst, (v2i64 (and (xor VR128:$src1,
1663 (bc_v2i64 (v4i32 immAllOnesV))),
1666 [(set VR128:$dst, (and (vnot (bc_v2i64 (v2f64 VR128:$src1))),
1667 (bc_v2i64 (v2f64 VR128:$src2))))],
1669 [(set VR128:$dst, (v2i64 (and (xor (bc_v2i64 (v4f32 VR128:$src1)),
1670 (bc_v2i64 (v4i32 immAllOnesV))),
1671 (memopv2i64 addr:$src2))))],
1673 [(set VR128:$dst, (and (vnot (bc_v2i64 (v2f64 VR128:$src1))),
1674 (memopv2i64 addr:$src2)))]]>;
1676 //===----------------------------------------------------------------------===//
1677 // SSE 1 & 2 - Arithmetic Instructions
1678 //===----------------------------------------------------------------------===//
1680 /// basic_sse12_fp_binop_xxx - SSE 1 & 2 binops come in both scalar and
1683 /// In addition, we also have a special variant of the scalar form here to
1684 /// represent the associated intrinsic operation. This form is unlike the
1685 /// plain scalar form, in that it takes an entire vector (instead of a scalar)
1686 /// and leaves the top elements unmodified (therefore these cannot be commuted).
1688 /// These three forms can each be reg+reg or reg+mem.
1691 /// FIXME: once all 256-bit intrinsics are matched, cleanup and refactor those
1693 multiclass basic_sse12_fp_binop_s<bits<8> opc, string OpcodeStr, SDNode OpNode,
1695 defm SS : sse12_fp_scalar<opc, !strconcat(OpcodeStr, "ss"),
1696 OpNode, FR32, f32mem, Is2Addr>, XS;
1697 defm SD : sse12_fp_scalar<opc, !strconcat(OpcodeStr, "sd"),
1698 OpNode, FR64, f64mem, Is2Addr>, XD;
1701 multiclass basic_sse12_fp_binop_p<bits<8> opc, string OpcodeStr, SDNode OpNode,
1703 let mayLoad = 0 in {
1704 defm PS : sse12_fp_packed<opc, !strconcat(OpcodeStr, "ps"), OpNode, VR128,
1705 v4f32, f128mem, memopv4f32, SSEPackedSingle, Is2Addr>, TB;
1706 defm PD : sse12_fp_packed<opc, !strconcat(OpcodeStr, "pd"), OpNode, VR128,
1707 v2f64, f128mem, memopv2f64, SSEPackedDouble, Is2Addr>, TB, OpSize;
1711 multiclass basic_sse12_fp_binop_p_y<bits<8> opc, string OpcodeStr,
1713 let mayLoad = 0 in {
1714 defm PSY : sse12_fp_packed<opc, !strconcat(OpcodeStr, "ps"), OpNode, VR256,
1715 v8f32, f256mem, memopv8f32, SSEPackedSingle, 0>, TB;
1716 defm PDY : sse12_fp_packed<opc, !strconcat(OpcodeStr, "pd"), OpNode, VR256,
1717 v4f64, f256mem, memopv4f64, SSEPackedDouble, 0>, TB, OpSize;
1721 multiclass basic_sse12_fp_binop_s_int<bits<8> opc, string OpcodeStr,
1723 defm SS : sse12_fp_scalar_int<opc, OpcodeStr, VR128,
1724 !strconcat(OpcodeStr, "ss"), "", "_ss", ssmem, sse_load_f32, Is2Addr>, XS;
1725 defm SD : sse12_fp_scalar_int<opc, OpcodeStr, VR128,
1726 !strconcat(OpcodeStr, "sd"), "2", "_sd", sdmem, sse_load_f64, Is2Addr>, XD;
1729 multiclass basic_sse12_fp_binop_p_int<bits<8> opc, string OpcodeStr,
1731 defm PS : sse12_fp_packed_int<opc, OpcodeStr, VR128,
1732 !strconcat(OpcodeStr, "ps"), "sse", "_ps", f128mem, memopv4f32,
1733 SSEPackedSingle, Is2Addr>, TB;
1735 defm PD : sse12_fp_packed_int<opc, OpcodeStr, VR128,
1736 !strconcat(OpcodeStr, "pd"), "sse2", "_pd", f128mem, memopv2f64,
1737 SSEPackedDouble, Is2Addr>, TB, OpSize;
1740 multiclass basic_sse12_fp_binop_p_y_int<bits<8> opc, string OpcodeStr> {
1741 defm PSY : sse12_fp_packed_int<opc, OpcodeStr, VR256,
1742 !strconcat(OpcodeStr, "ps"), "avx", "_ps_256", f256mem, memopv8f32,
1743 SSEPackedSingle, 0>, TB;
1745 defm PDY : sse12_fp_packed_int<opc, OpcodeStr, VR256,
1746 !strconcat(OpcodeStr, "pd"), "avx", "_pd_256", f256mem, memopv4f64,
1747 SSEPackedDouble, 0>, TB, OpSize;
1750 // Binary Arithmetic instructions
1751 let isAsmParserOnly = 1 in {
1752 defm VADD : basic_sse12_fp_binop_s<0x58, "add", fadd, 0>,
1753 basic_sse12_fp_binop_s_int<0x58, "add", 0>,
1754 basic_sse12_fp_binop_p<0x58, "add", fadd, 0>,
1755 basic_sse12_fp_binop_p_y<0x58, "add", fadd>, VEX_4V;
1756 defm VMUL : basic_sse12_fp_binop_s<0x59, "mul", fmul, 0>,
1757 basic_sse12_fp_binop_s_int<0x59, "mul", 0>,
1758 basic_sse12_fp_binop_p<0x59, "mul", fmul, 0>,
1759 basic_sse12_fp_binop_p_y<0x59, "mul", fmul>, VEX_4V;
1761 let isCommutable = 0 in {
1762 defm VSUB : basic_sse12_fp_binop_s<0x5C, "sub", fsub, 0>,
1763 basic_sse12_fp_binop_s_int<0x5C, "sub", 0>,
1764 basic_sse12_fp_binop_p<0x5C, "sub", fsub, 0>,
1765 basic_sse12_fp_binop_p_y<0x5C, "sub", fsub>, VEX_4V;
1766 defm VDIV : basic_sse12_fp_binop_s<0x5E, "div", fdiv, 0>,
1767 basic_sse12_fp_binop_s_int<0x5E, "div", 0>,
1768 basic_sse12_fp_binop_p<0x5E, "div", fdiv, 0>,
1769 basic_sse12_fp_binop_p_y<0x5E, "div", fdiv>, VEX_4V;
1770 defm VMAX : basic_sse12_fp_binop_s<0x5F, "max", X86fmax, 0>,
1771 basic_sse12_fp_binop_s_int<0x5F, "max", 0>,
1772 basic_sse12_fp_binop_p<0x5F, "max", X86fmax, 0>,
1773 basic_sse12_fp_binop_p_int<0x5F, "max", 0>,
1774 basic_sse12_fp_binop_p_y<0x5F, "max", X86fmax>,
1775 basic_sse12_fp_binop_p_y_int<0x5F, "max">, VEX_4V;
1776 defm VMIN : basic_sse12_fp_binop_s<0x5D, "min", X86fmin, 0>,
1777 basic_sse12_fp_binop_s_int<0x5D, "min", 0>,
1778 basic_sse12_fp_binop_p<0x5D, "min", X86fmin, 0>,
1779 basic_sse12_fp_binop_p_int<0x5D, "min", 0>,
1780 basic_sse12_fp_binop_p_y_int<0x5D, "min">,
1781 basic_sse12_fp_binop_p_y<0x5D, "min", X86fmin>, VEX_4V;
1785 let Constraints = "$src1 = $dst" in {
1786 defm ADD : basic_sse12_fp_binop_s<0x58, "add", fadd>,
1787 basic_sse12_fp_binop_p<0x58, "add", fadd>,
1788 basic_sse12_fp_binop_s_int<0x58, "add">;
1789 defm MUL : basic_sse12_fp_binop_s<0x59, "mul", fmul>,
1790 basic_sse12_fp_binop_p<0x59, "mul", fmul>,
1791 basic_sse12_fp_binop_s_int<0x59, "mul">;
1793 let isCommutable = 0 in {
1794 defm SUB : basic_sse12_fp_binop_s<0x5C, "sub", fsub>,
1795 basic_sse12_fp_binop_p<0x5C, "sub", fsub>,
1796 basic_sse12_fp_binop_s_int<0x5C, "sub">;
1797 defm DIV : basic_sse12_fp_binop_s<0x5E, "div", fdiv>,
1798 basic_sse12_fp_binop_p<0x5E, "div", fdiv>,
1799 basic_sse12_fp_binop_s_int<0x5E, "div">;
1800 defm MAX : basic_sse12_fp_binop_s<0x5F, "max", X86fmax>,
1801 basic_sse12_fp_binop_p<0x5F, "max", X86fmax>,
1802 basic_sse12_fp_binop_s_int<0x5F, "max">,
1803 basic_sse12_fp_binop_p_int<0x5F, "max">;
1804 defm MIN : basic_sse12_fp_binop_s<0x5D, "min", X86fmin>,
1805 basic_sse12_fp_binop_p<0x5D, "min", X86fmin>,
1806 basic_sse12_fp_binop_s_int<0x5D, "min">,
1807 basic_sse12_fp_binop_p_int<0x5D, "min">;
1812 /// In addition, we also have a special variant of the scalar form here to
1813 /// represent the associated intrinsic operation. This form is unlike the
1814 /// plain scalar form, in that it takes an entire vector (instead of a
1815 /// scalar) and leaves the top elements undefined.
1817 /// And, we have a special variant form for a full-vector intrinsic form.
1819 /// sse1_fp_unop_s - SSE1 unops in scalar form.
1820 multiclass sse1_fp_unop_s<bits<8> opc, string OpcodeStr,
1821 SDNode OpNode, Intrinsic F32Int> {
1822 def SSr : SSI<opc, MRMSrcReg, (outs FR32:$dst), (ins FR32:$src),
1823 !strconcat(OpcodeStr, "ss\t{$src, $dst|$dst, $src}"),
1824 [(set FR32:$dst, (OpNode FR32:$src))]>;
1825 // For scalar unary operations, fold a load into the operation
1826 // only in OptForSize mode. It eliminates an instruction, but it also
1827 // eliminates a whole-register clobber (the load), so it introduces a
1828 // partial register update condition.
1829 def SSm : I<opc, MRMSrcMem, (outs FR32:$dst), (ins f32mem:$src),
1830 !strconcat(OpcodeStr, "ss\t{$src, $dst|$dst, $src}"),
1831 [(set FR32:$dst, (OpNode (load addr:$src)))]>, XS,
1832 Requires<[HasSSE1, OptForSize]>;
1833 def SSr_Int : SSI<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1834 !strconcat(OpcodeStr, "ss\t{$src, $dst|$dst, $src}"),
1835 [(set VR128:$dst, (F32Int VR128:$src))]>;
1836 def SSm_Int : SSI<opc, MRMSrcMem, (outs VR128:$dst), (ins ssmem:$src),
1837 !strconcat(OpcodeStr, "ss\t{$src, $dst|$dst, $src}"),
1838 [(set VR128:$dst, (F32Int sse_load_f32:$src))]>;
1841 /// sse1_fp_unop_s_avx - AVX SSE1 unops in scalar form.
1842 multiclass sse1_fp_unop_s_avx<bits<8> opc, string OpcodeStr,
1843 SDNode OpNode, Intrinsic F32Int> {
1844 def SSr : SSI<opc, MRMSrcReg, (outs FR32:$dst), (ins FR32:$src1, FR32:$src2),
1845 !strconcat(OpcodeStr,
1846 "ss\t{$src2, $src1, $dst|$dst, $src1, $src2}"), []>;
1847 def SSm : I<opc, MRMSrcMem, (outs FR32:$dst), (ins FR32:$src1, f32mem:$src2),
1848 !strconcat(OpcodeStr,
1849 "ss\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
1850 []>, XS, Requires<[HasAVX, OptForSize]>;
1851 def SSr_Int : SSI<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1852 !strconcat(OpcodeStr,
1853 "ss\t{$src, $dst, $dst|$dst, $dst, $src}"),
1854 [(set VR128:$dst, (F32Int VR128:$src))]>;
1855 def SSm_Int : SSI<opc, MRMSrcMem, (outs VR128:$dst), (ins ssmem:$src),
1856 !strconcat(OpcodeStr,
1857 "ss\t{$src, $dst, $dst|$dst, $dst, $src}"),
1858 [(set VR128:$dst, (F32Int sse_load_f32:$src))]>;
1861 /// sse1_fp_unop_p - SSE1 unops in packed form.
1862 multiclass sse1_fp_unop_p<bits<8> opc, string OpcodeStr, SDNode OpNode> {
1863 def PSr : PSI<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1864 !strconcat(OpcodeStr, "ps\t{$src, $dst|$dst, $src}"),
1865 [(set VR128:$dst, (v4f32 (OpNode VR128:$src)))]>;
1866 def PSm : PSI<opc, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
1867 !strconcat(OpcodeStr, "ps\t{$src, $dst|$dst, $src}"),
1868 [(set VR128:$dst, (OpNode (memopv4f32 addr:$src)))]>;
1871 /// sse1_fp_unop_p_y - AVX 256-bit SSE1 unops in packed form.
1872 multiclass sse1_fp_unop_p_y<bits<8> opc, string OpcodeStr, SDNode OpNode> {
1873 def PSYr : PSI<opc, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src),
1874 !strconcat(OpcodeStr, "ps\t{$src, $dst|$dst, $src}"),
1875 [(set VR256:$dst, (v8f32 (OpNode VR256:$src)))]>;
1876 def PSYm : PSI<opc, MRMSrcMem, (outs VR256:$dst), (ins f256mem:$src),
1877 !strconcat(OpcodeStr, "ps\t{$src, $dst|$dst, $src}"),
1878 [(set VR256:$dst, (OpNode (memopv8f32 addr:$src)))]>;
1881 /// sse1_fp_unop_p_int - SSE1 intrinsics unops in packed forms.
1882 multiclass sse1_fp_unop_p_int<bits<8> opc, string OpcodeStr,
1883 Intrinsic V4F32Int> {
1884 def PSr_Int : PSI<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1885 !strconcat(OpcodeStr, "ps\t{$src, $dst|$dst, $src}"),
1886 [(set VR128:$dst, (V4F32Int VR128:$src))]>;
1887 def PSm_Int : PSI<opc, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
1888 !strconcat(OpcodeStr, "ps\t{$src, $dst|$dst, $src}"),
1889 [(set VR128:$dst, (V4F32Int (memopv4f32 addr:$src)))]>;
1892 /// sse1_fp_unop_p_y_int - AVX 256-bit intrinsics unops in packed forms.
1893 multiclass sse1_fp_unop_p_y_int<bits<8> opc, string OpcodeStr,
1894 Intrinsic V4F32Int> {
1895 def PSYr_Int : PSI<opc, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src),
1896 !strconcat(OpcodeStr, "ps\t{$src, $dst|$dst, $src}"),
1897 [(set VR256:$dst, (V4F32Int VR256:$src))]>;
1898 def PSYm_Int : PSI<opc, MRMSrcMem, (outs VR256:$dst), (ins f256mem:$src),
1899 !strconcat(OpcodeStr, "ps\t{$src, $dst|$dst, $src}"),
1900 [(set VR256:$dst, (V4F32Int (memopv8f32 addr:$src)))]>;
1903 /// sse2_fp_unop_s - SSE2 unops in scalar form.
1904 multiclass sse2_fp_unop_s<bits<8> opc, string OpcodeStr,
1905 SDNode OpNode, Intrinsic F64Int> {
1906 def SDr : SDI<opc, MRMSrcReg, (outs FR64:$dst), (ins FR64:$src),
1907 !strconcat(OpcodeStr, "sd\t{$src, $dst|$dst, $src}"),
1908 [(set FR64:$dst, (OpNode FR64:$src))]>;
1909 // See the comments in sse1_fp_unop_s for why this is OptForSize.
1910 def SDm : I<opc, MRMSrcMem, (outs FR64:$dst), (ins f64mem:$src),
1911 !strconcat(OpcodeStr, "sd\t{$src, $dst|$dst, $src}"),
1912 [(set FR64:$dst, (OpNode (load addr:$src)))]>, XD,
1913 Requires<[HasSSE2, OptForSize]>;
1914 def SDr_Int : SDI<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1915 !strconcat(OpcodeStr, "sd\t{$src, $dst|$dst, $src}"),
1916 [(set VR128:$dst, (F64Int VR128:$src))]>;
1917 def SDm_Int : SDI<opc, MRMSrcMem, (outs VR128:$dst), (ins sdmem:$src),
1918 !strconcat(OpcodeStr, "sd\t{$src, $dst|$dst, $src}"),
1919 [(set VR128:$dst, (F64Int sse_load_f64:$src))]>;
1922 /// sse2_fp_unop_s_avx - AVX SSE2 unops in scalar form.
1923 multiclass sse2_fp_unop_s_avx<bits<8> opc, string OpcodeStr,
1924 SDNode OpNode, Intrinsic F64Int> {
1925 def SDr : SDI<opc, MRMSrcReg, (outs FR64:$dst), (ins FR64:$src1, FR64:$src2),
1926 !strconcat(OpcodeStr,
1927 "sd\t{$src2, $src1, $dst|$dst, $src1, $src2}"), []>;
1928 def SDm : SDI<opc, MRMSrcMem, (outs FR64:$dst),
1929 (ins FR64:$src1, f64mem:$src2),
1930 !strconcat(OpcodeStr,
1931 "sd\t{$src2, $src1, $dst|$dst, $src1, $src2}"), []>;
1932 def SDr_Int : SDI<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1933 !strconcat(OpcodeStr, "sd\t{$src, $dst, $dst|$dst, $dst, $src}"),
1934 [(set VR128:$dst, (F64Int VR128:$src))]>;
1935 def SDm_Int : SDI<opc, MRMSrcMem, (outs VR128:$dst), (ins sdmem:$src),
1936 !strconcat(OpcodeStr, "sd\t{$src, $dst, $dst|$dst, $dst, $src}"),
1937 [(set VR128:$dst, (F64Int sse_load_f64:$src))]>;
1940 /// sse2_fp_unop_p - SSE2 unops in vector forms.
1941 multiclass sse2_fp_unop_p<bits<8> opc, string OpcodeStr,
1943 def PDr : PDI<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1944 !strconcat(OpcodeStr, "pd\t{$src, $dst|$dst, $src}"),
1945 [(set VR128:$dst, (v2f64 (OpNode VR128:$src)))]>;
1946 def PDm : PDI<opc, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
1947 !strconcat(OpcodeStr, "pd\t{$src, $dst|$dst, $src}"),
1948 [(set VR128:$dst, (OpNode (memopv2f64 addr:$src)))]>;
1951 /// sse2_fp_unop_p_y - AVX SSE2 256-bit unops in vector forms.
1952 multiclass sse2_fp_unop_p_y<bits<8> opc, string OpcodeStr, SDNode OpNode> {
1953 def PDYr : PDI<opc, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src),
1954 !strconcat(OpcodeStr, "pd\t{$src, $dst|$dst, $src}"),
1955 [(set VR256:$dst, (v4f64 (OpNode VR256:$src)))]>;
1956 def PDYm : PDI<opc, MRMSrcMem, (outs VR256:$dst), (ins f256mem:$src),
1957 !strconcat(OpcodeStr, "pd\t{$src, $dst|$dst, $src}"),
1958 [(set VR256:$dst, (OpNode (memopv4f64 addr:$src)))]>;
1961 /// sse2_fp_unop_p_int - SSE2 intrinsic unops in vector forms.
1962 multiclass sse2_fp_unop_p_int<bits<8> opc, string OpcodeStr,
1963 Intrinsic V2F64Int> {
1964 def PDr_Int : PDI<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
1965 !strconcat(OpcodeStr, "pd\t{$src, $dst|$dst, $src}"),
1966 [(set VR128:$dst, (V2F64Int VR128:$src))]>;
1967 def PDm_Int : PDI<opc, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
1968 !strconcat(OpcodeStr, "pd\t{$src, $dst|$dst, $src}"),
1969 [(set VR128:$dst, (V2F64Int (memopv2f64 addr:$src)))]>;
1972 /// sse2_fp_unop_p_y_int - AVX 256-bit intrinsic unops in vector forms.
1973 multiclass sse2_fp_unop_p_y_int<bits<8> opc, string OpcodeStr,
1974 Intrinsic V2F64Int> {
1975 def PDYr_Int : PDI<opc, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src),
1976 !strconcat(OpcodeStr, "pd\t{$src, $dst|$dst, $src}"),
1977 [(set VR256:$dst, (V2F64Int VR256:$src))]>;
1978 def PDYm_Int : PDI<opc, MRMSrcMem, (outs VR256:$dst), (ins f256mem:$src),
1979 !strconcat(OpcodeStr, "pd\t{$src, $dst|$dst, $src}"),
1980 [(set VR256:$dst, (V2F64Int (memopv4f64 addr:$src)))]>;
1983 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
1985 defm VSQRT : sse1_fp_unop_s_avx<0x51, "vsqrt", fsqrt, int_x86_sse_sqrt_ss>,
1986 sse2_fp_unop_s_avx<0x51, "vsqrt", fsqrt, int_x86_sse2_sqrt_sd>,
1989 defm VSQRT : sse1_fp_unop_p<0x51, "vsqrt", fsqrt>,
1990 sse2_fp_unop_p<0x51, "vsqrt", fsqrt>,
1991 sse1_fp_unop_p_y<0x51, "vsqrt", fsqrt>,
1992 sse2_fp_unop_p_y<0x51, "vsqrt", fsqrt>,
1993 sse1_fp_unop_p_int<0x51, "vsqrt", int_x86_sse_sqrt_ps>,
1994 sse2_fp_unop_p_int<0x51, "vsqrt", int_x86_sse2_sqrt_pd>,
1995 sse1_fp_unop_p_y_int<0x51, "vsqrt", int_x86_avx_sqrt_ps_256>,
1996 sse2_fp_unop_p_y_int<0x51, "vsqrt", int_x86_avx_sqrt_pd_256>,
1999 // Reciprocal approximations. Note that these typically require refinement
2000 // in order to obtain suitable precision.
2001 defm VRSQRT : sse1_fp_unop_s_avx<0x52, "vrsqrt", X86frsqrt,
2002 int_x86_sse_rsqrt_ss>, VEX_4V;
2003 defm VRSQRT : sse1_fp_unop_p<0x52, "vrsqrt", X86frsqrt>,
2004 sse1_fp_unop_p_y<0x52, "vrsqrt", X86frsqrt>,
2005 sse1_fp_unop_p_y_int<0x52, "vrsqrt", int_x86_avx_rsqrt_ps_256>,
2006 sse1_fp_unop_p_int<0x52, "vrsqrt", int_x86_sse_rsqrt_ps>, VEX;
2008 defm VRCP : sse1_fp_unop_s_avx<0x53, "vrcp", X86frcp, int_x86_sse_rcp_ss>,
2010 defm VRCP : sse1_fp_unop_p<0x53, "vrcp", X86frcp>,
2011 sse1_fp_unop_p_y<0x53, "vrcp", X86frcp>,
2012 sse1_fp_unop_p_y_int<0x53, "vrcp", int_x86_avx_rcp_ps_256>,
2013 sse1_fp_unop_p_int<0x53, "vrcp", int_x86_sse_rcp_ps>, VEX;
2017 defm SQRT : sse1_fp_unop_s<0x51, "sqrt", fsqrt, int_x86_sse_sqrt_ss>,
2018 sse1_fp_unop_p<0x51, "sqrt", fsqrt>,
2019 sse1_fp_unop_p_int<0x51, "sqrt", int_x86_sse_sqrt_ps>,
2020 sse2_fp_unop_s<0x51, "sqrt", fsqrt, int_x86_sse2_sqrt_sd>,
2021 sse2_fp_unop_p<0x51, "sqrt", fsqrt>,
2022 sse2_fp_unop_p_int<0x51, "sqrt", int_x86_sse2_sqrt_pd>;
2024 // Reciprocal approximations. Note that these typically require refinement
2025 // in order to obtain suitable precision.
2026 defm RSQRT : sse1_fp_unop_s<0x52, "rsqrt", X86frsqrt, int_x86_sse_rsqrt_ss>,
2027 sse1_fp_unop_p<0x52, "rsqrt", X86frsqrt>,
2028 sse1_fp_unop_p_int<0x52, "rsqrt", int_x86_sse_rsqrt_ps>;
2029 defm RCP : sse1_fp_unop_s<0x53, "rcp", X86frcp, int_x86_sse_rcp_ss>,
2030 sse1_fp_unop_p<0x53, "rcp", X86frcp>,
2031 sse1_fp_unop_p_int<0x53, "rcp", int_x86_sse_rcp_ps>;
2033 // There is no f64 version of the reciprocal approximation instructions.
2035 //===----------------------------------------------------------------------===//
2036 // SSE 1 & 2 - Non-temporal stores
2037 //===----------------------------------------------------------------------===//
2039 let isAsmParserOnly = 1 in {
2040 def VMOVNTPSmr_Int : VPSI<0x2B, MRMDestMem, (outs),
2041 (ins i128mem:$dst, VR128:$src),
2042 "movntps\t{$src, $dst|$dst, $src}",
2043 [(int_x86_sse_movnt_ps addr:$dst, VR128:$src)]>, VEX;
2044 def VMOVNTPDmr_Int : VPDI<0x2B, MRMDestMem, (outs),
2045 (ins i128mem:$dst, VR128:$src),
2046 "movntpd\t{$src, $dst|$dst, $src}",
2047 [(int_x86_sse2_movnt_pd addr:$dst, VR128:$src)]>, VEX;
2049 let ExeDomain = SSEPackedInt in
2050 def VMOVNTDQmr_Int : VPDI<0xE7, MRMDestMem, (outs),
2051 (ins f128mem:$dst, VR128:$src),
2052 "movntdq\t{$src, $dst|$dst, $src}",
2053 [(int_x86_sse2_movnt_dq addr:$dst, VR128:$src)]>, VEX;
2055 let AddedComplexity = 400 in { // Prefer non-temporal versions
2056 def VMOVNTPSmr : VPSI<0x2B, MRMDestMem, (outs),
2057 (ins f128mem:$dst, VR128:$src),
2058 "movntps\t{$src, $dst|$dst, $src}",
2059 [(alignednontemporalstore (v4f32 VR128:$src),
2061 def VMOVNTPDmr : VPDI<0x2B, MRMDestMem, (outs),
2062 (ins f128mem:$dst, VR128:$src),
2063 "movntpd\t{$src, $dst|$dst, $src}",
2064 [(alignednontemporalstore (v2f64 VR128:$src),
2066 def VMOVNTDQ_64mr : VPDI<0xE7, MRMDestMem, (outs),
2067 (ins f128mem:$dst, VR128:$src),
2068 "movntdq\t{$src, $dst|$dst, $src}",
2069 [(alignednontemporalstore (v2f64 VR128:$src),
2071 let ExeDomain = SSEPackedInt in
2072 def VMOVNTDQmr : VPDI<0xE7, MRMDestMem, (outs),
2073 (ins f128mem:$dst, VR128:$src),
2074 "movntdq\t{$src, $dst|$dst, $src}",
2075 [(alignednontemporalstore (v4f32 VR128:$src),
2078 def VMOVNTPSYmr : VPSI<0x2B, MRMDestMem, (outs),
2079 (ins f256mem:$dst, VR256:$src),
2080 "movntps\t{$src, $dst|$dst, $src}",
2081 [(alignednontemporalstore (v8f32 VR256:$src),
2083 def VMOVNTPDYmr : VPDI<0x2B, MRMDestMem, (outs),
2084 (ins f256mem:$dst, VR256:$src),
2085 "movntpd\t{$src, $dst|$dst, $src}",
2086 [(alignednontemporalstore (v4f64 VR256:$src),
2088 def VMOVNTDQY_64mr : VPDI<0xE7, MRMDestMem, (outs),
2089 (ins f256mem:$dst, VR256:$src),
2090 "movntdq\t{$src, $dst|$dst, $src}",
2091 [(alignednontemporalstore (v4f64 VR256:$src),
2093 let ExeDomain = SSEPackedInt in
2094 def VMOVNTDQYmr : VPDI<0xE7, MRMDestMem, (outs),
2095 (ins f256mem:$dst, VR256:$src),
2096 "movntdq\t{$src, $dst|$dst, $src}",
2097 [(alignednontemporalstore (v8f32 VR256:$src),
2102 def : Pat<(int_x86_avx_movnt_dq_256 addr:$dst, VR256:$src),
2103 (VMOVNTDQYmr addr:$dst, VR256:$src)>;
2104 def : Pat<(int_x86_avx_movnt_pd_256 addr:$dst, VR256:$src),
2105 (VMOVNTPDYmr addr:$dst, VR256:$src)>;
2106 def : Pat<(int_x86_avx_movnt_ps_256 addr:$dst, VR256:$src),
2107 (VMOVNTPSYmr addr:$dst, VR256:$src)>;
2109 def MOVNTPSmr_Int : PSI<0x2B, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src),
2110 "movntps\t{$src, $dst|$dst, $src}",
2111 [(int_x86_sse_movnt_ps addr:$dst, VR128:$src)]>;
2112 def MOVNTPDmr_Int : PDI<0x2B, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src),
2113 "movntpd\t{$src, $dst|$dst, $src}",
2114 [(int_x86_sse2_movnt_pd addr:$dst, VR128:$src)]>;
2116 let ExeDomain = SSEPackedInt in
2117 def MOVNTDQmr_Int : PDI<0xE7, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
2118 "movntdq\t{$src, $dst|$dst, $src}",
2119 [(int_x86_sse2_movnt_dq addr:$dst, VR128:$src)]>;
2121 let AddedComplexity = 400 in { // Prefer non-temporal versions
2122 def MOVNTPSmr : PSI<0x2B, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
2123 "movntps\t{$src, $dst|$dst, $src}",
2124 [(alignednontemporalstore (v4f32 VR128:$src), addr:$dst)]>;
2125 def MOVNTPDmr : PDI<0x2B, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
2126 "movntpd\t{$src, $dst|$dst, $src}",
2127 [(alignednontemporalstore(v2f64 VR128:$src), addr:$dst)]>;
2129 def MOVNTDQ_64mr : PDI<0xE7, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
2130 "movntdq\t{$src, $dst|$dst, $src}",
2131 [(alignednontemporalstore (v2f64 VR128:$src), addr:$dst)]>;
2133 let ExeDomain = SSEPackedInt in
2134 def MOVNTDQmr : PDI<0xE7, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
2135 "movntdq\t{$src, $dst|$dst, $src}",
2136 [(alignednontemporalstore (v4f32 VR128:$src), addr:$dst)]>;
2138 // There is no AVX form for instructions below this point
2139 def MOVNTImr : I<0xC3, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src),
2140 "movnti\t{$src, $dst|$dst, $src}",
2141 [(nontemporalstore (i32 GR32:$src), addr:$dst)]>,
2142 TB, Requires<[HasSSE2]>;
2144 def MOVNTI_64mr : RI<0xC3, MRMDestMem, (outs), (ins i64mem:$dst, GR64:$src),
2145 "movnti\t{$src, $dst|$dst, $src}",
2146 [(nontemporalstore (i64 GR64:$src), addr:$dst)]>,
2147 TB, Requires<[HasSSE2]>;
2150 def MOVNTImr_Int : I<0xC3, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src),
2151 "movnti\t{$src, $dst|$dst, $src}",
2152 [(int_x86_sse2_movnt_i addr:$dst, GR32:$src)]>,
2153 TB, Requires<[HasSSE2]>;
2155 //===----------------------------------------------------------------------===//
2156 // SSE 1 & 2 - Misc Instructions (No AVX form)
2157 //===----------------------------------------------------------------------===//
2159 // Prefetch intrinsic.
2160 def PREFETCHT0 : PSI<0x18, MRM1m, (outs), (ins i8mem:$src),
2161 "prefetcht0\t$src", [(prefetch addr:$src, imm, (i32 3))]>;
2162 def PREFETCHT1 : PSI<0x18, MRM2m, (outs), (ins i8mem:$src),
2163 "prefetcht1\t$src", [(prefetch addr:$src, imm, (i32 2))]>;
2164 def PREFETCHT2 : PSI<0x18, MRM3m, (outs), (ins i8mem:$src),
2165 "prefetcht2\t$src", [(prefetch addr:$src, imm, (i32 1))]>;
2166 def PREFETCHNTA : PSI<0x18, MRM0m, (outs), (ins i8mem:$src),
2167 "prefetchnta\t$src", [(prefetch addr:$src, imm, (i32 0))]>;
2169 // Load, store, and memory fence
2170 def SFENCE : I<0xAE, MRM_F8, (outs), (ins), "sfence", [(int_x86_sse_sfence)]>,
2171 TB, Requires<[HasSSE1]>;
2172 def : Pat<(X86SFence), (SFENCE)>;
2174 // Alias instructions that map zero vector to pxor / xorp* for sse.
2175 // We set canFoldAsLoad because this can be converted to a constant-pool
2176 // load of an all-zeros value if folding it would be beneficial.
2177 // FIXME: Change encoding to pseudo!
2178 let isReMaterializable = 1, isAsCheapAsAMove = 1, canFoldAsLoad = 1,
2179 isCodeGenOnly = 1 in {
2180 def V_SET0PS : PSI<0x57, MRMInitReg, (outs VR128:$dst), (ins), "",
2181 [(set VR128:$dst, (v4f32 immAllZerosV))]>;
2182 def V_SET0PD : PDI<0x57, MRMInitReg, (outs VR128:$dst), (ins), "",
2183 [(set VR128:$dst, (v2f64 immAllZerosV))]>;
2184 let ExeDomain = SSEPackedInt in
2185 def V_SET0PI : PDI<0xEF, MRMInitReg, (outs VR128:$dst), (ins), "",
2186 [(set VR128:$dst, (v4i32 immAllZerosV))]>;
2189 let isReMaterializable = 1, isAsCheapAsAMove = 1, canFoldAsLoad = 1,
2190 isCodeGenOnly = 1, Predicates = [HasAVX] in {
2191 def V_SET0PSY : PSI<0x57, MRMInitReg, (outs VR256:$dst), (ins), "",
2192 [(set VR256:$dst, (v8f32 immAllZerosV))]>, VEX_4V;
2193 def V_SET0PDY : PDI<0x57, MRMInitReg, (outs VR256:$dst), (ins), "",
2194 [(set VR256:$dst, (v4f64 immAllZerosV))]>, VEX_4V;
2197 def : Pat<(v2i64 immAllZerosV), (V_SET0PI)>;
2198 def : Pat<(v8i16 immAllZerosV), (V_SET0PI)>;
2199 def : Pat<(v16i8 immAllZerosV), (V_SET0PI)>;
2201 def : Pat<(f32 (vector_extract (v4f32 VR128:$src), (iPTR 0))),
2202 (f32 (EXTRACT_SUBREG (v4f32 VR128:$src), sub_ss))>;
2204 //===----------------------------------------------------------------------===//
2205 // SSE 1 & 2 - Load/Store XCSR register
2206 //===----------------------------------------------------------------------===//
2208 let isAsmParserOnly = 1 in {
2209 def VLDMXCSR : VPSI<0xAE, MRM2m, (outs), (ins i32mem:$src),
2210 "ldmxcsr\t$src", [(int_x86_sse_ldmxcsr addr:$src)]>, VEX;
2211 def VSTMXCSR : VPSI<0xAE, MRM3m, (outs), (ins i32mem:$dst),
2212 "stmxcsr\t$dst", [(int_x86_sse_stmxcsr addr:$dst)]>, VEX;
2215 def LDMXCSR : PSI<0xAE, MRM2m, (outs), (ins i32mem:$src),
2216 "ldmxcsr\t$src", [(int_x86_sse_ldmxcsr addr:$src)]>;
2217 def STMXCSR : PSI<0xAE, MRM3m, (outs), (ins i32mem:$dst),
2218 "stmxcsr\t$dst", [(int_x86_sse_stmxcsr addr:$dst)]>;
2220 //===---------------------------------------------------------------------===//
2221 // SSE2 - Move Aligned/Unaligned Packed Integer Instructions
2222 //===---------------------------------------------------------------------===//
2224 let ExeDomain = SSEPackedInt in { // SSE integer instructions
2226 let isAsmParserOnly = 1 in {
2227 let neverHasSideEffects = 1 in {
2228 def VMOVDQArr : VPDI<0x6F, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
2229 "movdqa\t{$src, $dst|$dst, $src}", []>, VEX;
2230 def VMOVDQAYrr : VPDI<0x6F, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src),
2231 "movdqa\t{$src, $dst|$dst, $src}", []>, VEX;
2233 def VMOVDQUrr : VPDI<0x6F, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
2234 "movdqu\t{$src, $dst|$dst, $src}", []>, XS, VEX;
2235 def VMOVDQUYrr : VPDI<0x6F, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src),
2236 "movdqu\t{$src, $dst|$dst, $src}", []>, XS, VEX;
2238 let canFoldAsLoad = 1, mayLoad = 1 in {
2239 def VMOVDQArm : VPDI<0x6F, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
2240 "movdqa\t{$src, $dst|$dst, $src}", []>, VEX;
2241 def VMOVDQAYrm : VPDI<0x6F, MRMSrcMem, (outs VR256:$dst), (ins i256mem:$src),
2242 "movdqa\t{$src, $dst|$dst, $src}", []>, VEX;
2243 let Predicates = [HasAVX] in {
2244 def VMOVDQUrm : I<0x6F, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
2245 "vmovdqu\t{$src, $dst|$dst, $src}",[]>, XS, VEX;
2246 def VMOVDQUYrm : I<0x6F, MRMSrcMem, (outs VR256:$dst), (ins i256mem:$src),
2247 "vmovdqu\t{$src, $dst|$dst, $src}",[]>, XS, VEX;
2251 let mayStore = 1 in {
2252 def VMOVDQAmr : VPDI<0x7F, MRMDestMem, (outs),
2253 (ins i128mem:$dst, VR128:$src),
2254 "movdqa\t{$src, $dst|$dst, $src}", []>, VEX;
2255 def VMOVDQAYmr : VPDI<0x7F, MRMDestMem, (outs),
2256 (ins i256mem:$dst, VR256:$src),
2257 "movdqa\t{$src, $dst|$dst, $src}", []>, VEX;
2258 let Predicates = [HasAVX] in {
2259 def VMOVDQUmr : I<0x7F, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src),
2260 "vmovdqu\t{$src, $dst|$dst, $src}",[]>, XS, VEX;
2261 def VMOVDQUYmr : I<0x7F, MRMDestMem, (outs), (ins i256mem:$dst, VR256:$src),
2262 "vmovdqu\t{$src, $dst|$dst, $src}",[]>, XS, VEX;
2267 let neverHasSideEffects = 1 in
2268 def MOVDQArr : PDI<0x6F, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
2269 "movdqa\t{$src, $dst|$dst, $src}", []>;
2271 let canFoldAsLoad = 1, mayLoad = 1 in {
2272 def MOVDQArm : PDI<0x6F, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
2273 "movdqa\t{$src, $dst|$dst, $src}",
2274 [/*(set VR128:$dst, (alignedloadv2i64 addr:$src))*/]>;
2275 def MOVDQUrm : I<0x6F, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
2276 "movdqu\t{$src, $dst|$dst, $src}",
2277 [/*(set VR128:$dst, (loadv2i64 addr:$src))*/]>,
2278 XS, Requires<[HasSSE2]>;
2281 let mayStore = 1 in {
2282 def MOVDQAmr : PDI<0x7F, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src),
2283 "movdqa\t{$src, $dst|$dst, $src}",
2284 [/*(alignedstore (v2i64 VR128:$src), addr:$dst)*/]>;
2285 def MOVDQUmr : I<0x7F, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src),
2286 "movdqu\t{$src, $dst|$dst, $src}",
2287 [/*(store (v2i64 VR128:$src), addr:$dst)*/]>,
2288 XS, Requires<[HasSSE2]>;
2291 // Intrinsic forms of MOVDQU load and store
2292 let isAsmParserOnly = 1 in {
2293 let canFoldAsLoad = 1 in
2294 def VMOVDQUrm_Int : I<0x6F, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
2295 "vmovdqu\t{$src, $dst|$dst, $src}",
2296 [(set VR128:$dst, (int_x86_sse2_loadu_dq addr:$src))]>,
2297 XS, VEX, Requires<[HasAVX]>;
2298 def VMOVDQUmr_Int : I<0x7F, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src),
2299 "vmovdqu\t{$src, $dst|$dst, $src}",
2300 [(int_x86_sse2_storeu_dq addr:$dst, VR128:$src)]>,
2301 XS, VEX, Requires<[HasAVX]>;
2304 let canFoldAsLoad = 1 in
2305 def MOVDQUrm_Int : I<0x6F, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
2306 "movdqu\t{$src, $dst|$dst, $src}",
2307 [(set VR128:$dst, (int_x86_sse2_loadu_dq addr:$src))]>,
2308 XS, Requires<[HasSSE2]>;
2309 def MOVDQUmr_Int : I<0x7F, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src),
2310 "movdqu\t{$src, $dst|$dst, $src}",
2311 [(int_x86_sse2_storeu_dq addr:$dst, VR128:$src)]>,
2312 XS, Requires<[HasSSE2]>;
2314 } // ExeDomain = SSEPackedInt
2316 def : Pat<(int_x86_avx_loadu_dq_256 addr:$src), (VMOVDQUYrm addr:$src)>;
2317 def : Pat<(int_x86_avx_storeu_dq_256 addr:$dst, VR256:$src),
2318 (VMOVDQUYmr addr:$dst, VR256:$src)>;
2320 //===---------------------------------------------------------------------===//
2321 // SSE2 - Packed Integer Arithmetic Instructions
2322 //===---------------------------------------------------------------------===//
2324 let ExeDomain = SSEPackedInt in { // SSE integer instructions
2326 multiclass PDI_binop_rm_int<bits<8> opc, string OpcodeStr, Intrinsic IntId,
2327 bit IsCommutable = 0, bit Is2Addr = 1> {
2328 let isCommutable = IsCommutable in
2329 def rr : PDI<opc, MRMSrcReg, (outs VR128:$dst),
2330 (ins VR128:$src1, VR128:$src2),
2332 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
2333 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
2334 [(set VR128:$dst, (IntId VR128:$src1, VR128:$src2))]>;
2335 def rm : PDI<opc, MRMSrcMem, (outs VR128:$dst),
2336 (ins VR128:$src1, i128mem:$src2),
2338 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
2339 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
2340 [(set VR128:$dst, (IntId VR128:$src1,
2341 (bitconvert (memopv2i64 addr:$src2))))]>;
2344 multiclass PDI_binop_rmi_int<bits<8> opc, bits<8> opc2, Format ImmForm,
2345 string OpcodeStr, Intrinsic IntId,
2346 Intrinsic IntId2, bit Is2Addr = 1> {
2347 def rr : PDI<opc, MRMSrcReg, (outs VR128:$dst),
2348 (ins VR128:$src1, VR128:$src2),
2350 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
2351 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
2352 [(set VR128:$dst, (IntId VR128:$src1, VR128:$src2))]>;
2353 def rm : PDI<opc, MRMSrcMem, (outs VR128:$dst),
2354 (ins VR128:$src1, i128mem:$src2),
2356 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
2357 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
2358 [(set VR128:$dst, (IntId VR128:$src1,
2359 (bitconvert (memopv2i64 addr:$src2))))]>;
2360 def ri : PDIi8<opc2, ImmForm, (outs VR128:$dst),
2361 (ins VR128:$src1, i32i8imm:$src2),
2363 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
2364 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
2365 [(set VR128:$dst, (IntId2 VR128:$src1, (i32 imm:$src2)))]>;
2368 /// PDI_binop_rm - Simple SSE2 binary operator.
2369 multiclass PDI_binop_rm<bits<8> opc, string OpcodeStr, SDNode OpNode,
2370 ValueType OpVT, bit IsCommutable = 0, bit Is2Addr = 1> {
2371 let isCommutable = IsCommutable in
2372 def rr : PDI<opc, MRMSrcReg, (outs VR128:$dst),
2373 (ins VR128:$src1, VR128:$src2),
2375 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
2376 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
2377 [(set VR128:$dst, (OpVT (OpNode VR128:$src1, VR128:$src2)))]>;
2378 def rm : PDI<opc, MRMSrcMem, (outs VR128:$dst),
2379 (ins VR128:$src1, i128mem:$src2),
2381 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
2382 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
2383 [(set VR128:$dst, (OpVT (OpNode VR128:$src1,
2384 (bitconvert (memopv2i64 addr:$src2)))))]>;
2387 /// PDI_binop_rm_v2i64 - Simple SSE2 binary operator whose type is v2i64.
2389 /// FIXME: we could eliminate this and use PDI_binop_rm instead if tblgen knew
2390 /// to collapse (bitconvert VT to VT) into its operand.
2392 multiclass PDI_binop_rm_v2i64<bits<8> opc, string OpcodeStr, SDNode OpNode,
2393 bit IsCommutable = 0, bit Is2Addr = 1> {
2394 let isCommutable = IsCommutable in
2395 def rr : PDI<opc, MRMSrcReg, (outs VR128:$dst),
2396 (ins VR128:$src1, VR128:$src2),
2398 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
2399 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
2400 [(set VR128:$dst, (v2i64 (OpNode VR128:$src1, VR128:$src2)))]>;
2401 def rm : PDI<opc, MRMSrcMem, (outs VR128:$dst),
2402 (ins VR128:$src1, i128mem:$src2),
2404 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
2405 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
2406 [(set VR128:$dst, (OpNode VR128:$src1, (memopv2i64 addr:$src2)))]>;
2409 } // ExeDomain = SSEPackedInt
2411 // 128-bit Integer Arithmetic
2413 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
2414 defm VPADDB : PDI_binop_rm<0xFC, "vpaddb", add, v16i8, 1, 0 /*3addr*/>, VEX_4V;
2415 defm VPADDW : PDI_binop_rm<0xFD, "vpaddw", add, v8i16, 1, 0>, VEX_4V;
2416 defm VPADDD : PDI_binop_rm<0xFE, "vpaddd", add, v4i32, 1, 0>, VEX_4V;
2417 defm VPADDQ : PDI_binop_rm_v2i64<0xD4, "vpaddq", add, 1, 0>, VEX_4V;
2418 defm VPMULLW : PDI_binop_rm<0xD5, "vpmullw", mul, v8i16, 1, 0>, VEX_4V;
2419 defm VPSUBB : PDI_binop_rm<0xF8, "vpsubb", sub, v16i8, 0, 0>, VEX_4V;
2420 defm VPSUBW : PDI_binop_rm<0xF9, "vpsubw", sub, v8i16, 0, 0>, VEX_4V;
2421 defm VPSUBD : PDI_binop_rm<0xFA, "vpsubd", sub, v4i32, 0, 0>, VEX_4V;
2422 defm VPSUBQ : PDI_binop_rm_v2i64<0xFB, "vpsubq", sub, 0, 0>, VEX_4V;
2425 defm VPSUBSB : PDI_binop_rm_int<0xE8, "vpsubsb" , int_x86_sse2_psubs_b, 0, 0>,
2427 defm VPSUBSW : PDI_binop_rm_int<0xE9, "vpsubsw" , int_x86_sse2_psubs_w, 0, 0>,
2429 defm VPSUBUSB : PDI_binop_rm_int<0xD8, "vpsubusb", int_x86_sse2_psubus_b, 0, 0>,
2431 defm VPSUBUSW : PDI_binop_rm_int<0xD9, "vpsubusw", int_x86_sse2_psubus_w, 0, 0>,
2433 defm VPADDSB : PDI_binop_rm_int<0xEC, "vpaddsb" , int_x86_sse2_padds_b, 1, 0>,
2435 defm VPADDSW : PDI_binop_rm_int<0xED, "vpaddsw" , int_x86_sse2_padds_w, 1, 0>,
2437 defm VPADDUSB : PDI_binop_rm_int<0xDC, "vpaddusb", int_x86_sse2_paddus_b, 1, 0>,
2439 defm VPADDUSW : PDI_binop_rm_int<0xDD, "vpaddusw", int_x86_sse2_paddus_w, 1, 0>,
2441 defm VPMULHUW : PDI_binop_rm_int<0xE4, "vpmulhuw", int_x86_sse2_pmulhu_w, 1, 0>,
2443 defm VPMULHW : PDI_binop_rm_int<0xE5, "vpmulhw" , int_x86_sse2_pmulh_w, 1, 0>,
2445 defm VPMULUDQ : PDI_binop_rm_int<0xF4, "vpmuludq", int_x86_sse2_pmulu_dq, 1, 0>,
2447 defm VPMADDWD : PDI_binop_rm_int<0xF5, "vpmaddwd", int_x86_sse2_pmadd_wd, 1, 0>,
2449 defm VPAVGB : PDI_binop_rm_int<0xE0, "vpavgb", int_x86_sse2_pavg_b, 1, 0>,
2451 defm VPAVGW : PDI_binop_rm_int<0xE3, "vpavgw", int_x86_sse2_pavg_w, 1, 0>,
2453 defm VPMINUB : PDI_binop_rm_int<0xDA, "vpminub", int_x86_sse2_pminu_b, 1, 0>,
2455 defm VPMINSW : PDI_binop_rm_int<0xEA, "vpminsw", int_x86_sse2_pmins_w, 1, 0>,
2457 defm VPMAXUB : PDI_binop_rm_int<0xDE, "vpmaxub", int_x86_sse2_pmaxu_b, 1, 0>,
2459 defm VPMAXSW : PDI_binop_rm_int<0xEE, "vpmaxsw", int_x86_sse2_pmaxs_w, 1, 0>,
2461 defm VPSADBW : PDI_binop_rm_int<0xF6, "vpsadbw", int_x86_sse2_psad_bw, 1, 0>,
2465 let Constraints = "$src1 = $dst" in {
2466 defm PADDB : PDI_binop_rm<0xFC, "paddb", add, v16i8, 1>;
2467 defm PADDW : PDI_binop_rm<0xFD, "paddw", add, v8i16, 1>;
2468 defm PADDD : PDI_binop_rm<0xFE, "paddd", add, v4i32, 1>;
2469 defm PADDQ : PDI_binop_rm_v2i64<0xD4, "paddq", add, 1>;
2470 defm PMULLW : PDI_binop_rm<0xD5, "pmullw", mul, v8i16, 1>;
2471 defm PSUBB : PDI_binop_rm<0xF8, "psubb", sub, v16i8>;
2472 defm PSUBW : PDI_binop_rm<0xF9, "psubw", sub, v8i16>;
2473 defm PSUBD : PDI_binop_rm<0xFA, "psubd", sub, v4i32>;
2474 defm PSUBQ : PDI_binop_rm_v2i64<0xFB, "psubq", sub>;
2477 defm PSUBSB : PDI_binop_rm_int<0xE8, "psubsb" , int_x86_sse2_psubs_b>;
2478 defm PSUBSW : PDI_binop_rm_int<0xE9, "psubsw" , int_x86_sse2_psubs_w>;
2479 defm PSUBUSB : PDI_binop_rm_int<0xD8, "psubusb", int_x86_sse2_psubus_b>;
2480 defm PSUBUSW : PDI_binop_rm_int<0xD9, "psubusw", int_x86_sse2_psubus_w>;
2481 defm PADDSB : PDI_binop_rm_int<0xEC, "paddsb" , int_x86_sse2_padds_b, 1>;
2482 defm PADDSW : PDI_binop_rm_int<0xED, "paddsw" , int_x86_sse2_padds_w, 1>;
2483 defm PADDUSB : PDI_binop_rm_int<0xDC, "paddusb", int_x86_sse2_paddus_b, 1>;
2484 defm PADDUSW : PDI_binop_rm_int<0xDD, "paddusw", int_x86_sse2_paddus_w, 1>;
2485 defm PMULHUW : PDI_binop_rm_int<0xE4, "pmulhuw", int_x86_sse2_pmulhu_w, 1>;
2486 defm PMULHW : PDI_binop_rm_int<0xE5, "pmulhw" , int_x86_sse2_pmulh_w, 1>;
2487 defm PMULUDQ : PDI_binop_rm_int<0xF4, "pmuludq", int_x86_sse2_pmulu_dq, 1>;
2488 defm PMADDWD : PDI_binop_rm_int<0xF5, "pmaddwd", int_x86_sse2_pmadd_wd, 1>;
2489 defm PAVGB : PDI_binop_rm_int<0xE0, "pavgb", int_x86_sse2_pavg_b, 1>;
2490 defm PAVGW : PDI_binop_rm_int<0xE3, "pavgw", int_x86_sse2_pavg_w, 1>;
2491 defm PMINUB : PDI_binop_rm_int<0xDA, "pminub", int_x86_sse2_pminu_b, 1>;
2492 defm PMINSW : PDI_binop_rm_int<0xEA, "pminsw", int_x86_sse2_pmins_w, 1>;
2493 defm PMAXUB : PDI_binop_rm_int<0xDE, "pmaxub", int_x86_sse2_pmaxu_b, 1>;
2494 defm PMAXSW : PDI_binop_rm_int<0xEE, "pmaxsw", int_x86_sse2_pmaxs_w, 1>;
2495 defm PSADBW : PDI_binop_rm_int<0xF6, "psadbw", int_x86_sse2_psad_bw, 1>;
2497 } // Constraints = "$src1 = $dst"
2499 //===---------------------------------------------------------------------===//
2500 // SSE2 - Packed Integer Logical Instructions
2501 //===---------------------------------------------------------------------===//
2503 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
2504 defm VPSLLW : PDI_binop_rmi_int<0xF1, 0x71, MRM6r, "vpsllw",
2505 int_x86_sse2_psll_w, int_x86_sse2_pslli_w, 0>,
2507 defm VPSLLD : PDI_binop_rmi_int<0xF2, 0x72, MRM6r, "vpslld",
2508 int_x86_sse2_psll_d, int_x86_sse2_pslli_d, 0>,
2510 defm VPSLLQ : PDI_binop_rmi_int<0xF3, 0x73, MRM6r, "vpsllq",
2511 int_x86_sse2_psll_q, int_x86_sse2_pslli_q, 0>,
2514 defm VPSRLW : PDI_binop_rmi_int<0xD1, 0x71, MRM2r, "vpsrlw",
2515 int_x86_sse2_psrl_w, int_x86_sse2_psrli_w, 0>,
2517 defm VPSRLD : PDI_binop_rmi_int<0xD2, 0x72, MRM2r, "vpsrld",
2518 int_x86_sse2_psrl_d, int_x86_sse2_psrli_d, 0>,
2520 defm VPSRLQ : PDI_binop_rmi_int<0xD3, 0x73, MRM2r, "vpsrlq",
2521 int_x86_sse2_psrl_q, int_x86_sse2_psrli_q, 0>,
2524 defm VPSRAW : PDI_binop_rmi_int<0xE1, 0x71, MRM4r, "vpsraw",
2525 int_x86_sse2_psra_w, int_x86_sse2_psrai_w, 0>,
2527 defm VPSRAD : PDI_binop_rmi_int<0xE2, 0x72, MRM4r, "vpsrad",
2528 int_x86_sse2_psra_d, int_x86_sse2_psrai_d, 0>,
2531 defm VPAND : PDI_binop_rm_v2i64<0xDB, "vpand", and, 1, 0>, VEX_4V;
2532 defm VPOR : PDI_binop_rm_v2i64<0xEB, "vpor" , or, 1, 0>, VEX_4V;
2533 defm VPXOR : PDI_binop_rm_v2i64<0xEF, "vpxor", xor, 1, 0>, VEX_4V;
2535 let ExeDomain = SSEPackedInt in {
2536 let neverHasSideEffects = 1 in {
2537 // 128-bit logical shifts.
2538 def VPSLLDQri : PDIi8<0x73, MRM7r,
2539 (outs VR128:$dst), (ins VR128:$src1, i32i8imm:$src2),
2540 "vpslldq\t{$src2, $src1, $dst|$dst, $src1, $src2}", []>,
2542 def VPSRLDQri : PDIi8<0x73, MRM3r,
2543 (outs VR128:$dst), (ins VR128:$src1, i32i8imm:$src2),
2544 "vpsrldq\t{$src2, $src1, $dst|$dst, $src1, $src2}", []>,
2546 // PSRADQri doesn't exist in SSE[1-3].
2548 def VPANDNrr : PDI<0xDF, MRMSrcReg,
2549 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
2550 "vpandn\t{$src2, $src1, $dst|$dst, $src1, $src2}",
2551 [(set VR128:$dst, (v2i64 (and (vnot VR128:$src1),
2552 VR128:$src2)))]>, VEX_4V;
2554 def VPANDNrm : PDI<0xDF, MRMSrcMem,
2555 (outs VR128:$dst), (ins VR128:$src1, i128mem:$src2),
2556 "vpandn\t{$src2, $src1, $dst|$dst, $src1, $src2}",
2557 [(set VR128:$dst, (v2i64 (and (vnot VR128:$src1),
2558 (memopv2i64 addr:$src2))))]>,
2563 let Constraints = "$src1 = $dst" in {
2564 defm PSLLW : PDI_binop_rmi_int<0xF1, 0x71, MRM6r, "psllw",
2565 int_x86_sse2_psll_w, int_x86_sse2_pslli_w>;
2566 defm PSLLD : PDI_binop_rmi_int<0xF2, 0x72, MRM6r, "pslld",
2567 int_x86_sse2_psll_d, int_x86_sse2_pslli_d>;
2568 defm PSLLQ : PDI_binop_rmi_int<0xF3, 0x73, MRM6r, "psllq",
2569 int_x86_sse2_psll_q, int_x86_sse2_pslli_q>;
2571 defm PSRLW : PDI_binop_rmi_int<0xD1, 0x71, MRM2r, "psrlw",
2572 int_x86_sse2_psrl_w, int_x86_sse2_psrli_w>;
2573 defm PSRLD : PDI_binop_rmi_int<0xD2, 0x72, MRM2r, "psrld",
2574 int_x86_sse2_psrl_d, int_x86_sse2_psrli_d>;
2575 defm PSRLQ : PDI_binop_rmi_int<0xD3, 0x73, MRM2r, "psrlq",
2576 int_x86_sse2_psrl_q, int_x86_sse2_psrli_q>;
2578 defm PSRAW : PDI_binop_rmi_int<0xE1, 0x71, MRM4r, "psraw",
2579 int_x86_sse2_psra_w, int_x86_sse2_psrai_w>;
2580 defm PSRAD : PDI_binop_rmi_int<0xE2, 0x72, MRM4r, "psrad",
2581 int_x86_sse2_psra_d, int_x86_sse2_psrai_d>;
2583 defm PAND : PDI_binop_rm_v2i64<0xDB, "pand", and, 1>;
2584 defm POR : PDI_binop_rm_v2i64<0xEB, "por" , or, 1>;
2585 defm PXOR : PDI_binop_rm_v2i64<0xEF, "pxor", xor, 1>;
2587 let ExeDomain = SSEPackedInt in {
2588 let neverHasSideEffects = 1 in {
2589 // 128-bit logical shifts.
2590 def PSLLDQri : PDIi8<0x73, MRM7r,
2591 (outs VR128:$dst), (ins VR128:$src1, i32i8imm:$src2),
2592 "pslldq\t{$src2, $dst|$dst, $src2}", []>;
2593 def PSRLDQri : PDIi8<0x73, MRM3r,
2594 (outs VR128:$dst), (ins VR128:$src1, i32i8imm:$src2),
2595 "psrldq\t{$src2, $dst|$dst, $src2}", []>;
2596 // PSRADQri doesn't exist in SSE[1-3].
2598 def PANDNrr : PDI<0xDF, MRMSrcReg,
2599 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
2600 "pandn\t{$src2, $dst|$dst, $src2}",
2601 [(set VR128:$dst, (v2i64 (and (vnot VR128:$src1),
2604 def PANDNrm : PDI<0xDF, MRMSrcMem,
2605 (outs VR128:$dst), (ins VR128:$src1, i128mem:$src2),
2606 "pandn\t{$src2, $dst|$dst, $src2}",
2607 [(set VR128:$dst, (v2i64 (and (vnot VR128:$src1),
2608 (memopv2i64 addr:$src2))))]>;
2610 } // Constraints = "$src1 = $dst"
2612 let Predicates = [HasAVX] in {
2613 def : Pat<(int_x86_sse2_psll_dq VR128:$src1, imm:$src2),
2614 (v2i64 (VPSLLDQri VR128:$src1, (BYTE_imm imm:$src2)))>;
2615 def : Pat<(int_x86_sse2_psrl_dq VR128:$src1, imm:$src2),
2616 (v2i64 (VPSRLDQri VR128:$src1, (BYTE_imm imm:$src2)))>;
2617 def : Pat<(int_x86_sse2_psll_dq_bs VR128:$src1, imm:$src2),
2618 (v2i64 (VPSLLDQri VR128:$src1, imm:$src2))>;
2619 def : Pat<(int_x86_sse2_psrl_dq_bs VR128:$src1, imm:$src2),
2620 (v2i64 (VPSRLDQri VR128:$src1, imm:$src2))>;
2621 def : Pat<(v2f64 (X86fsrl VR128:$src1, i32immSExt8:$src2)),
2622 (v2f64 (VPSRLDQri VR128:$src1, (BYTE_imm imm:$src2)))>;
2624 // Shift up / down and insert zero's.
2625 def : Pat<(v2i64 (X86vshl VR128:$src, (i8 imm:$amt))),
2626 (v2i64 (VPSLLDQri VR128:$src, (BYTE_imm imm:$amt)))>;
2627 def : Pat<(v2i64 (X86vshr VR128:$src, (i8 imm:$amt))),
2628 (v2i64 (VPSRLDQri VR128:$src, (BYTE_imm imm:$amt)))>;
2631 let Predicates = [HasSSE2] in {
2632 def : Pat<(int_x86_sse2_psll_dq VR128:$src1, imm:$src2),
2633 (v2i64 (PSLLDQri VR128:$src1, (BYTE_imm imm:$src2)))>;
2634 def : Pat<(int_x86_sse2_psrl_dq VR128:$src1, imm:$src2),
2635 (v2i64 (PSRLDQri VR128:$src1, (BYTE_imm imm:$src2)))>;
2636 def : Pat<(int_x86_sse2_psll_dq_bs VR128:$src1, imm:$src2),
2637 (v2i64 (PSLLDQri VR128:$src1, imm:$src2))>;
2638 def : Pat<(int_x86_sse2_psrl_dq_bs VR128:$src1, imm:$src2),
2639 (v2i64 (PSRLDQri VR128:$src1, imm:$src2))>;
2640 def : Pat<(v2f64 (X86fsrl VR128:$src1, i32immSExt8:$src2)),
2641 (v2f64 (PSRLDQri VR128:$src1, (BYTE_imm imm:$src2)))>;
2643 // Shift up / down and insert zero's.
2644 def : Pat<(v2i64 (X86vshl VR128:$src, (i8 imm:$amt))),
2645 (v2i64 (PSLLDQri VR128:$src, (BYTE_imm imm:$amt)))>;
2646 def : Pat<(v2i64 (X86vshr VR128:$src, (i8 imm:$amt))),
2647 (v2i64 (PSRLDQri VR128:$src, (BYTE_imm imm:$amt)))>;
2650 //===---------------------------------------------------------------------===//
2651 // SSE2 - Packed Integer Comparison Instructions
2652 //===---------------------------------------------------------------------===//
2654 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
2655 defm VPCMPEQB : PDI_binop_rm_int<0x74, "vpcmpeqb", int_x86_sse2_pcmpeq_b, 1,
2657 defm VPCMPEQW : PDI_binop_rm_int<0x75, "vpcmpeqw", int_x86_sse2_pcmpeq_w, 1,
2659 defm VPCMPEQD : PDI_binop_rm_int<0x76, "vpcmpeqd", int_x86_sse2_pcmpeq_d, 1,
2661 defm VPCMPGTB : PDI_binop_rm_int<0x64, "vpcmpgtb", int_x86_sse2_pcmpgt_b, 0,
2663 defm VPCMPGTW : PDI_binop_rm_int<0x65, "vpcmpgtw", int_x86_sse2_pcmpgt_w, 0,
2665 defm VPCMPGTD : PDI_binop_rm_int<0x66, "vpcmpgtd", int_x86_sse2_pcmpgt_d, 0,
2669 let Constraints = "$src1 = $dst" in {
2670 defm PCMPEQB : PDI_binop_rm_int<0x74, "pcmpeqb", int_x86_sse2_pcmpeq_b, 1>;
2671 defm PCMPEQW : PDI_binop_rm_int<0x75, "pcmpeqw", int_x86_sse2_pcmpeq_w, 1>;
2672 defm PCMPEQD : PDI_binop_rm_int<0x76, "pcmpeqd", int_x86_sse2_pcmpeq_d, 1>;
2673 defm PCMPGTB : PDI_binop_rm_int<0x64, "pcmpgtb", int_x86_sse2_pcmpgt_b>;
2674 defm PCMPGTW : PDI_binop_rm_int<0x65, "pcmpgtw", int_x86_sse2_pcmpgt_w>;
2675 defm PCMPGTD : PDI_binop_rm_int<0x66, "pcmpgtd", int_x86_sse2_pcmpgt_d>;
2676 } // Constraints = "$src1 = $dst"
2678 def : Pat<(v16i8 (X86pcmpeqb VR128:$src1, VR128:$src2)),
2679 (PCMPEQBrr VR128:$src1, VR128:$src2)>;
2680 def : Pat<(v16i8 (X86pcmpeqb VR128:$src1, (memop addr:$src2))),
2681 (PCMPEQBrm VR128:$src1, addr:$src2)>;
2682 def : Pat<(v8i16 (X86pcmpeqw VR128:$src1, VR128:$src2)),
2683 (PCMPEQWrr VR128:$src1, VR128:$src2)>;
2684 def : Pat<(v8i16 (X86pcmpeqw VR128:$src1, (memop addr:$src2))),
2685 (PCMPEQWrm VR128:$src1, addr:$src2)>;
2686 def : Pat<(v4i32 (X86pcmpeqd VR128:$src1, VR128:$src2)),
2687 (PCMPEQDrr VR128:$src1, VR128:$src2)>;
2688 def : Pat<(v4i32 (X86pcmpeqd VR128:$src1, (memop addr:$src2))),
2689 (PCMPEQDrm VR128:$src1, addr:$src2)>;
2691 def : Pat<(v16i8 (X86pcmpgtb VR128:$src1, VR128:$src2)),
2692 (PCMPGTBrr VR128:$src1, VR128:$src2)>;
2693 def : Pat<(v16i8 (X86pcmpgtb VR128:$src1, (memop addr:$src2))),
2694 (PCMPGTBrm VR128:$src1, addr:$src2)>;
2695 def : Pat<(v8i16 (X86pcmpgtw VR128:$src1, VR128:$src2)),
2696 (PCMPGTWrr VR128:$src1, VR128:$src2)>;
2697 def : Pat<(v8i16 (X86pcmpgtw VR128:$src1, (memop addr:$src2))),
2698 (PCMPGTWrm VR128:$src1, addr:$src2)>;
2699 def : Pat<(v4i32 (X86pcmpgtd VR128:$src1, VR128:$src2)),
2700 (PCMPGTDrr VR128:$src1, VR128:$src2)>;
2701 def : Pat<(v4i32 (X86pcmpgtd VR128:$src1, (memop addr:$src2))),
2702 (PCMPGTDrm VR128:$src1, addr:$src2)>;
2704 //===---------------------------------------------------------------------===//
2705 // SSE2 - Packed Integer Pack Instructions
2706 //===---------------------------------------------------------------------===//
2708 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
2709 defm VPACKSSWB : PDI_binop_rm_int<0x63, "vpacksswb", int_x86_sse2_packsswb_128,
2711 defm VPACKSSDW : PDI_binop_rm_int<0x6B, "vpackssdw", int_x86_sse2_packssdw_128,
2713 defm VPACKUSWB : PDI_binop_rm_int<0x67, "vpackuswb", int_x86_sse2_packuswb_128,
2717 let Constraints = "$src1 = $dst" in {
2718 defm PACKSSWB : PDI_binop_rm_int<0x63, "packsswb", int_x86_sse2_packsswb_128>;
2719 defm PACKSSDW : PDI_binop_rm_int<0x6B, "packssdw", int_x86_sse2_packssdw_128>;
2720 defm PACKUSWB : PDI_binop_rm_int<0x67, "packuswb", int_x86_sse2_packuswb_128>;
2721 } // Constraints = "$src1 = $dst"
2723 //===---------------------------------------------------------------------===//
2724 // SSE2 - Packed Integer Shuffle Instructions
2725 //===---------------------------------------------------------------------===//
2727 let ExeDomain = SSEPackedInt in {
2728 multiclass sse2_pshuffle<string OpcodeStr, ValueType vt, PatFrag pshuf_frag,
2730 def ri : Ii8<0x70, MRMSrcReg,
2731 (outs VR128:$dst), (ins VR128:$src1, i8imm:$src2),
2732 !strconcat(OpcodeStr,
2733 "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
2734 [(set VR128:$dst, (vt (pshuf_frag:$src2 VR128:$src1,
2736 def mi : Ii8<0x70, MRMSrcMem,
2737 (outs VR128:$dst), (ins i128mem:$src1, i8imm:$src2),
2738 !strconcat(OpcodeStr,
2739 "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
2740 [(set VR128:$dst, (vt (pshuf_frag:$src2
2741 (bc_frag (memopv2i64 addr:$src1)),
2744 } // ExeDomain = SSEPackedInt
2746 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
2747 let AddedComplexity = 5 in
2748 defm VPSHUFD : sse2_pshuffle<"vpshufd", v4i32, pshufd, bc_v4i32>, OpSize,
2751 // SSE2 with ImmT == Imm8 and XS prefix.
2752 defm VPSHUFHW : sse2_pshuffle<"vpshufhw", v8i16, pshufhw, bc_v8i16>, XS,
2755 // SSE2 with ImmT == Imm8 and XD prefix.
2756 defm VPSHUFLW : sse2_pshuffle<"vpshuflw", v8i16, pshuflw, bc_v8i16>, XD,
2760 let Predicates = [HasSSE2] in {
2761 let AddedComplexity = 5 in
2762 defm PSHUFD : sse2_pshuffle<"pshufd", v4i32, pshufd, bc_v4i32>, TB, OpSize;
2764 // SSE2 with ImmT == Imm8 and XS prefix.
2765 defm PSHUFHW : sse2_pshuffle<"pshufhw", v8i16, pshufhw, bc_v8i16>, XS;
2767 // SSE2 with ImmT == Imm8 and XD prefix.
2768 defm PSHUFLW : sse2_pshuffle<"pshuflw", v8i16, pshuflw, bc_v8i16>, XD;
2771 //===---------------------------------------------------------------------===//
2772 // SSE2 - Packed Integer Unpack Instructions
2773 //===---------------------------------------------------------------------===//
2775 let ExeDomain = SSEPackedInt in {
2776 multiclass sse2_unpack<bits<8> opc, string OpcodeStr, ValueType vt,
2777 PatFrag unp_frag, PatFrag bc_frag, bit Is2Addr = 1> {
2778 def rr : PDI<opc, MRMSrcReg,
2779 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
2781 !strconcat(OpcodeStr,"\t{$src2, $dst|$dst, $src2}"),
2782 !strconcat(OpcodeStr,"\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
2783 [(set VR128:$dst, (vt (unp_frag VR128:$src1, VR128:$src2)))]>;
2784 def rm : PDI<opc, MRMSrcMem,
2785 (outs VR128:$dst), (ins VR128:$src1, i128mem:$src2),
2787 !strconcat(OpcodeStr,"\t{$src2, $dst|$dst, $src2}"),
2788 !strconcat(OpcodeStr,"\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
2789 [(set VR128:$dst, (unp_frag VR128:$src1,
2790 (bc_frag (memopv2i64
2794 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
2795 defm VPUNPCKLBW : sse2_unpack<0x60, "vpunpcklbw", v16i8, unpckl, bc_v16i8,
2797 defm VPUNPCKLWD : sse2_unpack<0x61, "vpunpcklwd", v8i16, unpckl, bc_v8i16,
2799 defm VPUNPCKLDQ : sse2_unpack<0x62, "vpunpckldq", v4i32, unpckl, bc_v4i32,
2802 /// FIXME: we could eliminate this and use sse2_unpack instead if tblgen
2803 /// knew to collapse (bitconvert VT to VT) into its operand.
2804 def VPUNPCKLQDQrr : PDI<0x6C, MRMSrcReg,
2805 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
2806 "vpunpcklqdq\t{$src2, $src1, $dst|$dst, $src1, $src2}",
2808 (v2i64 (unpckl VR128:$src1, VR128:$src2)))]>, VEX_4V;
2809 def VPUNPCKLQDQrm : PDI<0x6C, MRMSrcMem,
2810 (outs VR128:$dst), (ins VR128:$src1, i128mem:$src2),
2811 "vpunpcklqdq\t{$src2, $src1, $dst|$dst, $src1, $src2}",
2813 (v2i64 (unpckl VR128:$src1,
2814 (memopv2i64 addr:$src2))))]>, VEX_4V;
2816 defm VPUNPCKHBW : sse2_unpack<0x68, "vpunpckhbw", v16i8, unpckh, bc_v16i8,
2818 defm VPUNPCKHWD : sse2_unpack<0x69, "vpunpckhwd", v8i16, unpckh, bc_v8i16,
2820 defm VPUNPCKHDQ : sse2_unpack<0x6A, "vpunpckhdq", v4i32, unpckh, bc_v4i32,
2823 /// FIXME: we could eliminate this and use sse2_unpack instead if tblgen
2824 /// knew to collapse (bitconvert VT to VT) into its operand.
2825 def VPUNPCKHQDQrr : PDI<0x6D, MRMSrcReg,
2826 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
2827 "vpunpckhqdq\t{$src2, $src1, $dst|$dst, $src1, $src2}",
2829 (v2i64 (unpckh VR128:$src1, VR128:$src2)))]>, VEX_4V;
2830 def VPUNPCKHQDQrm : PDI<0x6D, MRMSrcMem,
2831 (outs VR128:$dst), (ins VR128:$src1, i128mem:$src2),
2832 "vpunpckhqdq\t{$src2, $src1, $dst|$dst, $src1, $src2}",
2834 (v2i64 (unpckh VR128:$src1,
2835 (memopv2i64 addr:$src2))))]>, VEX_4V;
2838 let Constraints = "$src1 = $dst" in {
2839 defm PUNPCKLBW : sse2_unpack<0x60, "punpcklbw", v16i8, unpckl, bc_v16i8>;
2840 defm PUNPCKLWD : sse2_unpack<0x61, "punpcklwd", v8i16, unpckl, bc_v8i16>;
2841 defm PUNPCKLDQ : sse2_unpack<0x62, "punpckldq", v4i32, unpckl, bc_v4i32>;
2843 /// FIXME: we could eliminate this and use sse2_unpack instead if tblgen
2844 /// knew to collapse (bitconvert VT to VT) into its operand.
2845 def PUNPCKLQDQrr : PDI<0x6C, MRMSrcReg,
2846 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
2847 "punpcklqdq\t{$src2, $dst|$dst, $src2}",
2849 (v2i64 (unpckl VR128:$src1, VR128:$src2)))]>;
2850 def PUNPCKLQDQrm : PDI<0x6C, MRMSrcMem,
2851 (outs VR128:$dst), (ins VR128:$src1, i128mem:$src2),
2852 "punpcklqdq\t{$src2, $dst|$dst, $src2}",
2854 (v2i64 (unpckl VR128:$src1,
2855 (memopv2i64 addr:$src2))))]>;
2857 defm PUNPCKHBW : sse2_unpack<0x68, "punpckhbw", v16i8, unpckh, bc_v16i8>;
2858 defm PUNPCKHWD : sse2_unpack<0x69, "punpckhwd", v8i16, unpckh, bc_v8i16>;
2859 defm PUNPCKHDQ : sse2_unpack<0x6A, "punpckhdq", v4i32, unpckh, bc_v4i32>;
2861 /// FIXME: we could eliminate this and use sse2_unpack instead if tblgen
2862 /// knew to collapse (bitconvert VT to VT) into its operand.
2863 def PUNPCKHQDQrr : PDI<0x6D, MRMSrcReg,
2864 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
2865 "punpckhqdq\t{$src2, $dst|$dst, $src2}",
2867 (v2i64 (unpckh VR128:$src1, VR128:$src2)))]>;
2868 def PUNPCKHQDQrm : PDI<0x6D, MRMSrcMem,
2869 (outs VR128:$dst), (ins VR128:$src1, i128mem:$src2),
2870 "punpckhqdq\t{$src2, $dst|$dst, $src2}",
2872 (v2i64 (unpckh VR128:$src1,
2873 (memopv2i64 addr:$src2))))]>;
2876 } // ExeDomain = SSEPackedInt
2878 //===---------------------------------------------------------------------===//
2879 // SSE2 - Packed Integer Extract and Insert
2880 //===---------------------------------------------------------------------===//
2882 let ExeDomain = SSEPackedInt in {
2883 multiclass sse2_pinsrw<bit Is2Addr = 1> {
2884 def rri : Ii8<0xC4, MRMSrcReg,
2885 (outs VR128:$dst), (ins VR128:$src1,
2886 GR32:$src2, i32i8imm:$src3),
2888 "pinsrw\t{$src3, $src2, $dst|$dst, $src2, $src3}",
2889 "vpinsrw\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
2891 (X86pinsrw VR128:$src1, GR32:$src2, imm:$src3))]>;
2892 def rmi : Ii8<0xC4, MRMSrcMem,
2893 (outs VR128:$dst), (ins VR128:$src1,
2894 i16mem:$src2, i32i8imm:$src3),
2896 "pinsrw\t{$src3, $src2, $dst|$dst, $src2, $src3}",
2897 "vpinsrw\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
2899 (X86pinsrw VR128:$src1, (extloadi16 addr:$src2),
2904 let isAsmParserOnly = 1, Predicates = [HasAVX] in
2905 def VPEXTRWri : Ii8<0xC5, MRMSrcReg,
2906 (outs GR32:$dst), (ins VR128:$src1, i32i8imm:$src2),
2907 "vpextrw\t{$src2, $src1, $dst|$dst, $src1, $src2}",
2908 [(set GR32:$dst, (X86pextrw (v8i16 VR128:$src1),
2909 imm:$src2))]>, OpSize, VEX;
2910 def PEXTRWri : PDIi8<0xC5, MRMSrcReg,
2911 (outs GR32:$dst), (ins VR128:$src1, i32i8imm:$src2),
2912 "pextrw\t{$src2, $src1, $dst|$dst, $src1, $src2}",
2913 [(set GR32:$dst, (X86pextrw (v8i16 VR128:$src1),
2917 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
2918 defm VPINSRW : sse2_pinsrw<0>, OpSize, VEX_4V;
2919 def VPINSRWrr64i : Ii8<0xC4, MRMSrcReg, (outs VR128:$dst),
2920 (ins VR128:$src1, GR64:$src2, i32i8imm:$src3),
2921 "vpinsrw\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}",
2922 []>, OpSize, VEX_4V;
2925 let Constraints = "$src1 = $dst" in
2926 defm PINSRW : sse2_pinsrw, TB, OpSize, Requires<[HasSSE2]>;
2928 } // ExeDomain = SSEPackedInt
2930 //===---------------------------------------------------------------------===//
2931 // SSE2 - Packed Mask Creation
2932 //===---------------------------------------------------------------------===//
2934 let ExeDomain = SSEPackedInt in {
2936 let isAsmParserOnly = 1 in {
2937 def VPMOVMSKBrr : VPDI<0xD7, MRMSrcReg, (outs GR32:$dst), (ins VR128:$src),
2938 "pmovmskb\t{$src, $dst|$dst, $src}",
2939 [(set GR32:$dst, (int_x86_sse2_pmovmskb_128 VR128:$src))]>, VEX;
2940 def VPMOVMSKBr64r : VPDI<0xD7, MRMSrcReg, (outs GR64:$dst), (ins VR128:$src),
2941 "pmovmskb\t{$src, $dst|$dst, $src}", []>, VEX;
2943 def PMOVMSKBrr : PDI<0xD7, MRMSrcReg, (outs GR32:$dst), (ins VR128:$src),
2944 "pmovmskb\t{$src, $dst|$dst, $src}",
2945 [(set GR32:$dst, (int_x86_sse2_pmovmskb_128 VR128:$src))]>;
2947 } // ExeDomain = SSEPackedInt
2949 //===---------------------------------------------------------------------===//
2950 // SSE2 - Conditional Store
2951 //===---------------------------------------------------------------------===//
2953 let ExeDomain = SSEPackedInt in {
2955 let isAsmParserOnly = 1 in {
2957 def VMASKMOVDQU : VPDI<0xF7, MRMSrcReg, (outs),
2958 (ins VR128:$src, VR128:$mask),
2959 "maskmovdqu\t{$mask, $src|$src, $mask}",
2960 [(int_x86_sse2_maskmov_dqu VR128:$src, VR128:$mask, EDI)]>, VEX;
2962 def VMASKMOVDQU64 : VPDI<0xF7, MRMSrcReg, (outs),
2963 (ins VR128:$src, VR128:$mask),
2964 "maskmovdqu\t{$mask, $src|$src, $mask}",
2965 [(int_x86_sse2_maskmov_dqu VR128:$src, VR128:$mask, RDI)]>, VEX;
2969 def MASKMOVDQU : PDI<0xF7, MRMSrcReg, (outs), (ins VR128:$src, VR128:$mask),
2970 "maskmovdqu\t{$mask, $src|$src, $mask}",
2971 [(int_x86_sse2_maskmov_dqu VR128:$src, VR128:$mask, EDI)]>;
2973 def MASKMOVDQU64 : PDI<0xF7, MRMSrcReg, (outs), (ins VR128:$src, VR128:$mask),
2974 "maskmovdqu\t{$mask, $src|$src, $mask}",
2975 [(int_x86_sse2_maskmov_dqu VR128:$src, VR128:$mask, RDI)]>;
2977 } // ExeDomain = SSEPackedInt
2979 //===---------------------------------------------------------------------===//
2980 // SSE2 - Move Doubleword
2981 //===---------------------------------------------------------------------===//
2983 // Move Int Doubleword to Packed Double Int
2984 let isAsmParserOnly = 1 in {
2985 def VMOVDI2PDIrr : VPDI<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR32:$src),
2986 "movd\t{$src, $dst|$dst, $src}",
2988 (v4i32 (scalar_to_vector GR32:$src)))]>, VEX;
2989 def VMOVDI2PDIrm : VPDI<0x6E, MRMSrcMem, (outs VR128:$dst), (ins i32mem:$src),
2990 "movd\t{$src, $dst|$dst, $src}",
2992 (v4i32 (scalar_to_vector (loadi32 addr:$src))))]>,
2995 def MOVDI2PDIrr : PDI<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR32:$src),
2996 "movd\t{$src, $dst|$dst, $src}",
2998 (v4i32 (scalar_to_vector GR32:$src)))]>;
2999 def MOVDI2PDIrm : PDI<0x6E, MRMSrcMem, (outs VR128:$dst), (ins i32mem:$src),
3000 "movd\t{$src, $dst|$dst, $src}",
3002 (v4i32 (scalar_to_vector (loadi32 addr:$src))))]>;
3005 // Move Int Doubleword to Single Scalar
3006 let isAsmParserOnly = 1 in {
3007 def VMOVDI2SSrr : VPDI<0x6E, MRMSrcReg, (outs FR32:$dst), (ins GR32:$src),
3008 "movd\t{$src, $dst|$dst, $src}",
3009 [(set FR32:$dst, (bitconvert GR32:$src))]>, VEX;
3011 def VMOVDI2SSrm : VPDI<0x6E, MRMSrcMem, (outs FR32:$dst), (ins i32mem:$src),
3012 "movd\t{$src, $dst|$dst, $src}",
3013 [(set FR32:$dst, (bitconvert (loadi32 addr:$src)))]>,
3016 def MOVDI2SSrr : PDI<0x6E, MRMSrcReg, (outs FR32:$dst), (ins GR32:$src),
3017 "movd\t{$src, $dst|$dst, $src}",
3018 [(set FR32:$dst, (bitconvert GR32:$src))]>;
3020 def MOVDI2SSrm : PDI<0x6E, MRMSrcMem, (outs FR32:$dst), (ins i32mem:$src),
3021 "movd\t{$src, $dst|$dst, $src}",
3022 [(set FR32:$dst, (bitconvert (loadi32 addr:$src)))]>;
3024 // Move Packed Doubleword Int to Packed Double Int
3025 let isAsmParserOnly = 1 in {
3026 def VMOVPDI2DIrr : VPDI<0x7E, MRMDestReg, (outs GR32:$dst), (ins VR128:$src),
3027 "movd\t{$src, $dst|$dst, $src}",
3028 [(set GR32:$dst, (vector_extract (v4i32 VR128:$src),
3030 def VMOVPDI2DImr : VPDI<0x7E, MRMDestMem, (outs),
3031 (ins i32mem:$dst, VR128:$src),
3032 "movd\t{$src, $dst|$dst, $src}",
3033 [(store (i32 (vector_extract (v4i32 VR128:$src),
3034 (iPTR 0))), addr:$dst)]>, VEX;
3036 def MOVPDI2DIrr : PDI<0x7E, MRMDestReg, (outs GR32:$dst), (ins VR128:$src),
3037 "movd\t{$src, $dst|$dst, $src}",
3038 [(set GR32:$dst, (vector_extract (v4i32 VR128:$src),
3040 def MOVPDI2DImr : PDI<0x7E, MRMDestMem, (outs), (ins i32mem:$dst, VR128:$src),
3041 "movd\t{$src, $dst|$dst, $src}",
3042 [(store (i32 (vector_extract (v4i32 VR128:$src),
3043 (iPTR 0))), addr:$dst)]>;
3045 // Move Scalar Single to Double Int
3046 let isAsmParserOnly = 1 in {
3047 def VMOVSS2DIrr : VPDI<0x7E, MRMDestReg, (outs GR32:$dst), (ins FR32:$src),
3048 "movd\t{$src, $dst|$dst, $src}",
3049 [(set GR32:$dst, (bitconvert FR32:$src))]>, VEX;
3050 def VMOVSS2DImr : VPDI<0x7E, MRMDestMem, (outs), (ins i32mem:$dst, FR32:$src),
3051 "movd\t{$src, $dst|$dst, $src}",
3052 [(store (i32 (bitconvert FR32:$src)), addr:$dst)]>, VEX;
3054 def MOVSS2DIrr : PDI<0x7E, MRMDestReg, (outs GR32:$dst), (ins FR32:$src),
3055 "movd\t{$src, $dst|$dst, $src}",
3056 [(set GR32:$dst, (bitconvert FR32:$src))]>;
3057 def MOVSS2DImr : PDI<0x7E, MRMDestMem, (outs), (ins i32mem:$dst, FR32:$src),
3058 "movd\t{$src, $dst|$dst, $src}",
3059 [(store (i32 (bitconvert FR32:$src)), addr:$dst)]>;
3061 // movd / movq to XMM register zero-extends
3062 let AddedComplexity = 15, isAsmParserOnly = 1 in {
3063 def VMOVZDI2PDIrr : VPDI<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR32:$src),
3064 "movd\t{$src, $dst|$dst, $src}",
3065 [(set VR128:$dst, (v4i32 (X86vzmovl
3066 (v4i32 (scalar_to_vector GR32:$src)))))]>,
3068 def VMOVZQI2PQIrr : VPDI<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR64:$src),
3069 "mov{d|q}\t{$src, $dst|$dst, $src}", // X86-64 only
3070 [(set VR128:$dst, (v2i64 (X86vzmovl
3071 (v2i64 (scalar_to_vector GR64:$src)))))]>,
3074 let AddedComplexity = 15 in {
3075 def MOVZDI2PDIrr : PDI<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR32:$src),
3076 "movd\t{$src, $dst|$dst, $src}",
3077 [(set VR128:$dst, (v4i32 (X86vzmovl
3078 (v4i32 (scalar_to_vector GR32:$src)))))]>;
3079 def MOVZQI2PQIrr : RPDI<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR64:$src),
3080 "mov{d|q}\t{$src, $dst|$dst, $src}", // X86-64 only
3081 [(set VR128:$dst, (v2i64 (X86vzmovl
3082 (v2i64 (scalar_to_vector GR64:$src)))))]>;
3085 let AddedComplexity = 20 in {
3086 let isAsmParserOnly = 1 in
3087 def VMOVZDI2PDIrm : VPDI<0x6E, MRMSrcMem, (outs VR128:$dst), (ins i32mem:$src),
3088 "movd\t{$src, $dst|$dst, $src}",
3090 (v4i32 (X86vzmovl (v4i32 (scalar_to_vector
3091 (loadi32 addr:$src))))))]>,
3093 def MOVZDI2PDIrm : PDI<0x6E, MRMSrcMem, (outs VR128:$dst), (ins i32mem:$src),
3094 "movd\t{$src, $dst|$dst, $src}",
3096 (v4i32 (X86vzmovl (v4i32 (scalar_to_vector
3097 (loadi32 addr:$src))))))]>;
3099 def : Pat<(v4i32 (X86vzmovl (loadv4i32 addr:$src))),
3100 (MOVZDI2PDIrm addr:$src)>;
3101 def : Pat<(v4i32 (X86vzmovl (bc_v4i32 (loadv4f32 addr:$src)))),
3102 (MOVZDI2PDIrm addr:$src)>;
3103 def : Pat<(v4i32 (X86vzmovl (bc_v4i32 (loadv2i64 addr:$src)))),
3104 (MOVZDI2PDIrm addr:$src)>;
3107 //===---------------------------------------------------------------------===//
3108 // SSE2 - Move Quadword
3109 //===---------------------------------------------------------------------===//
3111 // Move Quadword Int to Packed Quadword Int
3112 let isAsmParserOnly = 1 in
3113 def VMOVQI2PQIrm : I<0x7E, MRMSrcMem, (outs VR128:$dst), (ins i64mem:$src),
3114 "vmovq\t{$src, $dst|$dst, $src}",
3116 (v2i64 (scalar_to_vector (loadi64 addr:$src))))]>, XS,
3117 VEX, Requires<[HasAVX]>;
3118 def MOVQI2PQIrm : I<0x7E, MRMSrcMem, (outs VR128:$dst), (ins i64mem:$src),
3119 "movq\t{$src, $dst|$dst, $src}",
3121 (v2i64 (scalar_to_vector (loadi64 addr:$src))))]>, XS,
3122 Requires<[HasSSE2]>; // SSE2 instruction with XS Prefix
3124 // Move Packed Quadword Int to Quadword Int
3125 let isAsmParserOnly = 1 in
3126 def VMOVPQI2QImr : VPDI<0xD6, MRMDestMem, (outs), (ins i64mem:$dst, VR128:$src),
3127 "movq\t{$src, $dst|$dst, $src}",
3128 [(store (i64 (vector_extract (v2i64 VR128:$src),
3129 (iPTR 0))), addr:$dst)]>, VEX;
3130 def MOVPQI2QImr : PDI<0xD6, MRMDestMem, (outs), (ins i64mem:$dst, VR128:$src),
3131 "movq\t{$src, $dst|$dst, $src}",
3132 [(store (i64 (vector_extract (v2i64 VR128:$src),
3133 (iPTR 0))), addr:$dst)]>;
3135 def : Pat<(f64 (vector_extract (v2f64 VR128:$src), (iPTR 0))),
3136 (f64 (EXTRACT_SUBREG (v2f64 VR128:$src), sub_sd))>;
3138 // Store / copy lower 64-bits of a XMM register.
3139 let isAsmParserOnly = 1 in
3140 def VMOVLQ128mr : VPDI<0xD6, MRMDestMem, (outs), (ins i64mem:$dst, VR128:$src),
3141 "movq\t{$src, $dst|$dst, $src}",
3142 [(int_x86_sse2_storel_dq addr:$dst, VR128:$src)]>, VEX;
3143 def MOVLQ128mr : PDI<0xD6, MRMDestMem, (outs), (ins i64mem:$dst, VR128:$src),
3144 "movq\t{$src, $dst|$dst, $src}",
3145 [(int_x86_sse2_storel_dq addr:$dst, VR128:$src)]>;
3147 let AddedComplexity = 20, isAsmParserOnly = 1 in
3148 def VMOVZQI2PQIrm : I<0x7E, MRMSrcMem, (outs VR128:$dst), (ins i64mem:$src),
3149 "vmovq\t{$src, $dst|$dst, $src}",
3151 (v2i64 (X86vzmovl (v2i64 (scalar_to_vector
3152 (loadi64 addr:$src))))))]>,
3153 XS, VEX, Requires<[HasAVX]>;
3155 let AddedComplexity = 20 in {
3156 def MOVZQI2PQIrm : I<0x7E, MRMSrcMem, (outs VR128:$dst), (ins i64mem:$src),
3157 "movq\t{$src, $dst|$dst, $src}",
3159 (v2i64 (X86vzmovl (v2i64 (scalar_to_vector
3160 (loadi64 addr:$src))))))]>,
3161 XS, Requires<[HasSSE2]>;
3163 def : Pat<(v2i64 (X86vzmovl (loadv2i64 addr:$src))),
3164 (MOVZQI2PQIrm addr:$src)>;
3165 def : Pat<(v2i64 (X86vzmovl (bc_v2i64 (loadv4f32 addr:$src)))),
3166 (MOVZQI2PQIrm addr:$src)>;
3167 def : Pat<(v2i64 (X86vzload addr:$src)), (MOVZQI2PQIrm addr:$src)>;
3170 // Moving from XMM to XMM and clear upper 64 bits. Note, there is a bug in
3171 // IA32 document. movq xmm1, xmm2 does clear the high bits.
3172 let isAsmParserOnly = 1, AddedComplexity = 15 in
3173 def VMOVZPQILo2PQIrr : I<0x7E, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
3174 "vmovq\t{$src, $dst|$dst, $src}",
3175 [(set VR128:$dst, (v2i64 (X86vzmovl (v2i64 VR128:$src))))]>,
3176 XS, VEX, Requires<[HasAVX]>;
3177 let AddedComplexity = 15 in
3178 def MOVZPQILo2PQIrr : I<0x7E, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
3179 "movq\t{$src, $dst|$dst, $src}",
3180 [(set VR128:$dst, (v2i64 (X86vzmovl (v2i64 VR128:$src))))]>,
3181 XS, Requires<[HasSSE2]>;
3183 let AddedComplexity = 20, isAsmParserOnly = 1 in
3184 def VMOVZPQILo2PQIrm : I<0x7E, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
3185 "vmovq\t{$src, $dst|$dst, $src}",
3186 [(set VR128:$dst, (v2i64 (X86vzmovl
3187 (loadv2i64 addr:$src))))]>,
3188 XS, VEX, Requires<[HasAVX]>;
3189 let AddedComplexity = 20 in {
3190 def MOVZPQILo2PQIrm : I<0x7E, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
3191 "movq\t{$src, $dst|$dst, $src}",
3192 [(set VR128:$dst, (v2i64 (X86vzmovl
3193 (loadv2i64 addr:$src))))]>,
3194 XS, Requires<[HasSSE2]>;
3196 def : Pat<(v2i64 (X86vzmovl (bc_v2i64 (loadv4i32 addr:$src)))),
3197 (MOVZPQILo2PQIrm addr:$src)>;
3200 // Instructions to match in the assembler
3201 let isAsmParserOnly = 1 in {
3202 def VMOVQs64rr : VPDI<0x6E, MRMSrcReg, (outs VR128:$dst), (ins GR64:$src),
3203 "movq\t{$src, $dst|$dst, $src}", []>, VEX, VEX_W;
3204 def VMOVQd64rr : VPDI<0x7E, MRMDestReg, (outs GR64:$dst), (ins VR128:$src),
3205 "movq\t{$src, $dst|$dst, $src}", []>, VEX, VEX_W;
3206 // Recognize "movd" with GR64 destination, but encode as a "movq"
3207 def VMOVQd64rr_alt : VPDI<0x7E, MRMDestReg, (outs GR64:$dst), (ins VR128:$src),
3208 "movd\t{$src, $dst|$dst, $src}", []>, VEX, VEX_W;
3211 // Instructions for the disassembler
3212 // xr = XMM register
3215 let isAsmParserOnly = 1, Predicates = [HasAVX] in
3216 def VMOVQxrxr: I<0x7E, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
3217 "vmovq\t{$src, $dst|$dst, $src}", []>, VEX, XS;
3218 def MOVQxrxr : I<0x7E, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
3219 "movq\t{$src, $dst|$dst, $src}", []>, XS;
3221 //===---------------------------------------------------------------------===//
3222 // SSE2 - Misc Instructions
3223 //===---------------------------------------------------------------------===//
3226 def CLFLUSH : I<0xAE, MRM7m, (outs), (ins i8mem:$src),
3227 "clflush\t$src", [(int_x86_sse2_clflush addr:$src)]>,
3228 TB, Requires<[HasSSE2]>;
3230 // Load, store, and memory fence
3231 def LFENCE : I<0xAE, MRM_E8, (outs), (ins),
3232 "lfence", [(int_x86_sse2_lfence)]>, TB, Requires<[HasSSE2]>;
3233 def MFENCE : I<0xAE, MRM_F0, (outs), (ins),
3234 "mfence", [(int_x86_sse2_mfence)]>, TB, Requires<[HasSSE2]>;
3235 def : Pat<(X86LFence), (LFENCE)>;
3236 def : Pat<(X86MFence), (MFENCE)>;
3239 // Pause. This "instruction" is encoded as "rep; nop", so even though it
3240 // was introduced with SSE2, it's backward compatible.
3241 def PAUSE : I<0x90, RawFrm, (outs), (ins), "pause", []>, REP;
3243 // Alias instructions that map zero vector to pxor / xorp* for sse.
3244 // We set canFoldAsLoad because this can be converted to a constant-pool
3245 // load of an all-ones value if folding it would be beneficial.
3246 let isReMaterializable = 1, isAsCheapAsAMove = 1, canFoldAsLoad = 1,
3247 isCodeGenOnly = 1, ExeDomain = SSEPackedInt in
3248 // FIXME: Change encoding to pseudo.
3249 def V_SETALLONES : PDI<0x76, MRMInitReg, (outs VR128:$dst), (ins), "",
3250 [(set VR128:$dst, (v4i32 immAllOnesV))]>;
3252 //===---------------------------------------------------------------------===//
3253 // SSE3 - Conversion Instructions
3254 //===---------------------------------------------------------------------===//
3256 // Convert Packed Double FP to Packed DW Integers
3257 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
3258 // The assembler can recognize rr 256-bit instructions by seeing a ymm
3259 // register, but the same isn't true when using memory operands instead.
3260 // Provide other assembly rr and rm forms to address this explicitly.
3261 def VCVTPD2DQrr : S3DI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
3262 "vcvtpd2dq\t{$src, $dst|$dst, $src}", []>, VEX;
3263 def VCVTPD2DQXrYr : S3DI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR256:$src),
3264 "vcvtpd2dq\t{$src, $dst|$dst, $src}", []>, VEX;
3267 def VCVTPD2DQXrr : S3DI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
3268 "vcvtpd2dqx\t{$src, $dst|$dst, $src}", []>, VEX;
3269 def VCVTPD2DQXrm : S3DI<0xE6, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
3270 "vcvtpd2dqx\t{$src, $dst|$dst, $src}", []>, VEX;
3273 def VCVTPD2DQYrr : S3DI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR256:$src),
3274 "vcvtpd2dqy\t{$src, $dst|$dst, $src}", []>, VEX;
3275 def VCVTPD2DQYrm : S3DI<0xE6, MRMSrcMem, (outs VR128:$dst), (ins f256mem:$src),
3276 "vcvtpd2dqy\t{$src, $dst|$dst, $src}", []>, VEX, VEX_L;
3279 def CVTPD2DQrm : S3DI<0xE6, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
3280 "cvtpd2dq\t{$src, $dst|$dst, $src}", []>;
3281 def CVTPD2DQrr : S3DI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
3282 "cvtpd2dq\t{$src, $dst|$dst, $src}", []>;
3284 // Convert Packed DW Integers to Packed Double FP
3285 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
3286 def VCVTDQ2PDrm : S3SI<0xE6, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
3287 "vcvtdq2pd\t{$src, $dst|$dst, $src}", []>, VEX;
3288 def VCVTDQ2PDrr : S3SI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
3289 "vcvtdq2pd\t{$src, $dst|$dst, $src}", []>, VEX;
3290 def VCVTDQ2PDYrm : S3SI<0xE6, MRMSrcMem, (outs VR256:$dst), (ins f128mem:$src),
3291 "vcvtdq2pd\t{$src, $dst|$dst, $src}", []>, VEX;
3292 def VCVTDQ2PDYrr : S3SI<0xE6, MRMSrcReg, (outs VR256:$dst), (ins VR128:$src),
3293 "vcvtdq2pd\t{$src, $dst|$dst, $src}", []>, VEX;
3296 def CVTDQ2PDrm : S3SI<0xE6, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
3297 "cvtdq2pd\t{$src, $dst|$dst, $src}", []>;
3298 def CVTDQ2PDrr : S3SI<0xE6, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
3299 "cvtdq2pd\t{$src, $dst|$dst, $src}", []>;
3301 // AVX 256-bit register conversion intrinsics
3302 def : Pat<(int_x86_avx_cvtdq2_pd_256 VR128:$src),
3303 (VCVTDQ2PDYrr VR128:$src)>;
3304 def : Pat<(int_x86_avx_cvtdq2_pd_256 (memopv4i32 addr:$src)),
3305 (VCVTDQ2PDYrm addr:$src)>;
3307 def : Pat<(int_x86_avx_cvt_pd2dq_256 VR256:$src),
3308 (VCVTPD2DQYrr VR256:$src)>;
3309 def : Pat<(int_x86_avx_cvt_pd2dq_256 (memopv4f64 addr:$src)),
3310 (VCVTPD2DQYrm addr:$src)>;
3312 //===---------------------------------------------------------------------===//
3313 // SSE3 - Move Instructions
3314 //===---------------------------------------------------------------------===//
3316 // Replicate Single FP
3317 multiclass sse3_replicate_sfp<bits<8> op, PatFrag rep_frag, string OpcodeStr> {
3318 def rr : S3SI<op, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
3319 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
3320 [(set VR128:$dst, (v4f32 (rep_frag
3321 VR128:$src, (undef))))]>;
3322 def rm : S3SI<op, MRMSrcMem, (outs VR128:$dst), (ins f128mem:$src),
3323 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
3324 [(set VR128:$dst, (rep_frag
3325 (memopv4f32 addr:$src), (undef)))]>;
3328 multiclass sse3_replicate_sfp_y<bits<8> op, PatFrag rep_frag,
3330 def rr : S3SI<op, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src),
3331 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"), []>;
3332 def rm : S3SI<op, MRMSrcMem, (outs VR256:$dst), (ins f256mem:$src),
3333 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"), []>;
3336 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
3337 // FIXME: Merge above classes when we have patterns for the ymm version
3338 defm VMOVSHDUP : sse3_replicate_sfp<0x16, movshdup, "vmovshdup">, VEX;
3339 defm VMOVSLDUP : sse3_replicate_sfp<0x12, movsldup, "vmovsldup">, VEX;
3340 defm VMOVSHDUPY : sse3_replicate_sfp_y<0x16, movshdup, "vmovshdup">, VEX;
3341 defm VMOVSLDUPY : sse3_replicate_sfp_y<0x12, movsldup, "vmovsldup">, VEX;
3343 defm MOVSHDUP : sse3_replicate_sfp<0x16, movshdup, "movshdup">;
3344 defm MOVSLDUP : sse3_replicate_sfp<0x12, movsldup, "movsldup">;
3346 // Replicate Double FP
3347 multiclass sse3_replicate_dfp<string OpcodeStr> {
3348 def rr : S3DI<0x12, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
3349 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
3350 [(set VR128:$dst,(v2f64 (movddup VR128:$src, (undef))))]>;
3351 def rm : S3DI<0x12, MRMSrcMem, (outs VR128:$dst), (ins f64mem:$src),
3352 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
3354 (v2f64 (movddup (scalar_to_vector (loadf64 addr:$src)),
3358 multiclass sse3_replicate_dfp_y<string OpcodeStr> {
3359 def rr : S3DI<0x12, MRMSrcReg, (outs VR256:$dst), (ins VR256:$src),
3360 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
3362 def rm : S3DI<0x12, MRMSrcMem, (outs VR256:$dst), (ins f256mem:$src),
3363 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
3367 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
3368 // FIXME: Merge above classes when we have patterns for the ymm version
3369 defm VMOVDDUP : sse3_replicate_dfp<"vmovddup">, VEX;
3370 defm VMOVDDUPY : sse3_replicate_dfp_y<"vmovddup">, VEX;
3372 defm MOVDDUP : sse3_replicate_dfp<"movddup">;
3374 // Move Unaligned Integer
3375 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
3376 def VLDDQUrm : S3DI<0xF0, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
3377 "vlddqu\t{$src, $dst|$dst, $src}",
3378 [(set VR128:$dst, (int_x86_sse3_ldu_dq addr:$src))]>, VEX;
3379 def VLDDQUYrm : S3DI<0xF0, MRMSrcMem, (outs VR256:$dst), (ins i256mem:$src),
3380 "vlddqu\t{$src, $dst|$dst, $src}",
3381 [(set VR256:$dst, (int_x86_avx_ldu_dq_256 addr:$src))]>, VEX;
3383 def LDDQUrm : S3DI<0xF0, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
3384 "lddqu\t{$src, $dst|$dst, $src}",
3385 [(set VR128:$dst, (int_x86_sse3_ldu_dq addr:$src))]>;
3387 def : Pat<(movddup (bc_v2f64 (v2i64 (scalar_to_vector (loadi64 addr:$src)))),
3389 (MOVDDUPrm addr:$src)>, Requires<[HasSSE3]>;
3391 // Several Move patterns
3392 let AddedComplexity = 5 in {
3393 def : Pat<(movddup (memopv2f64 addr:$src), (undef)),
3394 (MOVDDUPrm addr:$src)>, Requires<[HasSSE3]>;
3395 def : Pat<(movddup (bc_v4f32 (memopv2f64 addr:$src)), (undef)),
3396 (MOVDDUPrm addr:$src)>, Requires<[HasSSE3]>;
3397 def : Pat<(movddup (memopv2i64 addr:$src), (undef)),
3398 (MOVDDUPrm addr:$src)>, Requires<[HasSSE3]>;
3399 def : Pat<(movddup (bc_v4i32 (memopv2i64 addr:$src)), (undef)),
3400 (MOVDDUPrm addr:$src)>, Requires<[HasSSE3]>;
3403 // vector_shuffle v1, <undef> <1, 1, 3, 3>
3404 let AddedComplexity = 15 in
3405 def : Pat<(v4i32 (movshdup VR128:$src, (undef))),
3406 (MOVSHDUPrr VR128:$src)>, Requires<[HasSSE3]>;
3407 let AddedComplexity = 20 in
3408 def : Pat<(v4i32 (movshdup (bc_v4i32 (memopv2i64 addr:$src)), (undef))),
3409 (MOVSHDUPrm addr:$src)>, Requires<[HasSSE3]>;
3411 // vector_shuffle v1, <undef> <0, 0, 2, 2>
3412 let AddedComplexity = 15 in
3413 def : Pat<(v4i32 (movsldup VR128:$src, (undef))),
3414 (MOVSLDUPrr VR128:$src)>, Requires<[HasSSE3]>;
3415 let AddedComplexity = 20 in
3416 def : Pat<(v4i32 (movsldup (bc_v4i32 (memopv2i64 addr:$src)), (undef))),
3417 (MOVSLDUPrm addr:$src)>, Requires<[HasSSE3]>;
3419 //===---------------------------------------------------------------------===//
3420 // SSE3 - Arithmetic
3421 //===---------------------------------------------------------------------===//
3423 multiclass sse3_addsub<Intrinsic Int, string OpcodeStr, RegisterClass RC,
3424 X86MemOperand x86memop, bit Is2Addr = 1> {
3425 def rr : I<0xD0, MRMSrcReg,
3426 (outs RC:$dst), (ins RC:$src1, RC:$src2),
3428 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
3429 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
3430 [(set RC:$dst, (Int RC:$src1, RC:$src2))]>;
3431 def rm : I<0xD0, MRMSrcMem,
3432 (outs RC:$dst), (ins RC:$src1, x86memop:$src2),
3434 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
3435 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
3436 [(set RC:$dst, (Int RC:$src1, (memop addr:$src2)))]>;
3439 let isAsmParserOnly = 1, Predicates = [HasAVX],
3440 ExeDomain = SSEPackedDouble in {
3441 defm VADDSUBPS : sse3_addsub<int_x86_sse3_addsub_ps, "vaddsubps", VR128,
3442 f128mem, 0>, XD, VEX_4V;
3443 defm VADDSUBPD : sse3_addsub<int_x86_sse3_addsub_pd, "vaddsubpd", VR128,
3444 f128mem, 0>, OpSize, VEX_4V;
3445 defm VADDSUBPSY : sse3_addsub<int_x86_avx_addsub_ps_256, "vaddsubps", VR256,
3446 f256mem, 0>, XD, VEX_4V;
3447 defm VADDSUBPDY : sse3_addsub<int_x86_avx_addsub_pd_256, "vaddsubpd", VR256,
3448 f256mem, 0>, OpSize, VEX_4V;
3450 let Constraints = "$src1 = $dst", Predicates = [HasSSE3],
3451 ExeDomain = SSEPackedDouble in {
3452 defm ADDSUBPS : sse3_addsub<int_x86_sse3_addsub_ps, "addsubps", VR128,
3454 defm ADDSUBPD : sse3_addsub<int_x86_sse3_addsub_pd, "addsubpd", VR128,
3455 f128mem>, TB, OpSize;
3458 //===---------------------------------------------------------------------===//
3459 // SSE3 Instructions
3460 //===---------------------------------------------------------------------===//
3463 multiclass S3D_Int<bits<8> o, string OpcodeStr, ValueType vt, RegisterClass RC,
3464 X86MemOperand x86memop, Intrinsic IntId, bit Is2Addr = 1> {
3465 def rr : S3DI<o, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, RC:$src2),
3467 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
3468 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
3469 [(set RC:$dst, (vt (IntId RC:$src1, RC:$src2)))]>;
3471 def rm : S3DI<o, MRMSrcMem, (outs RC:$dst), (ins RC:$src1, x86memop:$src2),
3473 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
3474 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
3475 [(set RC:$dst, (vt (IntId RC:$src1, (memop addr:$src2))))]>;
3477 multiclass S3_Int<bits<8> o, string OpcodeStr, ValueType vt, RegisterClass RC,
3478 X86MemOperand x86memop, Intrinsic IntId, bit Is2Addr = 1> {
3479 def rr : S3I<o, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, RC:$src2),
3481 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
3482 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
3483 [(set RC:$dst, (vt (IntId RC:$src1, RC:$src2)))]>;
3485 def rm : S3I<o, MRMSrcMem, (outs RC:$dst), (ins RC:$src1, x86memop:$src2),
3487 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
3488 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
3489 [(set RC:$dst, (vt (IntId RC:$src1, (memop addr:$src2))))]>;
3492 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
3493 defm VHADDPS : S3D_Int<0x7C, "vhaddps", v4f32, VR128, f128mem,
3494 int_x86_sse3_hadd_ps, 0>, VEX_4V;
3495 defm VHADDPD : S3_Int <0x7C, "vhaddpd", v2f64, VR128, f128mem,
3496 int_x86_sse3_hadd_pd, 0>, VEX_4V;
3497 defm VHSUBPS : S3D_Int<0x7D, "vhsubps", v4f32, VR128, f128mem,
3498 int_x86_sse3_hsub_ps, 0>, VEX_4V;
3499 defm VHSUBPD : S3_Int <0x7D, "vhsubpd", v2f64, VR128, f128mem,
3500 int_x86_sse3_hsub_pd, 0>, VEX_4V;
3501 defm VHADDPSY : S3D_Int<0x7C, "vhaddps", v8f32, VR256, f256mem,
3502 int_x86_avx_hadd_ps_256, 0>, VEX_4V;
3503 defm VHADDPDY : S3_Int <0x7C, "vhaddpd", v4f64, VR256, f256mem,
3504 int_x86_avx_hadd_pd_256, 0>, VEX_4V;
3505 defm VHSUBPSY : S3D_Int<0x7D, "vhsubps", v8f32, VR256, f256mem,
3506 int_x86_avx_hsub_ps_256, 0>, VEX_4V;
3507 defm VHSUBPDY : S3_Int <0x7D, "vhsubpd", v4f64, VR256, f256mem,
3508 int_x86_avx_hsub_pd_256, 0>, VEX_4V;
3511 let Constraints = "$src1 = $dst" in {
3512 defm HADDPS : S3D_Int<0x7C, "haddps", v4f32, VR128, f128mem,
3513 int_x86_sse3_hadd_ps>;
3514 defm HADDPD : S3_Int<0x7C, "haddpd", v2f64, VR128, f128mem,
3515 int_x86_sse3_hadd_pd>;
3516 defm HSUBPS : S3D_Int<0x7D, "hsubps", v4f32, VR128, f128mem,
3517 int_x86_sse3_hsub_ps>;
3518 defm HSUBPD : S3_Int<0x7D, "hsubpd", v2f64, VR128, f128mem,
3519 int_x86_sse3_hsub_pd>;
3522 //===---------------------------------------------------------------------===//
3523 // SSSE3 - Packed Absolute Instructions
3524 //===---------------------------------------------------------------------===//
3526 /// SS3I_unop_rm_int - Simple SSSE3 unary op whose type can be v*{i8,i16,i32}.
3527 multiclass SS3I_unop_rm_int<bits<8> opc, string OpcodeStr,
3528 PatFrag mem_frag64, PatFrag mem_frag128,
3529 Intrinsic IntId64, Intrinsic IntId128> {
3530 def rr64 : SS38I<opc, MRMSrcReg, (outs VR64:$dst), (ins VR64:$src),
3531 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
3532 [(set VR64:$dst, (IntId64 VR64:$src))]>;
3534 def rm64 : SS38I<opc, MRMSrcMem, (outs VR64:$dst), (ins i64mem:$src),
3535 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
3537 (IntId64 (bitconvert (mem_frag64 addr:$src))))]>;
3539 def rr128 : SS38I<opc, MRMSrcReg, (outs VR128:$dst),
3541 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
3542 [(set VR128:$dst, (IntId128 VR128:$src))]>,
3545 def rm128 : SS38I<opc, MRMSrcMem, (outs VR128:$dst),
3547 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
3550 (bitconvert (mem_frag128 addr:$src))))]>, OpSize;
3553 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
3554 defm VPABSB : SS3I_unop_rm_int<0x1C, "vpabsb", memopv8i8, memopv16i8,
3555 int_x86_ssse3_pabs_b,
3556 int_x86_ssse3_pabs_b_128>, VEX;
3557 defm VPABSW : SS3I_unop_rm_int<0x1D, "vpabsw", memopv4i16, memopv8i16,
3558 int_x86_ssse3_pabs_w,
3559 int_x86_ssse3_pabs_w_128>, VEX;
3560 defm VPABSD : SS3I_unop_rm_int<0x1E, "vpabsd", memopv2i32, memopv4i32,
3561 int_x86_ssse3_pabs_d,
3562 int_x86_ssse3_pabs_d_128>, VEX;
3565 defm PABSB : SS3I_unop_rm_int<0x1C, "pabsb", memopv8i8, memopv16i8,
3566 int_x86_ssse3_pabs_b,
3567 int_x86_ssse3_pabs_b_128>;
3568 defm PABSW : SS3I_unop_rm_int<0x1D, "pabsw", memopv4i16, memopv8i16,
3569 int_x86_ssse3_pabs_w,
3570 int_x86_ssse3_pabs_w_128>;
3571 defm PABSD : SS3I_unop_rm_int<0x1E, "pabsd", memopv2i32, memopv4i32,
3572 int_x86_ssse3_pabs_d,
3573 int_x86_ssse3_pabs_d_128>;
3575 //===---------------------------------------------------------------------===//
3576 // SSSE3 - Packed Binary Operator Instructions
3577 //===---------------------------------------------------------------------===//
3579 /// SS3I_binop_rm_int - Simple SSSE3 bin op whose type can be v*{i8,i16,i32}.
3580 multiclass SS3I_binop_rm_int<bits<8> opc, string OpcodeStr,
3581 PatFrag mem_frag64, PatFrag mem_frag128,
3582 Intrinsic IntId64, Intrinsic IntId128,
3584 let isCommutable = 1 in
3585 def rr64 : SS38I<opc, MRMSrcReg, (outs VR64:$dst),
3586 (ins VR64:$src1, VR64:$src2),
3588 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
3589 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
3590 [(set VR64:$dst, (IntId64 VR64:$src1, VR64:$src2))]>;
3591 def rm64 : SS38I<opc, MRMSrcMem, (outs VR64:$dst),
3592 (ins VR64:$src1, i64mem:$src2),
3594 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
3595 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
3597 (IntId64 VR64:$src1,
3598 (bitconvert (memopv8i8 addr:$src2))))]>;
3600 let isCommutable = 1 in
3601 def rr128 : SS38I<opc, MRMSrcReg, (outs VR128:$dst),
3602 (ins VR128:$src1, VR128:$src2),
3604 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
3605 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
3606 [(set VR128:$dst, (IntId128 VR128:$src1, VR128:$src2))]>,
3608 def rm128 : SS38I<opc, MRMSrcMem, (outs VR128:$dst),
3609 (ins VR128:$src1, i128mem:$src2),
3611 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
3612 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
3614 (IntId128 VR128:$src1,
3615 (bitconvert (memopv16i8 addr:$src2))))]>, OpSize;
3618 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
3619 let isCommutable = 0 in {
3620 defm VPHADDW : SS3I_binop_rm_int<0x01, "vphaddw", memopv4i16, memopv8i16,
3621 int_x86_ssse3_phadd_w,
3622 int_x86_ssse3_phadd_w_128, 0>, VEX_4V;
3623 defm VPHADDD : SS3I_binop_rm_int<0x02, "vphaddd", memopv2i32, memopv4i32,
3624 int_x86_ssse3_phadd_d,
3625 int_x86_ssse3_phadd_d_128, 0>, VEX_4V;
3626 defm VPHADDSW : SS3I_binop_rm_int<0x03, "vphaddsw", memopv4i16, memopv8i16,
3627 int_x86_ssse3_phadd_sw,
3628 int_x86_ssse3_phadd_sw_128, 0>, VEX_4V;
3629 defm VPHSUBW : SS3I_binop_rm_int<0x05, "vphsubw", memopv4i16, memopv8i16,
3630 int_x86_ssse3_phsub_w,
3631 int_x86_ssse3_phsub_w_128, 0>, VEX_4V;
3632 defm VPHSUBD : SS3I_binop_rm_int<0x06, "vphsubd", memopv2i32, memopv4i32,
3633 int_x86_ssse3_phsub_d,
3634 int_x86_ssse3_phsub_d_128, 0>, VEX_4V;
3635 defm VPHSUBSW : SS3I_binop_rm_int<0x07, "vphsubsw", memopv4i16, memopv8i16,
3636 int_x86_ssse3_phsub_sw,
3637 int_x86_ssse3_phsub_sw_128, 0>, VEX_4V;
3638 defm VPMADDUBSW : SS3I_binop_rm_int<0x04, "vpmaddubsw", memopv8i8, memopv16i8,
3639 int_x86_ssse3_pmadd_ub_sw,
3640 int_x86_ssse3_pmadd_ub_sw_128, 0>, VEX_4V;
3641 defm VPSHUFB : SS3I_binop_rm_int<0x00, "vpshufb", memopv8i8, memopv16i8,
3642 int_x86_ssse3_pshuf_b,
3643 int_x86_ssse3_pshuf_b_128, 0>, VEX_4V;
3644 defm VPSIGNB : SS3I_binop_rm_int<0x08, "vpsignb", memopv8i8, memopv16i8,
3645 int_x86_ssse3_psign_b,
3646 int_x86_ssse3_psign_b_128, 0>, VEX_4V;
3647 defm VPSIGNW : SS3I_binop_rm_int<0x09, "vpsignw", memopv4i16, memopv8i16,
3648 int_x86_ssse3_psign_w,
3649 int_x86_ssse3_psign_w_128, 0>, VEX_4V;
3650 defm VPSIGND : SS3I_binop_rm_int<0x0A, "vpsignd", memopv2i32, memopv4i32,
3651 int_x86_ssse3_psign_d,
3652 int_x86_ssse3_psign_d_128, 0>, VEX_4V;
3654 defm VPMULHRSW : SS3I_binop_rm_int<0x0B, "vpmulhrsw", memopv4i16, memopv8i16,
3655 int_x86_ssse3_pmul_hr_sw,
3656 int_x86_ssse3_pmul_hr_sw_128, 0>, VEX_4V;
3659 // None of these have i8 immediate fields.
3660 let ImmT = NoImm, Constraints = "$src1 = $dst" in {
3661 let isCommutable = 0 in {
3662 defm PHADDW : SS3I_binop_rm_int<0x01, "phaddw", memopv4i16, memopv8i16,
3663 int_x86_ssse3_phadd_w,
3664 int_x86_ssse3_phadd_w_128>;
3665 defm PHADDD : SS3I_binop_rm_int<0x02, "phaddd", memopv2i32, memopv4i32,
3666 int_x86_ssse3_phadd_d,
3667 int_x86_ssse3_phadd_d_128>;
3668 defm PHADDSW : SS3I_binop_rm_int<0x03, "phaddsw", memopv4i16, memopv8i16,
3669 int_x86_ssse3_phadd_sw,
3670 int_x86_ssse3_phadd_sw_128>;
3671 defm PHSUBW : SS3I_binop_rm_int<0x05, "phsubw", memopv4i16, memopv8i16,
3672 int_x86_ssse3_phsub_w,
3673 int_x86_ssse3_phsub_w_128>;
3674 defm PHSUBD : SS3I_binop_rm_int<0x06, "phsubd", memopv2i32, memopv4i32,
3675 int_x86_ssse3_phsub_d,
3676 int_x86_ssse3_phsub_d_128>;
3677 defm PHSUBSW : SS3I_binop_rm_int<0x07, "phsubsw", memopv4i16, memopv8i16,
3678 int_x86_ssse3_phsub_sw,
3679 int_x86_ssse3_phsub_sw_128>;
3680 defm PMADDUBSW : SS3I_binop_rm_int<0x04, "pmaddubsw", memopv8i8, memopv16i8,
3681 int_x86_ssse3_pmadd_ub_sw,
3682 int_x86_ssse3_pmadd_ub_sw_128>;
3683 defm PSHUFB : SS3I_binop_rm_int<0x00, "pshufb", memopv8i8, memopv16i8,
3684 int_x86_ssse3_pshuf_b,
3685 int_x86_ssse3_pshuf_b_128>;
3686 defm PSIGNB : SS3I_binop_rm_int<0x08, "psignb", memopv8i8, memopv16i8,
3687 int_x86_ssse3_psign_b,
3688 int_x86_ssse3_psign_b_128>;
3689 defm PSIGNW : SS3I_binop_rm_int<0x09, "psignw", memopv4i16, memopv8i16,
3690 int_x86_ssse3_psign_w,
3691 int_x86_ssse3_psign_w_128>;
3692 defm PSIGND : SS3I_binop_rm_int<0x0A, "psignd", memopv2i32, memopv4i32,
3693 int_x86_ssse3_psign_d,
3694 int_x86_ssse3_psign_d_128>;
3696 defm PMULHRSW : SS3I_binop_rm_int<0x0B, "pmulhrsw", memopv4i16, memopv8i16,
3697 int_x86_ssse3_pmul_hr_sw,
3698 int_x86_ssse3_pmul_hr_sw_128>;
3701 def : Pat<(X86pshufb VR128:$src, VR128:$mask),
3702 (PSHUFBrr128 VR128:$src, VR128:$mask)>, Requires<[HasSSSE3]>;
3703 def : Pat<(X86pshufb VR128:$src, (bc_v16i8 (memopv2i64 addr:$mask))),
3704 (PSHUFBrm128 VR128:$src, addr:$mask)>, Requires<[HasSSSE3]>;
3706 //===---------------------------------------------------------------------===//
3707 // SSSE3 - Packed Align Instruction Patterns
3708 //===---------------------------------------------------------------------===//
3710 multiclass sse3_palign<string asm, bit Is2Addr = 1> {
3711 def R64rr : SS3AI<0x0F, MRMSrcReg, (outs VR64:$dst),
3712 (ins VR64:$src1, VR64:$src2, i8imm:$src3),
3714 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
3716 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
3718 def R64rm : SS3AI<0x0F, MRMSrcMem, (outs VR64:$dst),
3719 (ins VR64:$src1, i64mem:$src2, i8imm:$src3),
3721 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
3723 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
3726 def R128rr : SS3AI<0x0F, MRMSrcReg, (outs VR128:$dst),
3727 (ins VR128:$src1, VR128:$src2, i8imm:$src3),
3729 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
3731 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
3733 def R128rm : SS3AI<0x0F, MRMSrcMem, (outs VR128:$dst),
3734 (ins VR128:$src1, i128mem:$src2, i8imm:$src3),
3736 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
3738 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
3742 let isAsmParserOnly = 1, Predicates = [HasAVX] in
3743 defm VPALIGN : sse3_palign<"vpalignr", 0>, VEX_4V;
3744 let Constraints = "$src1 = $dst" in
3745 defm PALIGN : sse3_palign<"palignr">;
3747 let AddedComplexity = 5 in {
3749 def : Pat<(v1i64 (palign:$src3 VR64:$src1, VR64:$src2)),
3750 (PALIGNR64rr VR64:$src2, VR64:$src1,
3751 (SHUFFLE_get_palign_imm VR64:$src3))>,
3752 Requires<[HasSSSE3]>;
3753 def : Pat<(v2i32 (palign:$src3 VR64:$src1, VR64:$src2)),
3754 (PALIGNR64rr VR64:$src2, VR64:$src1,
3755 (SHUFFLE_get_palign_imm VR64:$src3))>,
3756 Requires<[HasSSSE3]>;
3757 def : Pat<(v4i16 (palign:$src3 VR64:$src1, VR64:$src2)),
3758 (PALIGNR64rr VR64:$src2, VR64:$src1,
3759 (SHUFFLE_get_palign_imm VR64:$src3))>,
3760 Requires<[HasSSSE3]>;
3761 def : Pat<(v8i8 (palign:$src3 VR64:$src1, VR64:$src2)),
3762 (PALIGNR64rr VR64:$src2, VR64:$src1,
3763 (SHUFFLE_get_palign_imm VR64:$src3))>,
3764 Requires<[HasSSSE3]>;
3766 def : Pat<(v4i32 (palign:$src3 VR128:$src1, VR128:$src2)),
3767 (PALIGNR128rr VR128:$src2, VR128:$src1,
3768 (SHUFFLE_get_palign_imm VR128:$src3))>,
3769 Requires<[HasSSSE3]>;
3770 def : Pat<(v4f32 (palign:$src3 VR128:$src1, VR128:$src2)),
3771 (PALIGNR128rr VR128:$src2, VR128:$src1,
3772 (SHUFFLE_get_palign_imm VR128:$src3))>,
3773 Requires<[HasSSSE3]>;
3774 def : Pat<(v8i16 (palign:$src3 VR128:$src1, VR128:$src2)),
3775 (PALIGNR128rr VR128:$src2, VR128:$src1,
3776 (SHUFFLE_get_palign_imm VR128:$src3))>,
3777 Requires<[HasSSSE3]>;
3778 def : Pat<(v16i8 (palign:$src3 VR128:$src1, VR128:$src2)),
3779 (PALIGNR128rr VR128:$src2, VR128:$src1,
3780 (SHUFFLE_get_palign_imm VR128:$src3))>,
3781 Requires<[HasSSSE3]>;
3784 //===---------------------------------------------------------------------===//
3785 // SSSE3 Misc Instructions
3786 //===---------------------------------------------------------------------===//
3788 // Thread synchronization
3789 def MONITOR : I<0x01, MRM_C8, (outs), (ins), "monitor",
3790 [(int_x86_sse3_monitor EAX, ECX, EDX)]>,TB, Requires<[HasSSE3]>;
3791 def MWAIT : I<0x01, MRM_C9, (outs), (ins), "mwait",
3792 [(int_x86_sse3_mwait ECX, EAX)]>, TB, Requires<[HasSSE3]>;
3794 //===---------------------------------------------------------------------===//
3795 // Non-Instruction Patterns
3796 //===---------------------------------------------------------------------===//
3798 // extload f32 -> f64. This matches load+fextend because we have a hack in
3799 // the isel (PreprocessForFPConvert) that can introduce loads after dag
3801 // Since these loads aren't folded into the fextend, we have to match it
3803 let Predicates = [HasSSE2] in
3804 def : Pat<(fextend (loadf32 addr:$src)),
3805 (CVTSS2SDrm addr:$src)>;
3808 let Predicates = [HasSSE2] in {
3809 def : Pat<(v2i64 (bitconvert (v4i32 VR128:$src))), (v2i64 VR128:$src)>;
3810 def : Pat<(v2i64 (bitconvert (v8i16 VR128:$src))), (v2i64 VR128:$src)>;
3811 def : Pat<(v2i64 (bitconvert (v16i8 VR128:$src))), (v2i64 VR128:$src)>;
3812 def : Pat<(v2i64 (bitconvert (v2f64 VR128:$src))), (v2i64 VR128:$src)>;
3813 def : Pat<(v2i64 (bitconvert (v4f32 VR128:$src))), (v2i64 VR128:$src)>;
3814 def : Pat<(v4i32 (bitconvert (v2i64 VR128:$src))), (v4i32 VR128:$src)>;
3815 def : Pat<(v4i32 (bitconvert (v8i16 VR128:$src))), (v4i32 VR128:$src)>;
3816 def : Pat<(v4i32 (bitconvert (v16i8 VR128:$src))), (v4i32 VR128:$src)>;
3817 def : Pat<(v4i32 (bitconvert (v2f64 VR128:$src))), (v4i32 VR128:$src)>;
3818 def : Pat<(v4i32 (bitconvert (v4f32 VR128:$src))), (v4i32 VR128:$src)>;
3819 def : Pat<(v8i16 (bitconvert (v2i64 VR128:$src))), (v8i16 VR128:$src)>;
3820 def : Pat<(v8i16 (bitconvert (v4i32 VR128:$src))), (v8i16 VR128:$src)>;
3821 def : Pat<(v8i16 (bitconvert (v16i8 VR128:$src))), (v8i16 VR128:$src)>;
3822 def : Pat<(v8i16 (bitconvert (v2f64 VR128:$src))), (v8i16 VR128:$src)>;
3823 def : Pat<(v8i16 (bitconvert (v4f32 VR128:$src))), (v8i16 VR128:$src)>;
3824 def : Pat<(v16i8 (bitconvert (v2i64 VR128:$src))), (v16i8 VR128:$src)>;
3825 def : Pat<(v16i8 (bitconvert (v4i32 VR128:$src))), (v16i8 VR128:$src)>;
3826 def : Pat<(v16i8 (bitconvert (v8i16 VR128:$src))), (v16i8 VR128:$src)>;
3827 def : Pat<(v16i8 (bitconvert (v2f64 VR128:$src))), (v16i8 VR128:$src)>;
3828 def : Pat<(v16i8 (bitconvert (v4f32 VR128:$src))), (v16i8 VR128:$src)>;
3829 def : Pat<(v4f32 (bitconvert (v2i64 VR128:$src))), (v4f32 VR128:$src)>;
3830 def : Pat<(v4f32 (bitconvert (v4i32 VR128:$src))), (v4f32 VR128:$src)>;
3831 def : Pat<(v4f32 (bitconvert (v8i16 VR128:$src))), (v4f32 VR128:$src)>;
3832 def : Pat<(v4f32 (bitconvert (v16i8 VR128:$src))), (v4f32 VR128:$src)>;
3833 def : Pat<(v4f32 (bitconvert (v2f64 VR128:$src))), (v4f32 VR128:$src)>;
3834 def : Pat<(v2f64 (bitconvert (v2i64 VR128:$src))), (v2f64 VR128:$src)>;
3835 def : Pat<(v2f64 (bitconvert (v4i32 VR128:$src))), (v2f64 VR128:$src)>;
3836 def : Pat<(v2f64 (bitconvert (v8i16 VR128:$src))), (v2f64 VR128:$src)>;
3837 def : Pat<(v2f64 (bitconvert (v16i8 VR128:$src))), (v2f64 VR128:$src)>;
3838 def : Pat<(v2f64 (bitconvert (v4f32 VR128:$src))), (v2f64 VR128:$src)>;
3841 // Move scalar to XMM zero-extended
3842 // movd to XMM register zero-extends
3843 let AddedComplexity = 15 in {
3844 // Zeroing a VR128 then do a MOVS{S|D} to the lower bits.
3845 def : Pat<(v2f64 (X86vzmovl (v2f64 (scalar_to_vector FR64:$src)))),
3846 (MOVSDrr (v2f64 (V_SET0PS)), FR64:$src)>;
3847 def : Pat<(v4f32 (X86vzmovl (v4f32 (scalar_to_vector FR32:$src)))),
3848 (MOVSSrr (v4f32 (V_SET0PS)), FR32:$src)>;
3849 def : Pat<(v4f32 (X86vzmovl (v4f32 VR128:$src))),
3850 (MOVSSrr (v4f32 (V_SET0PS)),
3851 (f32 (EXTRACT_SUBREG (v4f32 VR128:$src), sub_ss)))>;
3852 def : Pat<(v4i32 (X86vzmovl (v4i32 VR128:$src))),
3853 (MOVSSrr (v4i32 (V_SET0PI)),
3854 (EXTRACT_SUBREG (v4i32 VR128:$src), sub_ss))>;
3857 // Splat v2f64 / v2i64
3858 let AddedComplexity = 10 in {
3859 def : Pat<(splat_lo (v2f64 VR128:$src), (undef)),
3860 (UNPCKLPDrr VR128:$src, VR128:$src)>, Requires<[HasSSE2]>;
3861 def : Pat<(unpckh (v2f64 VR128:$src), (undef)),
3862 (UNPCKHPDrr VR128:$src, VR128:$src)>, Requires<[HasSSE2]>;
3863 def : Pat<(splat_lo (v2i64 VR128:$src), (undef)),
3864 (PUNPCKLQDQrr VR128:$src, VR128:$src)>, Requires<[HasSSE2]>;
3865 def : Pat<(unpckh (v2i64 VR128:$src), (undef)),
3866 (PUNPCKHQDQrr VR128:$src, VR128:$src)>, Requires<[HasSSE2]>;
3869 // Special unary SHUFPSrri case.
3870 def : Pat<(v4f32 (pshufd:$src3 VR128:$src1, (undef))),
3871 (SHUFPSrri VR128:$src1, VR128:$src1,
3872 (SHUFFLE_get_shuf_imm VR128:$src3))>;
3873 let AddedComplexity = 5 in
3874 def : Pat<(v4f32 (pshufd:$src2 VR128:$src1, (undef))),
3875 (PSHUFDri VR128:$src1, (SHUFFLE_get_shuf_imm VR128:$src2))>,
3876 Requires<[HasSSE2]>;
3877 // Special unary SHUFPDrri case.
3878 def : Pat<(v2i64 (pshufd:$src3 VR128:$src1, (undef))),
3879 (SHUFPDrri VR128:$src1, VR128:$src1,
3880 (SHUFFLE_get_shuf_imm VR128:$src3))>,
3881 Requires<[HasSSE2]>;
3882 // Special unary SHUFPDrri case.
3883 def : Pat<(v2f64 (pshufd:$src3 VR128:$src1, (undef))),
3884 (SHUFPDrri VR128:$src1, VR128:$src1,
3885 (SHUFFLE_get_shuf_imm VR128:$src3))>,
3886 Requires<[HasSSE2]>;
3887 // Unary v4f32 shuffle with PSHUF* in order to fold a load.
3888 def : Pat<(pshufd:$src2 (bc_v4i32 (memopv4f32 addr:$src1)), (undef)),
3889 (PSHUFDmi addr:$src1, (SHUFFLE_get_shuf_imm VR128:$src2))>,
3890 Requires<[HasSSE2]>;
3892 // Special binary v4i32 shuffle cases with SHUFPS.
3893 def : Pat<(v4i32 (shufp:$src3 VR128:$src1, (v4i32 VR128:$src2))),
3894 (SHUFPSrri VR128:$src1, VR128:$src2,
3895 (SHUFFLE_get_shuf_imm VR128:$src3))>,
3896 Requires<[HasSSE2]>;
3897 def : Pat<(v4i32 (shufp:$src3 VR128:$src1, (bc_v4i32 (memopv2i64 addr:$src2)))),
3898 (SHUFPSrmi VR128:$src1, addr:$src2,
3899 (SHUFFLE_get_shuf_imm VR128:$src3))>,
3900 Requires<[HasSSE2]>;
3901 // Special binary v2i64 shuffle cases using SHUFPDrri.
3902 def : Pat<(v2i64 (shufp:$src3 VR128:$src1, VR128:$src2)),
3903 (SHUFPDrri VR128:$src1, VR128:$src2,
3904 (SHUFFLE_get_shuf_imm VR128:$src3))>,
3905 Requires<[HasSSE2]>;
3907 // vector_shuffle v1, <undef>, <0, 0, 1, 1, ...>
3908 let AddedComplexity = 15 in {
3909 def : Pat<(v4i32 (unpckl_undef:$src2 VR128:$src, (undef))),
3910 (PSHUFDri VR128:$src, (SHUFFLE_get_shuf_imm VR128:$src2))>,
3911 Requires<[OptForSpeed, HasSSE2]>;
3912 def : Pat<(v4f32 (unpckl_undef:$src2 VR128:$src, (undef))),
3913 (PSHUFDri VR128:$src, (SHUFFLE_get_shuf_imm VR128:$src2))>,
3914 Requires<[OptForSpeed, HasSSE2]>;
3916 let AddedComplexity = 10 in {
3917 def : Pat<(v4f32 (unpckl_undef VR128:$src, (undef))),
3918 (UNPCKLPSrr VR128:$src, VR128:$src)>;
3919 def : Pat<(v16i8 (unpckl_undef VR128:$src, (undef))),
3920 (PUNPCKLBWrr VR128:$src, VR128:$src)>;
3921 def : Pat<(v8i16 (unpckl_undef VR128:$src, (undef))),
3922 (PUNPCKLWDrr VR128:$src, VR128:$src)>;
3923 def : Pat<(v4i32 (unpckl_undef VR128:$src, (undef))),
3924 (PUNPCKLDQrr VR128:$src, VR128:$src)>;
3927 // vector_shuffle v1, <undef>, <2, 2, 3, 3, ...>
3928 let AddedComplexity = 15 in {
3929 def : Pat<(v4i32 (unpckh_undef:$src2 VR128:$src, (undef))),
3930 (PSHUFDri VR128:$src, (SHUFFLE_get_shuf_imm VR128:$src2))>,
3931 Requires<[OptForSpeed, HasSSE2]>;
3932 def : Pat<(v4f32 (unpckh_undef:$src2 VR128:$src, (undef))),
3933 (PSHUFDri VR128:$src, (SHUFFLE_get_shuf_imm VR128:$src2))>,
3934 Requires<[OptForSpeed, HasSSE2]>;
3936 let AddedComplexity = 10 in {
3937 def : Pat<(v4f32 (unpckh_undef VR128:$src, (undef))),
3938 (UNPCKHPSrr VR128:$src, VR128:$src)>;
3939 def : Pat<(v16i8 (unpckh_undef VR128:$src, (undef))),
3940 (PUNPCKHBWrr VR128:$src, VR128:$src)>;
3941 def : Pat<(v8i16 (unpckh_undef VR128:$src, (undef))),
3942 (PUNPCKHWDrr VR128:$src, VR128:$src)>;
3943 def : Pat<(v4i32 (unpckh_undef VR128:$src, (undef))),
3944 (PUNPCKHDQrr VR128:$src, VR128:$src)>;
3947 let AddedComplexity = 20 in {
3948 // vector_shuffle v1, v2 <0, 1, 4, 5> using MOVLHPS
3949 def : Pat<(v4i32 (movlhps VR128:$src1, VR128:$src2)),
3950 (MOVLHPSrr VR128:$src1, VR128:$src2)>;
3952 // vector_shuffle v1, v2 <6, 7, 2, 3> using MOVHLPS
3953 def : Pat<(v4i32 (movhlps VR128:$src1, VR128:$src2)),
3954 (MOVHLPSrr VR128:$src1, VR128:$src2)>;
3956 // vector_shuffle v1, undef <2, ?, ?, ?> using MOVHLPS
3957 def : Pat<(v4f32 (movhlps_undef VR128:$src1, (undef))),
3958 (MOVHLPSrr VR128:$src1, VR128:$src1)>;
3959 def : Pat<(v4i32 (movhlps_undef VR128:$src1, (undef))),
3960 (MOVHLPSrr VR128:$src1, VR128:$src1)>;
3963 let AddedComplexity = 20 in {
3964 // vector_shuffle v1, (load v2) <4, 5, 2, 3> using MOVLPS
3965 def : Pat<(v4f32 (movlp VR128:$src1, (load addr:$src2))),
3966 (MOVLPSrm VR128:$src1, addr:$src2)>;
3967 def : Pat<(v2f64 (movlp VR128:$src1, (load addr:$src2))),
3968 (MOVLPDrm VR128:$src1, addr:$src2)>;
3969 def : Pat<(v4i32 (movlp VR128:$src1, (load addr:$src2))),
3970 (MOVLPSrm VR128:$src1, addr:$src2)>;
3971 def : Pat<(v2i64 (movlp VR128:$src1, (load addr:$src2))),
3972 (MOVLPDrm VR128:$src1, addr:$src2)>;
3975 // (store (vector_shuffle (load addr), v2, <4, 5, 2, 3>), addr) using MOVLPS
3976 def : Pat<(store (v4f32 (movlp (load addr:$src1), VR128:$src2)), addr:$src1),
3977 (MOVLPSmr addr:$src1, VR128:$src2)>;
3978 def : Pat<(store (v2f64 (movlp (load addr:$src1), VR128:$src2)), addr:$src1),
3979 (MOVLPDmr addr:$src1, VR128:$src2)>;
3980 def : Pat<(store (v4i32 (movlp (bc_v4i32 (loadv2i64 addr:$src1)), VR128:$src2)),
3982 (MOVLPSmr addr:$src1, VR128:$src2)>;
3983 def : Pat<(store (v2i64 (movlp (load addr:$src1), VR128:$src2)), addr:$src1),
3984 (MOVLPDmr addr:$src1, VR128:$src2)>;
3986 let AddedComplexity = 15 in {
3987 // Setting the lowest element in the vector.
3988 def : Pat<(v4i32 (movl VR128:$src1, VR128:$src2)),
3989 (MOVSSrr (v4i32 VR128:$src1),
3990 (EXTRACT_SUBREG (v4i32 VR128:$src2), sub_ss))>;
3991 def : Pat<(v2i64 (movl VR128:$src1, VR128:$src2)),
3992 (MOVSDrr (v2i64 VR128:$src1),
3993 (EXTRACT_SUBREG (v2i64 VR128:$src2), sub_sd))>;
3995 // vector_shuffle v1, v2 <4, 5, 2, 3> using movsd
3996 def : Pat<(v4f32 (movlp VR128:$src1, VR128:$src2)),
3997 (MOVSDrr VR128:$src1, (EXTRACT_SUBREG VR128:$src2, sub_sd))>,
3998 Requires<[HasSSE2]>;
3999 def : Pat<(v4i32 (movlp VR128:$src1, VR128:$src2)),
4000 (MOVSDrr VR128:$src1, (EXTRACT_SUBREG VR128:$src2, sub_sd))>,
4001 Requires<[HasSSE2]>;
4004 // vector_shuffle v1, v2 <4, 5, 2, 3> using SHUFPSrri (we prefer movsd, but
4005 // fall back to this for SSE1)
4006 def : Pat<(v4f32 (movlp:$src3 VR128:$src1, (v4f32 VR128:$src2))),
4007 (SHUFPSrri VR128:$src2, VR128:$src1,
4008 (SHUFFLE_get_shuf_imm VR128:$src3))>;
4010 // Set lowest element and zero upper elements.
4011 def : Pat<(v2f64 (X86vzmovl (v2f64 VR128:$src))),
4012 (MOVZPQILo2PQIrr VR128:$src)>, Requires<[HasSSE2]>;
4014 // Some special case pandn patterns.
4015 def : Pat<(v2i64 (and (xor VR128:$src1, (bc_v2i64 (v4i32 immAllOnesV))),
4017 (PANDNrr VR128:$src1, VR128:$src2)>, Requires<[HasSSE2]>;
4018 def : Pat<(v2i64 (and (xor VR128:$src1, (bc_v2i64 (v8i16 immAllOnesV))),
4020 (PANDNrr VR128:$src1, VR128:$src2)>, Requires<[HasSSE2]>;
4021 def : Pat<(v2i64 (and (xor VR128:$src1, (bc_v2i64 (v16i8 immAllOnesV))),
4023 (PANDNrr VR128:$src1, VR128:$src2)>, Requires<[HasSSE2]>;
4025 def : Pat<(v2i64 (and (xor VR128:$src1, (bc_v2i64 (v4i32 immAllOnesV))),
4026 (memop addr:$src2))),
4027 (PANDNrm VR128:$src1, addr:$src2)>, Requires<[HasSSE2]>;
4028 def : Pat<(v2i64 (and (xor VR128:$src1, (bc_v2i64 (v8i16 immAllOnesV))),
4029 (memop addr:$src2))),
4030 (PANDNrm VR128:$src1, addr:$src2)>, Requires<[HasSSE2]>;
4031 def : Pat<(v2i64 (and (xor VR128:$src1, (bc_v2i64 (v16i8 immAllOnesV))),
4032 (memop addr:$src2))),
4033 (PANDNrm VR128:$src1, addr:$src2)>, Requires<[HasSSE2]>;
4035 // vector -> vector casts
4036 def : Pat<(v4f32 (sint_to_fp (v4i32 VR128:$src))),
4037 (Int_CVTDQ2PSrr VR128:$src)>, Requires<[HasSSE2]>;
4038 def : Pat<(v4i32 (fp_to_sint (v4f32 VR128:$src))),
4039 (Int_CVTTPS2DQrr VR128:$src)>, Requires<[HasSSE2]>;
4040 def : Pat<(v2f64 (sint_to_fp (v2i32 VR64:$src))),
4041 (Int_CVTPI2PDrr VR64:$src)>, Requires<[HasSSE2]>;
4042 def : Pat<(v2i32 (fp_to_sint (v2f64 VR128:$src))),
4043 (Int_CVTTPD2PIrr VR128:$src)>, Requires<[HasSSE2]>;
4045 // Use movaps / movups for SSE integer load / store (one byte shorter).
4046 def : Pat<(alignedloadv4i32 addr:$src),
4047 (MOVAPSrm addr:$src)>;
4048 def : Pat<(loadv4i32 addr:$src),
4049 (MOVUPSrm addr:$src)>;
4050 def : Pat<(alignedloadv2i64 addr:$src),
4051 (MOVAPSrm addr:$src)>;
4052 def : Pat<(loadv2i64 addr:$src),
4053 (MOVUPSrm addr:$src)>;
4055 def : Pat<(alignedstore (v2i64 VR128:$src), addr:$dst),
4056 (MOVAPSmr addr:$dst, VR128:$src)>;
4057 def : Pat<(alignedstore (v4i32 VR128:$src), addr:$dst),
4058 (MOVAPSmr addr:$dst, VR128:$src)>;
4059 def : Pat<(alignedstore (v8i16 VR128:$src), addr:$dst),
4060 (MOVAPSmr addr:$dst, VR128:$src)>;
4061 def : Pat<(alignedstore (v16i8 VR128:$src), addr:$dst),
4062 (MOVAPSmr addr:$dst, VR128:$src)>;
4063 def : Pat<(store (v2i64 VR128:$src), addr:$dst),
4064 (MOVUPSmr addr:$dst, VR128:$src)>;
4065 def : Pat<(store (v4i32 VR128:$src), addr:$dst),
4066 (MOVUPSmr addr:$dst, VR128:$src)>;
4067 def : Pat<(store (v8i16 VR128:$src), addr:$dst),
4068 (MOVUPSmr addr:$dst, VR128:$src)>;
4069 def : Pat<(store (v16i8 VR128:$src), addr:$dst),
4070 (MOVUPSmr addr:$dst, VR128:$src)>;
4072 //===----------------------------------------------------------------------===//
4073 // SSE4.1 - Packed Move with Sign/Zero Extend
4074 //===----------------------------------------------------------------------===//
4076 multiclass SS41I_binop_rm_int8<bits<8> opc, string OpcodeStr, Intrinsic IntId> {
4077 def rr : SS48I<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
4078 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
4079 [(set VR128:$dst, (IntId VR128:$src))]>, OpSize;
4081 def rm : SS48I<opc, MRMSrcMem, (outs VR128:$dst), (ins i64mem:$src),
4082 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
4084 (IntId (bitconvert (v2i64 (scalar_to_vector (loadi64 addr:$src))))))]>,
4088 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
4089 defm VPMOVSXBW : SS41I_binop_rm_int8<0x20, "vpmovsxbw", int_x86_sse41_pmovsxbw>,
4091 defm VPMOVSXWD : SS41I_binop_rm_int8<0x23, "vpmovsxwd", int_x86_sse41_pmovsxwd>,
4093 defm VPMOVSXDQ : SS41I_binop_rm_int8<0x25, "vpmovsxdq", int_x86_sse41_pmovsxdq>,
4095 defm VPMOVZXBW : SS41I_binop_rm_int8<0x30, "vpmovzxbw", int_x86_sse41_pmovzxbw>,
4097 defm VPMOVZXWD : SS41I_binop_rm_int8<0x33, "vpmovzxwd", int_x86_sse41_pmovzxwd>,
4099 defm VPMOVZXDQ : SS41I_binop_rm_int8<0x35, "vpmovzxdq", int_x86_sse41_pmovzxdq>,
4103 defm PMOVSXBW : SS41I_binop_rm_int8<0x20, "pmovsxbw", int_x86_sse41_pmovsxbw>;
4104 defm PMOVSXWD : SS41I_binop_rm_int8<0x23, "pmovsxwd", int_x86_sse41_pmovsxwd>;
4105 defm PMOVSXDQ : SS41I_binop_rm_int8<0x25, "pmovsxdq", int_x86_sse41_pmovsxdq>;
4106 defm PMOVZXBW : SS41I_binop_rm_int8<0x30, "pmovzxbw", int_x86_sse41_pmovzxbw>;
4107 defm PMOVZXWD : SS41I_binop_rm_int8<0x33, "pmovzxwd", int_x86_sse41_pmovzxwd>;
4108 defm PMOVZXDQ : SS41I_binop_rm_int8<0x35, "pmovzxdq", int_x86_sse41_pmovzxdq>;
4110 // Common patterns involving scalar load.
4111 def : Pat<(int_x86_sse41_pmovsxbw (vzmovl_v2i64 addr:$src)),
4112 (PMOVSXBWrm addr:$src)>, Requires<[HasSSE41]>;
4113 def : Pat<(int_x86_sse41_pmovsxbw (vzload_v2i64 addr:$src)),
4114 (PMOVSXBWrm addr:$src)>, Requires<[HasSSE41]>;
4116 def : Pat<(int_x86_sse41_pmovsxwd (vzmovl_v2i64 addr:$src)),
4117 (PMOVSXWDrm addr:$src)>, Requires<[HasSSE41]>;
4118 def : Pat<(int_x86_sse41_pmovsxwd (vzload_v2i64 addr:$src)),
4119 (PMOVSXWDrm addr:$src)>, Requires<[HasSSE41]>;
4121 def : Pat<(int_x86_sse41_pmovsxdq (vzmovl_v2i64 addr:$src)),
4122 (PMOVSXDQrm addr:$src)>, Requires<[HasSSE41]>;
4123 def : Pat<(int_x86_sse41_pmovsxdq (vzload_v2i64 addr:$src)),
4124 (PMOVSXDQrm addr:$src)>, Requires<[HasSSE41]>;
4126 def : Pat<(int_x86_sse41_pmovzxbw (vzmovl_v2i64 addr:$src)),
4127 (PMOVZXBWrm addr:$src)>, Requires<[HasSSE41]>;
4128 def : Pat<(int_x86_sse41_pmovzxbw (vzload_v2i64 addr:$src)),
4129 (PMOVZXBWrm addr:$src)>, Requires<[HasSSE41]>;
4131 def : Pat<(int_x86_sse41_pmovzxwd (vzmovl_v2i64 addr:$src)),
4132 (PMOVZXWDrm addr:$src)>, Requires<[HasSSE41]>;
4133 def : Pat<(int_x86_sse41_pmovzxwd (vzload_v2i64 addr:$src)),
4134 (PMOVZXWDrm addr:$src)>, Requires<[HasSSE41]>;
4136 def : Pat<(int_x86_sse41_pmovzxdq (vzmovl_v2i64 addr:$src)),
4137 (PMOVZXDQrm addr:$src)>, Requires<[HasSSE41]>;
4138 def : Pat<(int_x86_sse41_pmovzxdq (vzload_v2i64 addr:$src)),
4139 (PMOVZXDQrm addr:$src)>, Requires<[HasSSE41]>;
4142 multiclass SS41I_binop_rm_int4<bits<8> opc, string OpcodeStr, Intrinsic IntId> {
4143 def rr : SS48I<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
4144 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
4145 [(set VR128:$dst, (IntId VR128:$src))]>, OpSize;
4147 def rm : SS48I<opc, MRMSrcMem, (outs VR128:$dst), (ins i32mem:$src),
4148 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
4150 (IntId (bitconvert (v4i32 (scalar_to_vector (loadi32 addr:$src))))))]>,
4154 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
4155 defm VPMOVSXBD : SS41I_binop_rm_int4<0x21, "vpmovsxbd", int_x86_sse41_pmovsxbd>,
4157 defm VPMOVSXWQ : SS41I_binop_rm_int4<0x24, "vpmovsxwq", int_x86_sse41_pmovsxwq>,
4159 defm VPMOVZXBD : SS41I_binop_rm_int4<0x31, "vpmovzxbd", int_x86_sse41_pmovzxbd>,
4161 defm VPMOVZXWQ : SS41I_binop_rm_int4<0x34, "vpmovzxwq", int_x86_sse41_pmovzxwq>,
4165 defm PMOVSXBD : SS41I_binop_rm_int4<0x21, "pmovsxbd", int_x86_sse41_pmovsxbd>;
4166 defm PMOVSXWQ : SS41I_binop_rm_int4<0x24, "pmovsxwq", int_x86_sse41_pmovsxwq>;
4167 defm PMOVZXBD : SS41I_binop_rm_int4<0x31, "pmovzxbd", int_x86_sse41_pmovzxbd>;
4168 defm PMOVZXWQ : SS41I_binop_rm_int4<0x34, "pmovzxwq", int_x86_sse41_pmovzxwq>;
4170 // Common patterns involving scalar load
4171 def : Pat<(int_x86_sse41_pmovsxbd (vzmovl_v4i32 addr:$src)),
4172 (PMOVSXBDrm addr:$src)>, Requires<[HasSSE41]>;
4173 def : Pat<(int_x86_sse41_pmovsxwq (vzmovl_v4i32 addr:$src)),
4174 (PMOVSXWQrm addr:$src)>, Requires<[HasSSE41]>;
4176 def : Pat<(int_x86_sse41_pmovzxbd (vzmovl_v4i32 addr:$src)),
4177 (PMOVZXBDrm addr:$src)>, Requires<[HasSSE41]>;
4178 def : Pat<(int_x86_sse41_pmovzxwq (vzmovl_v4i32 addr:$src)),
4179 (PMOVZXWQrm addr:$src)>, Requires<[HasSSE41]>;
4182 multiclass SS41I_binop_rm_int2<bits<8> opc, string OpcodeStr, Intrinsic IntId> {
4183 def rr : SS48I<opc, MRMSrcReg, (outs VR128:$dst), (ins VR128:$src),
4184 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
4185 [(set VR128:$dst, (IntId VR128:$src))]>, OpSize;
4187 // Expecting a i16 load any extended to i32 value.
4188 def rm : SS48I<opc, MRMSrcMem, (outs VR128:$dst), (ins i16mem:$src),
4189 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
4190 [(set VR128:$dst, (IntId (bitconvert
4191 (v4i32 (scalar_to_vector (loadi16_anyext addr:$src))))))]>,
4195 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
4196 defm VPMOVSXBQ : SS41I_binop_rm_int2<0x22, "vpmovsxbq", int_x86_sse41_pmovsxbq>,
4198 defm VPMOVZXBQ : SS41I_binop_rm_int2<0x32, "vpmovzxbq", int_x86_sse41_pmovzxbq>,
4201 defm PMOVSXBQ : SS41I_binop_rm_int2<0x22, "pmovsxbq", int_x86_sse41_pmovsxbq>;
4202 defm PMOVZXBQ : SS41I_binop_rm_int2<0x32, "pmovzxbq", int_x86_sse41_pmovzxbq>;
4204 // Common patterns involving scalar load
4205 def : Pat<(int_x86_sse41_pmovsxbq
4206 (bitconvert (v4i32 (X86vzmovl
4207 (v4i32 (scalar_to_vector (loadi32 addr:$src))))))),
4208 (PMOVSXBQrm addr:$src)>, Requires<[HasSSE41]>;
4210 def : Pat<(int_x86_sse41_pmovzxbq
4211 (bitconvert (v4i32 (X86vzmovl
4212 (v4i32 (scalar_to_vector (loadi32 addr:$src))))))),
4213 (PMOVZXBQrm addr:$src)>, Requires<[HasSSE41]>;
4215 //===----------------------------------------------------------------------===//
4216 // SSE4.1 - Extract Instructions
4217 //===----------------------------------------------------------------------===//
4219 /// SS41I_binop_ext8 - SSE 4.1 extract 8 bits to 32 bit reg or 8 bit mem
4220 multiclass SS41I_extract8<bits<8> opc, string OpcodeStr> {
4221 def rr : SS4AIi8<opc, MRMDestReg, (outs GR32:$dst),
4222 (ins VR128:$src1, i32i8imm:$src2),
4223 !strconcat(OpcodeStr,
4224 "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4225 [(set GR32:$dst, (X86pextrb (v16i8 VR128:$src1), imm:$src2))]>,
4227 def mr : SS4AIi8<opc, MRMDestMem, (outs),
4228 (ins i8mem:$dst, VR128:$src1, i32i8imm:$src2),
4229 !strconcat(OpcodeStr,
4230 "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4233 // There's an AssertZext in the way of writing the store pattern
4234 // (store (i8 (trunc (X86pextrb (v16i8 VR128:$src1), imm:$src2))), addr:$dst)
4237 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
4238 defm VPEXTRB : SS41I_extract8<0x14, "vpextrb">, VEX;
4239 def VPEXTRBrr64 : SS4AIi8<0x14, MRMDestReg, (outs GR64:$dst),
4240 (ins VR128:$src1, i32i8imm:$src2),
4241 "vpextrb\t{$src2, $src1, $dst|$dst, $src1, $src2}", []>, OpSize, VEX;
4244 defm PEXTRB : SS41I_extract8<0x14, "pextrb">;
4247 /// SS41I_extract16 - SSE 4.1 extract 16 bits to memory destination
4248 multiclass SS41I_extract16<bits<8> opc, string OpcodeStr> {
4249 def mr : SS4AIi8<opc, MRMDestMem, (outs),
4250 (ins i16mem:$dst, VR128:$src1, i32i8imm:$src2),
4251 !strconcat(OpcodeStr,
4252 "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4255 // There's an AssertZext in the way of writing the store pattern
4256 // (store (i16 (trunc (X86pextrw (v16i8 VR128:$src1), imm:$src2))), addr:$dst)
4259 let isAsmParserOnly = 1, Predicates = [HasAVX] in
4260 defm VPEXTRW : SS41I_extract16<0x15, "vpextrw">, VEX;
4262 defm PEXTRW : SS41I_extract16<0x15, "pextrw">;
4265 /// SS41I_extract32 - SSE 4.1 extract 32 bits to int reg or memory destination
4266 multiclass SS41I_extract32<bits<8> opc, string OpcodeStr> {
4267 def rr : SS4AIi8<opc, MRMDestReg, (outs GR32:$dst),
4268 (ins VR128:$src1, i32i8imm:$src2),
4269 !strconcat(OpcodeStr,
4270 "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4272 (extractelt (v4i32 VR128:$src1), imm:$src2))]>, OpSize;
4273 def mr : SS4AIi8<opc, MRMDestMem, (outs),
4274 (ins i32mem:$dst, VR128:$src1, i32i8imm:$src2),
4275 !strconcat(OpcodeStr,
4276 "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4277 [(store (extractelt (v4i32 VR128:$src1), imm:$src2),
4278 addr:$dst)]>, OpSize;
4281 let isAsmParserOnly = 1, Predicates = [HasAVX] in
4282 defm VPEXTRD : SS41I_extract32<0x16, "vpextrd">, VEX;
4284 defm PEXTRD : SS41I_extract32<0x16, "pextrd">;
4286 /// SS41I_extract32 - SSE 4.1 extract 32 bits to int reg or memory destination
4287 multiclass SS41I_extract64<bits<8> opc, string OpcodeStr> {
4288 def rr : SS4AIi8<opc, MRMDestReg, (outs GR64:$dst),
4289 (ins VR128:$src1, i32i8imm:$src2),
4290 !strconcat(OpcodeStr,
4291 "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4293 (extractelt (v2i64 VR128:$src1), imm:$src2))]>, OpSize, REX_W;
4294 def mr : SS4AIi8<opc, MRMDestMem, (outs),
4295 (ins i64mem:$dst, VR128:$src1, i32i8imm:$src2),
4296 !strconcat(OpcodeStr,
4297 "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4298 [(store (extractelt (v2i64 VR128:$src1), imm:$src2),
4299 addr:$dst)]>, OpSize, REX_W;
4302 let isAsmParserOnly = 1, Predicates = [HasAVX] in
4303 defm VPEXTRQ : SS41I_extract64<0x16, "vpextrq">, VEX, VEX_W;
4305 defm PEXTRQ : SS41I_extract64<0x16, "pextrq">;
4307 /// SS41I_extractf32 - SSE 4.1 extract 32 bits fp value to int reg or memory
4309 multiclass SS41I_extractf32<bits<8> opc, string OpcodeStr> {
4310 def rr : SS4AIi8<opc, MRMDestReg, (outs GR32:$dst),
4311 (ins VR128:$src1, i32i8imm:$src2),
4312 !strconcat(OpcodeStr,
4313 "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4315 (extractelt (bc_v4i32 (v4f32 VR128:$src1)), imm:$src2))]>,
4317 def mr : SS4AIi8<opc, MRMDestMem, (outs),
4318 (ins f32mem:$dst, VR128:$src1, i32i8imm:$src2),
4319 !strconcat(OpcodeStr,
4320 "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4321 [(store (extractelt (bc_v4i32 (v4f32 VR128:$src1)), imm:$src2),
4322 addr:$dst)]>, OpSize;
4325 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
4326 defm VEXTRACTPS : SS41I_extractf32<0x17, "vextractps">, VEX;
4327 def VEXTRACTPSrr64 : SS4AIi8<0x17, MRMDestReg, (outs GR64:$dst),
4328 (ins VR128:$src1, i32i8imm:$src2),
4329 "vextractps \t{$src2, $src1, $dst|$dst, $src1, $src2}",
4332 defm EXTRACTPS : SS41I_extractf32<0x17, "extractps">;
4334 // Also match an EXTRACTPS store when the store is done as f32 instead of i32.
4335 def : Pat<(store (f32 (bitconvert (extractelt (bc_v4i32 (v4f32 VR128:$src1)),
4338 (EXTRACTPSmr addr:$dst, VR128:$src1, imm:$src2)>,
4339 Requires<[HasSSE41]>;
4341 //===----------------------------------------------------------------------===//
4342 // SSE4.1 - Insert Instructions
4343 //===----------------------------------------------------------------------===//
4345 multiclass SS41I_insert8<bits<8> opc, string asm, bit Is2Addr = 1> {
4346 def rr : SS4AIi8<opc, MRMSrcReg, (outs VR128:$dst),
4347 (ins VR128:$src1, GR32:$src2, i32i8imm:$src3),
4349 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4351 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4353 (X86pinsrb VR128:$src1, GR32:$src2, imm:$src3))]>, OpSize;
4354 def rm : SS4AIi8<opc, MRMSrcMem, (outs VR128:$dst),
4355 (ins VR128:$src1, i8mem:$src2, i32i8imm:$src3),
4357 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4359 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4361 (X86pinsrb VR128:$src1, (extloadi8 addr:$src2),
4362 imm:$src3))]>, OpSize;
4365 let isAsmParserOnly = 1, Predicates = [HasAVX] in
4366 defm VPINSRB : SS41I_insert8<0x20, "vpinsrb", 0>, VEX_4V;
4367 let Constraints = "$src1 = $dst" in
4368 defm PINSRB : SS41I_insert8<0x20, "pinsrb">;
4370 multiclass SS41I_insert32<bits<8> opc, string asm, bit Is2Addr = 1> {
4371 def rr : SS4AIi8<opc, MRMSrcReg, (outs VR128:$dst),
4372 (ins VR128:$src1, GR32:$src2, i32i8imm:$src3),
4374 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4376 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4378 (v4i32 (insertelt VR128:$src1, GR32:$src2, imm:$src3)))]>,
4380 def rm : SS4AIi8<opc, MRMSrcMem, (outs VR128:$dst),
4381 (ins VR128:$src1, i32mem:$src2, i32i8imm:$src3),
4383 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4385 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4387 (v4i32 (insertelt VR128:$src1, (loadi32 addr:$src2),
4388 imm:$src3)))]>, OpSize;
4391 let isAsmParserOnly = 1, Predicates = [HasAVX] in
4392 defm VPINSRD : SS41I_insert32<0x22, "vpinsrd", 0>, VEX_4V;
4393 let Constraints = "$src1 = $dst" in
4394 defm PINSRD : SS41I_insert32<0x22, "pinsrd">;
4396 multiclass SS41I_insert64<bits<8> opc, string asm, bit Is2Addr = 1> {
4397 def rr : SS4AIi8<opc, MRMSrcReg, (outs VR128:$dst),
4398 (ins VR128:$src1, GR64:$src2, i32i8imm:$src3),
4400 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4402 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4404 (v2i64 (insertelt VR128:$src1, GR64:$src2, imm:$src3)))]>,
4406 def rm : SS4AIi8<opc, MRMSrcMem, (outs VR128:$dst),
4407 (ins VR128:$src1, i64mem:$src2, i32i8imm:$src3),
4409 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4411 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4413 (v2i64 (insertelt VR128:$src1, (loadi64 addr:$src2),
4414 imm:$src3)))]>, OpSize;
4417 let isAsmParserOnly = 1, Predicates = [HasAVX] in
4418 defm VPINSRQ : SS41I_insert64<0x22, "vpinsrq", 0>, VEX_4V, VEX_W;
4419 let Constraints = "$src1 = $dst" in
4420 defm PINSRQ : SS41I_insert64<0x22, "pinsrq">, REX_W;
4422 // insertps has a few different modes, there's the first two here below which
4423 // are optimized inserts that won't zero arbitrary elements in the destination
4424 // vector. The next one matches the intrinsic and could zero arbitrary elements
4425 // in the target vector.
4426 multiclass SS41I_insertf32<bits<8> opc, string asm, bit Is2Addr = 1> {
4427 def rr : SS4AIi8<opc, MRMSrcReg, (outs VR128:$dst),
4428 (ins VR128:$src1, VR128:$src2, i32i8imm:$src3),
4430 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4432 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4434 (X86insrtps VR128:$src1, VR128:$src2, imm:$src3))]>,
4436 def rm : SS4AIi8<opc, MRMSrcMem, (outs VR128:$dst),
4437 (ins VR128:$src1, f32mem:$src2, i32i8imm:$src3),
4439 !strconcat(asm, "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4441 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4443 (X86insrtps VR128:$src1,
4444 (v4f32 (scalar_to_vector (loadf32 addr:$src2))),
4445 imm:$src3))]>, OpSize;
4448 let Constraints = "$src1 = $dst" in
4449 defm INSERTPS : SS41I_insertf32<0x21, "insertps">;
4450 let isAsmParserOnly = 1, Predicates = [HasAVX] in
4451 defm VINSERTPS : SS41I_insertf32<0x21, "vinsertps", 0>, VEX_4V;
4453 def : Pat<(int_x86_sse41_insertps VR128:$src1, VR128:$src2, imm:$src3),
4454 (VINSERTPSrr VR128:$src1, VR128:$src2, imm:$src3)>,
4456 def : Pat<(int_x86_sse41_insertps VR128:$src1, VR128:$src2, imm:$src3),
4457 (INSERTPSrr VR128:$src1, VR128:$src2, imm:$src3)>,
4458 Requires<[HasSSE41]>;
4460 //===----------------------------------------------------------------------===//
4461 // SSE4.1 - Round Instructions
4462 //===----------------------------------------------------------------------===//
4464 multiclass sse41_fp_unop_rm<bits<8> opcps, bits<8> opcpd, string OpcodeStr,
4465 X86MemOperand x86memop, RegisterClass RC,
4466 PatFrag mem_frag32, PatFrag mem_frag64,
4467 Intrinsic V4F32Int, Intrinsic V2F64Int> {
4468 // Intrinsic operation, reg.
4469 // Vector intrinsic operation, reg
4470 def PSr_Int : SS4AIi8<opcps, MRMSrcReg,
4471 (outs RC:$dst), (ins RC:$src1, i32i8imm:$src2),
4472 !strconcat(OpcodeStr,
4473 "ps\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4474 [(set RC:$dst, (V4F32Int RC:$src1, imm:$src2))]>,
4477 // Vector intrinsic operation, mem
4478 def PSm_Int : Ii8<opcps, MRMSrcMem,
4479 (outs RC:$dst), (ins f256mem:$src1, i32i8imm:$src2),
4480 !strconcat(OpcodeStr,
4481 "ps\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4483 (V4F32Int (mem_frag32 addr:$src1),imm:$src2))]>,
4485 Requires<[HasSSE41]>;
4487 // Vector intrinsic operation, reg
4488 def PDr_Int : SS4AIi8<opcpd, MRMSrcReg,
4489 (outs RC:$dst), (ins RC:$src1, i32i8imm:$src2),
4490 !strconcat(OpcodeStr,
4491 "pd\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4492 [(set RC:$dst, (V2F64Int RC:$src1, imm:$src2))]>,
4495 // Vector intrinsic operation, mem
4496 def PDm_Int : SS4AIi8<opcpd, MRMSrcMem,
4497 (outs RC:$dst), (ins f256mem:$src1, i32i8imm:$src2),
4498 !strconcat(OpcodeStr,
4499 "pd\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4501 (V2F64Int (mem_frag64 addr:$src1),imm:$src2))]>,
4505 multiclass sse41_fp_unop_rm_avx_p<bits<8> opcps, bits<8> opcpd,
4506 RegisterClass RC, X86MemOperand x86memop, string OpcodeStr> {
4507 // Intrinsic operation, reg.
4508 // Vector intrinsic operation, reg
4509 def PSr : SS4AIi8<opcps, MRMSrcReg,
4510 (outs RC:$dst), (ins RC:$src1, i32i8imm:$src2),
4511 !strconcat(OpcodeStr,
4512 "ps\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4515 // Vector intrinsic operation, mem
4516 def PSm : Ii8<opcps, MRMSrcMem,
4517 (outs RC:$dst), (ins x86memop:$src1, i32i8imm:$src2),
4518 !strconcat(OpcodeStr,
4519 "ps\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4520 []>, TA, OpSize, Requires<[HasSSE41]>;
4522 // Vector intrinsic operation, reg
4523 def PDr : SS4AIi8<opcpd, MRMSrcReg,
4524 (outs RC:$dst), (ins RC:$src1, i32i8imm:$src2),
4525 !strconcat(OpcodeStr,
4526 "pd\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4529 // Vector intrinsic operation, mem
4530 def PDm : SS4AIi8<opcpd, MRMSrcMem,
4531 (outs RC:$dst), (ins x86memop:$src1, i32i8imm:$src2),
4532 !strconcat(OpcodeStr,
4533 "pd\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
4537 multiclass sse41_fp_binop_rm<bits<8> opcss, bits<8> opcsd,
4540 Intrinsic F64Int, bit Is2Addr = 1> {
4541 // Intrinsic operation, reg.
4542 def SSr_Int : SS4AIi8<opcss, MRMSrcReg,
4543 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2, i32i8imm:$src3),
4545 !strconcat(OpcodeStr,
4546 "ss\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4547 !strconcat(OpcodeStr,
4548 "ss\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4549 [(set VR128:$dst, (F32Int VR128:$src1, VR128:$src2, imm:$src3))]>,
4552 // Intrinsic operation, mem.
4553 def SSm_Int : SS4AIi8<opcss, MRMSrcMem,
4554 (outs VR128:$dst), (ins VR128:$src1, ssmem:$src2, i32i8imm:$src3),
4556 !strconcat(OpcodeStr,
4557 "ss\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4558 !strconcat(OpcodeStr,
4559 "ss\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4561 (F32Int VR128:$src1, sse_load_f32:$src2, imm:$src3))]>,
4564 // Intrinsic operation, reg.
4565 def SDr_Int : SS4AIi8<opcsd, MRMSrcReg,
4566 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2, i32i8imm:$src3),
4568 !strconcat(OpcodeStr,
4569 "sd\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4570 !strconcat(OpcodeStr,
4571 "sd\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4572 [(set VR128:$dst, (F64Int VR128:$src1, VR128:$src2, imm:$src3))]>,
4575 // Intrinsic operation, mem.
4576 def SDm_Int : SS4AIi8<opcsd, MRMSrcMem,
4577 (outs VR128:$dst), (ins VR128:$src1, sdmem:$src2, i32i8imm:$src3),
4579 !strconcat(OpcodeStr,
4580 "sd\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4581 !strconcat(OpcodeStr,
4582 "sd\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4584 (F64Int VR128:$src1, sse_load_f64:$src2, imm:$src3))]>,
4588 multiclass sse41_fp_binop_rm_avx_s<bits<8> opcss, bits<8> opcsd,
4590 // Intrinsic operation, reg.
4591 def SSr : SS4AIi8<opcss, MRMSrcReg,
4592 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2, i32i8imm:$src3),
4593 !strconcat(OpcodeStr,
4594 "ss\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
4597 // Intrinsic operation, mem.
4598 def SSm : SS4AIi8<opcss, MRMSrcMem,
4599 (outs VR128:$dst), (ins VR128:$src1, ssmem:$src2, i32i8imm:$src3),
4600 !strconcat(OpcodeStr,
4601 "ss\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
4604 // Intrinsic operation, reg.
4605 def SDr : SS4AIi8<opcsd, MRMSrcReg,
4606 (outs VR128:$dst), (ins VR128:$src1, VR128:$src2, i32i8imm:$src3),
4607 !strconcat(OpcodeStr,
4608 "sd\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
4611 // Intrinsic operation, mem.
4612 def SDm : SS4AIi8<opcsd, MRMSrcMem,
4613 (outs VR128:$dst), (ins VR128:$src1, sdmem:$src2, i32i8imm:$src3),
4614 !strconcat(OpcodeStr,
4615 "sd\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
4619 // FP round - roundss, roundps, roundsd, roundpd
4620 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
4622 defm VROUND : sse41_fp_unop_rm<0x08, 0x09, "vround", f128mem, VR128,
4623 memopv4f32, memopv2f64,
4624 int_x86_sse41_round_ps,
4625 int_x86_sse41_round_pd>, VEX;
4626 defm VROUNDY : sse41_fp_unop_rm<0x08, 0x09, "vround", f256mem, VR256,
4627 memopv8f32, memopv4f64,
4628 int_x86_avx_round_ps_256,
4629 int_x86_avx_round_pd_256>, VEX;
4630 defm VROUND : sse41_fp_binop_rm<0x0A, 0x0B, "vround",
4631 int_x86_sse41_round_ss,
4632 int_x86_sse41_round_sd, 0>, VEX_4V;
4634 // Instructions for the assembler
4635 defm VROUND : sse41_fp_unop_rm_avx_p<0x08, 0x09, VR128, f128mem, "vround">,
4637 defm VROUNDY : sse41_fp_unop_rm_avx_p<0x08, 0x09, VR256, f256mem, "vround">,
4639 defm VROUND : sse41_fp_binop_rm_avx_s<0x0A, 0x0B, "vround">, VEX_4V;
4642 defm ROUND : sse41_fp_unop_rm<0x08, 0x09, "round", f128mem, VR128,
4643 memopv4f32, memopv2f64,
4644 int_x86_sse41_round_ps, int_x86_sse41_round_pd>;
4645 let Constraints = "$src1 = $dst" in
4646 defm ROUND : sse41_fp_binop_rm<0x0A, 0x0B, "round",
4647 int_x86_sse41_round_ss, int_x86_sse41_round_sd>;
4649 //===----------------------------------------------------------------------===//
4650 // SSE4.1 - Packed Bit Test
4651 //===----------------------------------------------------------------------===//
4653 // ptest instruction we'll lower to this in X86ISelLowering primarily from
4654 // the intel intrinsic that corresponds to this.
4655 let Defs = [EFLAGS], isAsmParserOnly = 1, Predicates = [HasAVX] in {
4656 def VPTESTrr : SS48I<0x17, MRMSrcReg, (outs), (ins VR128:$src1, VR128:$src2),
4657 "vptest\t{$src2, $src1|$src1, $src2}",
4658 [(set EFLAGS, (X86ptest VR128:$src1, (v4f32 VR128:$src2)))]>,
4660 def VPTESTrm : SS48I<0x17, MRMSrcMem, (outs), (ins VR128:$src1, f128mem:$src2),
4661 "vptest\t{$src2, $src1|$src1, $src2}",
4662 [(set EFLAGS,(X86ptest VR128:$src1, (memopv4f32 addr:$src2)))]>,
4665 def VPTESTYrr : SS48I<0x17, MRMSrcReg, (outs), (ins VR256:$src1, VR256:$src2),
4666 "vptest\t{$src2, $src1|$src1, $src2}",
4667 [(set EFLAGS, (X86ptest VR256:$src1, (v4i64 VR256:$src2)))]>,
4669 def VPTESTYrm : SS48I<0x17, MRMSrcMem, (outs), (ins VR256:$src1, i256mem:$src2),
4670 "vptest\t{$src2, $src1|$src1, $src2}",
4671 [(set EFLAGS,(X86ptest VR256:$src1, (memopv4i64 addr:$src2)))]>,
4675 let Defs = [EFLAGS] in {
4676 def PTESTrr : SS48I<0x17, MRMSrcReg, (outs), (ins VR128:$src1, VR128:$src2),
4677 "ptest \t{$src2, $src1|$src1, $src2}",
4678 [(set EFLAGS, (X86ptest VR128:$src1, (v4f32 VR128:$src2)))]>,
4680 def PTESTrm : SS48I<0x17, MRMSrcMem, (outs), (ins VR128:$src1, f128mem:$src2),
4681 "ptest \t{$src2, $src1|$src1, $src2}",
4682 [(set EFLAGS, (X86ptest VR128:$src1, (memopv4f32 addr:$src2)))]>,
4686 // The bit test instructions below are AVX only
4687 multiclass avx_bittest<bits<8> opc, string OpcodeStr, RegisterClass RC,
4688 X86MemOperand x86memop, PatFrag mem_frag, ValueType vt> {
4689 def rr : SS48I<opc, MRMSrcReg, (outs), (ins RC:$src1, RC:$src2),
4690 !strconcat(OpcodeStr, "\t{$src2, $src1|$src1, $src2}"),
4691 [(set EFLAGS, (X86testp RC:$src1, (vt RC:$src2)))]>, OpSize, VEX;
4692 def rm : SS48I<opc, MRMSrcMem, (outs), (ins RC:$src1, x86memop:$src2),
4693 !strconcat(OpcodeStr, "\t{$src2, $src1|$src1, $src2}"),
4694 [(set EFLAGS, (X86testp RC:$src1, (mem_frag addr:$src2)))]>,
4698 let Defs = [EFLAGS], isAsmParserOnly = 1, Predicates = [HasAVX] in {
4699 defm VTESTPS : avx_bittest<0x0E, "vtestps", VR128, f128mem, memopv4f32, v4f32>;
4700 defm VTESTPSY : avx_bittest<0x0E, "vtestps", VR256, f256mem, memopv8f32, v8f32>;
4701 defm VTESTPD : avx_bittest<0x0F, "vtestpd", VR128, f128mem, memopv2f64, v2f64>;
4702 defm VTESTPDY : avx_bittest<0x0F, "vtestpd", VR256, f256mem, memopv4f64, v4f64>;
4705 //===----------------------------------------------------------------------===//
4706 // SSE4.1 - Misc Instructions
4707 //===----------------------------------------------------------------------===//
4709 // SS41I_unop_rm_int_v16 - SSE 4.1 unary operator whose type is v8i16.
4710 multiclass SS41I_unop_rm_int_v16<bits<8> opc, string OpcodeStr,
4711 Intrinsic IntId128> {
4712 def rr128 : SS48I<opc, MRMSrcReg, (outs VR128:$dst),
4714 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
4715 [(set VR128:$dst, (IntId128 VR128:$src))]>, OpSize;
4716 def rm128 : SS48I<opc, MRMSrcMem, (outs VR128:$dst),
4718 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
4721 (bitconvert (memopv8i16 addr:$src))))]>, OpSize;
4724 let isAsmParserOnly = 1, Predicates = [HasAVX] in
4725 defm VPHMINPOSUW : SS41I_unop_rm_int_v16 <0x41, "vphminposuw",
4726 int_x86_sse41_phminposuw>, VEX;
4727 defm PHMINPOSUW : SS41I_unop_rm_int_v16 <0x41, "phminposuw",
4728 int_x86_sse41_phminposuw>;
4730 /// SS41I_binop_rm_int - Simple SSE 4.1 binary operator
4731 multiclass SS41I_binop_rm_int<bits<8> opc, string OpcodeStr,
4732 Intrinsic IntId128, bit Is2Addr = 1> {
4733 let isCommutable = 1 in
4734 def rr : SS48I<opc, MRMSrcReg, (outs VR128:$dst),
4735 (ins VR128:$src1, VR128:$src2),
4737 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
4738 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
4739 [(set VR128:$dst, (IntId128 VR128:$src1, VR128:$src2))]>, OpSize;
4740 def rm : SS48I<opc, MRMSrcMem, (outs VR128:$dst),
4741 (ins VR128:$src1, i128mem:$src2),
4743 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
4744 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
4746 (IntId128 VR128:$src1,
4747 (bitconvert (memopv16i8 addr:$src2))))]>, OpSize;
4750 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
4751 let isCommutable = 0 in
4752 defm VPACKUSDW : SS41I_binop_rm_int<0x2B, "vpackusdw", int_x86_sse41_packusdw,
4754 defm VPCMPEQQ : SS41I_binop_rm_int<0x29, "vpcmpeqq", int_x86_sse41_pcmpeqq,
4756 defm VPMINSB : SS41I_binop_rm_int<0x38, "vpminsb", int_x86_sse41_pminsb,
4758 defm VPMINSD : SS41I_binop_rm_int<0x39, "vpminsd", int_x86_sse41_pminsd,
4760 defm VPMINUD : SS41I_binop_rm_int<0x3B, "vpminud", int_x86_sse41_pminud,
4762 defm VPMINUW : SS41I_binop_rm_int<0x3A, "vpminuw", int_x86_sse41_pminuw,
4764 defm VPMAXSB : SS41I_binop_rm_int<0x3C, "vpmaxsb", int_x86_sse41_pmaxsb,
4766 defm VPMAXSD : SS41I_binop_rm_int<0x3D, "vpmaxsd", int_x86_sse41_pmaxsd,
4768 defm VPMAXUD : SS41I_binop_rm_int<0x3F, "vpmaxud", int_x86_sse41_pmaxud,
4770 defm VPMAXUW : SS41I_binop_rm_int<0x3E, "vpmaxuw", int_x86_sse41_pmaxuw,
4772 defm VPMULDQ : SS41I_binop_rm_int<0x28, "vpmuldq", int_x86_sse41_pmuldq,
4776 let Constraints = "$src1 = $dst" in {
4777 let isCommutable = 0 in
4778 defm PACKUSDW : SS41I_binop_rm_int<0x2B, "packusdw", int_x86_sse41_packusdw>;
4779 defm PCMPEQQ : SS41I_binop_rm_int<0x29, "pcmpeqq", int_x86_sse41_pcmpeqq>;
4780 defm PMINSB : SS41I_binop_rm_int<0x38, "pminsb", int_x86_sse41_pminsb>;
4781 defm PMINSD : SS41I_binop_rm_int<0x39, "pminsd", int_x86_sse41_pminsd>;
4782 defm PMINUD : SS41I_binop_rm_int<0x3B, "pminud", int_x86_sse41_pminud>;
4783 defm PMINUW : SS41I_binop_rm_int<0x3A, "pminuw", int_x86_sse41_pminuw>;
4784 defm PMAXSB : SS41I_binop_rm_int<0x3C, "pmaxsb", int_x86_sse41_pmaxsb>;
4785 defm PMAXSD : SS41I_binop_rm_int<0x3D, "pmaxsd", int_x86_sse41_pmaxsd>;
4786 defm PMAXUD : SS41I_binop_rm_int<0x3F, "pmaxud", int_x86_sse41_pmaxud>;
4787 defm PMAXUW : SS41I_binop_rm_int<0x3E, "pmaxuw", int_x86_sse41_pmaxuw>;
4788 defm PMULDQ : SS41I_binop_rm_int<0x28, "pmuldq", int_x86_sse41_pmuldq>;
4791 def : Pat<(v2i64 (X86pcmpeqq VR128:$src1, VR128:$src2)),
4792 (PCMPEQQrr VR128:$src1, VR128:$src2)>;
4793 def : Pat<(v2i64 (X86pcmpeqq VR128:$src1, (memop addr:$src2))),
4794 (PCMPEQQrm VR128:$src1, addr:$src2)>;
4796 /// SS48I_binop_rm - Simple SSE41 binary operator.
4797 multiclass SS48I_binop_rm<bits<8> opc, string OpcodeStr, SDNode OpNode,
4798 ValueType OpVT, bit Is2Addr = 1> {
4799 let isCommutable = 1 in
4800 def rr : SS48I<opc, MRMSrcReg, (outs VR128:$dst),
4801 (ins VR128:$src1, VR128:$src2),
4803 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
4804 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
4805 [(set VR128:$dst, (OpVT (OpNode VR128:$src1, VR128:$src2)))]>,
4807 def rm : SS48I<opc, MRMSrcMem, (outs VR128:$dst),
4808 (ins VR128:$src1, i128mem:$src2),
4810 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
4811 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
4812 [(set VR128:$dst, (OpNode VR128:$src1,
4813 (bc_v4i32 (memopv2i64 addr:$src2))))]>,
4817 let isAsmParserOnly = 1, Predicates = [HasAVX] in
4818 defm VPMULLD : SS48I_binop_rm<0x40, "vpmulld", mul, v4i32, 0>, VEX_4V;
4819 let Constraints = "$src1 = $dst" in
4820 defm PMULLD : SS48I_binop_rm<0x40, "pmulld", mul, v4i32>;
4822 /// SS41I_binop_rmi_int - SSE 4.1 binary operator with 8-bit immediate
4823 multiclass SS41I_binop_rmi_int<bits<8> opc, string OpcodeStr,
4824 Intrinsic IntId, RegisterClass RC, PatFrag memop_frag,
4825 X86MemOperand x86memop, bit Is2Addr = 1> {
4826 let isCommutable = 1 in
4827 def rri : SS4AIi8<opc, MRMSrcReg, (outs RC:$dst),
4828 (ins RC:$src1, RC:$src2, i32i8imm:$src3),
4830 !strconcat(OpcodeStr,
4831 "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4832 !strconcat(OpcodeStr,
4833 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4834 [(set RC:$dst, (IntId RC:$src1, RC:$src2, imm:$src3))]>,
4836 def rmi : SS4AIi8<opc, MRMSrcMem, (outs RC:$dst),
4837 (ins RC:$src1, x86memop:$src2, i32i8imm:$src3),
4839 !strconcat(OpcodeStr,
4840 "\t{$src3, $src2, $dst|$dst, $src2, $src3}"),
4841 !strconcat(OpcodeStr,
4842 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}")),
4845 (bitconvert (memop_frag addr:$src2)), imm:$src3))]>,
4849 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
4850 let isCommutable = 0 in {
4851 defm VBLENDPS : SS41I_binop_rmi_int<0x0C, "vblendps", int_x86_sse41_blendps,
4852 VR128, memopv16i8, i128mem, 0>, VEX_4V;
4853 defm VBLENDPD : SS41I_binop_rmi_int<0x0D, "vblendpd", int_x86_sse41_blendpd,
4854 VR128, memopv16i8, i128mem, 0>, VEX_4V;
4855 defm VBLENDPSY : SS41I_binop_rmi_int<0x0C, "vblendps",
4856 int_x86_avx_blend_ps_256, VR256, memopv32i8, i256mem, 0>, VEX_4V;
4857 defm VBLENDPDY : SS41I_binop_rmi_int<0x0D, "vblendpd",
4858 int_x86_avx_blend_pd_256, VR256, memopv32i8, i256mem, 0>, VEX_4V;
4859 defm VPBLENDW : SS41I_binop_rmi_int<0x0E, "vpblendw", int_x86_sse41_pblendw,
4860 VR128, memopv16i8, i128mem, 0>, VEX_4V;
4861 defm VMPSADBW : SS41I_binop_rmi_int<0x42, "vmpsadbw", int_x86_sse41_mpsadbw,
4862 VR128, memopv16i8, i128mem, 0>, VEX_4V;
4864 defm VDPPS : SS41I_binop_rmi_int<0x40, "vdpps", int_x86_sse41_dpps,
4865 VR128, memopv16i8, i128mem, 0>, VEX_4V;
4866 defm VDPPD : SS41I_binop_rmi_int<0x41, "vdppd", int_x86_sse41_dppd,
4867 VR128, memopv16i8, i128mem, 0>, VEX_4V;
4868 defm VDPPSY : SS41I_binop_rmi_int<0x40, "vdpps", int_x86_avx_dp_ps_256,
4869 VR256, memopv32i8, i256mem, 0>, VEX_4V;
4872 let Constraints = "$src1 = $dst" in {
4873 let isCommutable = 0 in {
4874 defm BLENDPS : SS41I_binop_rmi_int<0x0C, "blendps", int_x86_sse41_blendps,
4875 VR128, memopv16i8, i128mem>;
4876 defm BLENDPD : SS41I_binop_rmi_int<0x0D, "blendpd", int_x86_sse41_blendpd,
4877 VR128, memopv16i8, i128mem>;
4878 defm PBLENDW : SS41I_binop_rmi_int<0x0E, "pblendw", int_x86_sse41_pblendw,
4879 VR128, memopv16i8, i128mem>;
4880 defm MPSADBW : SS41I_binop_rmi_int<0x42, "mpsadbw", int_x86_sse41_mpsadbw,
4881 VR128, memopv16i8, i128mem>;
4883 defm DPPS : SS41I_binop_rmi_int<0x40, "dpps", int_x86_sse41_dpps,
4884 VR128, memopv16i8, i128mem>;
4885 defm DPPD : SS41I_binop_rmi_int<0x41, "dppd", int_x86_sse41_dppd,
4886 VR128, memopv16i8, i128mem>;
4889 /// SS41I_quaternary_int_avx - AVX SSE 4.1 with 4 operators
4890 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
4891 multiclass SS41I_quaternary_int_avx<bits<8> opc, string OpcodeStr,
4892 RegisterClass RC, X86MemOperand x86memop,
4893 PatFrag mem_frag, Intrinsic IntId> {
4894 def rr : I<opc, MRMSrcReg, (outs RC:$dst),
4895 (ins RC:$src1, RC:$src2, RC:$src3),
4896 !strconcat(OpcodeStr,
4897 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
4898 [(set RC:$dst, (IntId RC:$src1, RC:$src2, RC:$src3))],
4899 SSEPackedInt>, OpSize, TA, VEX_4V, VEX_I8IMM;
4901 def rm : I<opc, MRMSrcMem, (outs RC:$dst),
4902 (ins RC:$src1, x86memop:$src2, RC:$src3),
4903 !strconcat(OpcodeStr,
4904 "\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}"),
4906 (IntId RC:$src1, (bitconvert (mem_frag addr:$src2)),
4908 SSEPackedInt>, OpSize, TA, VEX_4V, VEX_I8IMM;
4912 defm VBLENDVPD : SS41I_quaternary_int_avx<0x4B, "vblendvpd", VR128, i128mem,
4913 memopv16i8, int_x86_sse41_blendvpd>;
4914 defm VBLENDVPS : SS41I_quaternary_int_avx<0x4A, "vblendvps", VR128, i128mem,
4915 memopv16i8, int_x86_sse41_blendvps>;
4916 defm VPBLENDVB : SS41I_quaternary_int_avx<0x4C, "vpblendvb", VR128, i128mem,
4917 memopv16i8, int_x86_sse41_pblendvb>;
4918 defm VBLENDVPDY : SS41I_quaternary_int_avx<0x4B, "vblendvpd", VR256, i256mem,
4919 memopv32i8, int_x86_avx_blendv_pd_256>;
4920 defm VBLENDVPSY : SS41I_quaternary_int_avx<0x4A, "vblendvps", VR256, i256mem,
4921 memopv32i8, int_x86_avx_blendv_ps_256>;
4923 /// SS41I_ternary_int - SSE 4.1 ternary operator
4924 let Uses = [XMM0], Constraints = "$src1 = $dst" in {
4925 multiclass SS41I_ternary_int<bits<8> opc, string OpcodeStr, Intrinsic IntId> {
4926 def rr0 : SS48I<opc, MRMSrcReg, (outs VR128:$dst),
4927 (ins VR128:$src1, VR128:$src2),
4928 !strconcat(OpcodeStr,
4929 "\t{%xmm0, $src2, $dst|$dst, $src2, %xmm0}"),
4930 [(set VR128:$dst, (IntId VR128:$src1, VR128:$src2, XMM0))]>,
4933 def rm0 : SS48I<opc, MRMSrcMem, (outs VR128:$dst),
4934 (ins VR128:$src1, i128mem:$src2),
4935 !strconcat(OpcodeStr,
4936 "\t{%xmm0, $src2, $dst|$dst, $src2, %xmm0}"),
4939 (bitconvert (memopv16i8 addr:$src2)), XMM0))]>, OpSize;
4943 defm BLENDVPD : SS41I_ternary_int<0x15, "blendvpd", int_x86_sse41_blendvpd>;
4944 defm BLENDVPS : SS41I_ternary_int<0x14, "blendvps", int_x86_sse41_blendvps>;
4945 defm PBLENDVB : SS41I_ternary_int<0x10, "pblendvb", int_x86_sse41_pblendvb>;
4947 let isAsmParserOnly = 1, Predicates = [HasAVX] in
4948 def VMOVNTDQArm : SS48I<0x2A, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
4949 "vmovntdqa\t{$src, $dst|$dst, $src}",
4950 [(set VR128:$dst, (int_x86_sse41_movntdqa addr:$src))]>,
4952 def MOVNTDQArm : SS48I<0x2A, MRMSrcMem, (outs VR128:$dst), (ins i128mem:$src),
4953 "movntdqa\t{$src, $dst|$dst, $src}",
4954 [(set VR128:$dst, (int_x86_sse41_movntdqa addr:$src))]>,
4957 //===----------------------------------------------------------------------===//
4958 // SSE4.2 - Compare Instructions
4959 //===----------------------------------------------------------------------===//
4961 /// SS42I_binop_rm_int - Simple SSE 4.2 binary operator
4962 multiclass SS42I_binop_rm_int<bits<8> opc, string OpcodeStr,
4963 Intrinsic IntId128, bit Is2Addr = 1> {
4964 def rr : SS428I<opc, MRMSrcReg, (outs VR128:$dst),
4965 (ins VR128:$src1, VR128:$src2),
4967 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
4968 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
4969 [(set VR128:$dst, (IntId128 VR128:$src1, VR128:$src2))]>,
4971 def rm : SS428I<opc, MRMSrcMem, (outs VR128:$dst),
4972 (ins VR128:$src1, i128mem:$src2),
4974 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
4975 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
4977 (IntId128 VR128:$src1,
4978 (bitconvert (memopv16i8 addr:$src2))))]>, OpSize;
4981 let isAsmParserOnly = 1, Predicates = [HasAVX] in
4982 defm VPCMPGTQ : SS42I_binop_rm_int<0x37, "vpcmpgtq", int_x86_sse42_pcmpgtq,
4984 let Constraints = "$src1 = $dst" in
4985 defm PCMPGTQ : SS42I_binop_rm_int<0x37, "pcmpgtq", int_x86_sse42_pcmpgtq>;
4987 def : Pat<(v2i64 (X86pcmpgtq VR128:$src1, VR128:$src2)),
4988 (PCMPGTQrr VR128:$src1, VR128:$src2)>;
4989 def : Pat<(v2i64 (X86pcmpgtq VR128:$src1, (memop addr:$src2))),
4990 (PCMPGTQrm VR128:$src1, addr:$src2)>;
4992 //===----------------------------------------------------------------------===//
4993 // SSE4.2 - String/text Processing Instructions
4994 //===----------------------------------------------------------------------===//
4996 // Packed Compare Implicit Length Strings, Return Mask
4997 multiclass pseudo_pcmpistrm<string asm> {
4998 def REG : Ii8<0, Pseudo, (outs VR128:$dst),
4999 (ins VR128:$src1, VR128:$src2, i8imm:$src3), !strconcat(asm, "rr PSEUDO"),
5000 [(set VR128:$dst, (int_x86_sse42_pcmpistrm128 VR128:$src1, VR128:$src2,
5002 def MEM : Ii8<0, Pseudo, (outs VR128:$dst),
5003 (ins VR128:$src1, i128mem:$src2, i8imm:$src3), !strconcat(asm, "rm PSEUDO"),
5004 [(set VR128:$dst, (int_x86_sse42_pcmpistrm128
5005 VR128:$src1, (load addr:$src2), imm:$src3))]>;
5008 let Defs = [EFLAGS], usesCustomInserter = 1 in {
5009 defm PCMPISTRM128 : pseudo_pcmpistrm<"#PCMPISTRM128">, Requires<[HasSSE42]>;
5010 defm VPCMPISTRM128 : pseudo_pcmpistrm<"#VPCMPISTRM128">, Requires<[HasAVX]>;
5013 let Defs = [XMM0, EFLAGS], isAsmParserOnly = 1,
5014 Predicates = [HasAVX] in {
5015 def VPCMPISTRM128rr : SS42AI<0x62, MRMSrcReg, (outs),
5016 (ins VR128:$src1, VR128:$src2, i8imm:$src3),
5017 "vpcmpistrm\t{$src3, $src2, $src1|$src1, $src2, $src3}", []>, OpSize, VEX;
5018 def VPCMPISTRM128rm : SS42AI<0x62, MRMSrcMem, (outs),
5019 (ins VR128:$src1, i128mem:$src2, i8imm:$src3),
5020 "vpcmpistrm\t{$src3, $src2, $src1|$src1, $src2, $src3}", []>, OpSize, VEX;
5023 let Defs = [XMM0, EFLAGS] in {
5024 def PCMPISTRM128rr : SS42AI<0x62, MRMSrcReg, (outs),
5025 (ins VR128:$src1, VR128:$src2, i8imm:$src3),
5026 "pcmpistrm\t{$src3, $src2, $src1|$src1, $src2, $src3}", []>, OpSize;
5027 def PCMPISTRM128rm : SS42AI<0x62, MRMSrcMem, (outs),
5028 (ins VR128:$src1, i128mem:$src2, i8imm:$src3),
5029 "pcmpistrm\t{$src3, $src2, $src1|$src1, $src2, $src3}", []>, OpSize;
5032 // Packed Compare Explicit Length Strings, Return Mask
5033 multiclass pseudo_pcmpestrm<string asm> {
5034 def REG : Ii8<0, Pseudo, (outs VR128:$dst),
5035 (ins VR128:$src1, VR128:$src3, i8imm:$src5), !strconcat(asm, "rr PSEUDO"),
5036 [(set VR128:$dst, (int_x86_sse42_pcmpestrm128
5037 VR128:$src1, EAX, VR128:$src3, EDX, imm:$src5))]>;
5038 def MEM : Ii8<0, Pseudo, (outs VR128:$dst),
5039 (ins VR128:$src1, i128mem:$src3, i8imm:$src5), !strconcat(asm, "rm PSEUDO"),
5040 [(set VR128:$dst, (int_x86_sse42_pcmpestrm128
5041 VR128:$src1, EAX, (load addr:$src3), EDX, imm:$src5))]>;
5044 let Defs = [EFLAGS], Uses = [EAX, EDX], usesCustomInserter = 1 in {
5045 defm PCMPESTRM128 : pseudo_pcmpestrm<"#PCMPESTRM128">, Requires<[HasSSE42]>;
5046 defm VPCMPESTRM128 : pseudo_pcmpestrm<"#VPCMPESTRM128">, Requires<[HasAVX]>;
5049 let isAsmParserOnly = 1, Predicates = [HasAVX],
5050 Defs = [XMM0, EFLAGS], Uses = [EAX, EDX] in {
5051 def VPCMPESTRM128rr : SS42AI<0x60, MRMSrcReg, (outs),
5052 (ins VR128:$src1, VR128:$src3, i8imm:$src5),
5053 "vpcmpestrm\t{$src5, $src3, $src1|$src1, $src3, $src5}", []>, OpSize, VEX;
5054 def VPCMPESTRM128rm : SS42AI<0x60, MRMSrcMem, (outs),
5055 (ins VR128:$src1, i128mem:$src3, i8imm:$src5),
5056 "vpcmpestrm\t{$src5, $src3, $src1|$src1, $src3, $src5}", []>, OpSize, VEX;
5059 let Defs = [XMM0, EFLAGS], Uses = [EAX, EDX] in {
5060 def PCMPESTRM128rr : SS42AI<0x60, MRMSrcReg, (outs),
5061 (ins VR128:$src1, VR128:$src3, i8imm:$src5),
5062 "pcmpestrm\t{$src5, $src3, $src1|$src1, $src3, $src5}", []>, OpSize;
5063 def PCMPESTRM128rm : SS42AI<0x60, MRMSrcMem, (outs),
5064 (ins VR128:$src1, i128mem:$src3, i8imm:$src5),
5065 "pcmpestrm\t{$src5, $src3, $src1|$src1, $src3, $src5}", []>, OpSize;
5068 // Packed Compare Implicit Length Strings, Return Index
5069 let Defs = [ECX, EFLAGS] in {
5070 multiclass SS42AI_pcmpistri<Intrinsic IntId128, string asm = "pcmpistri"> {
5071 def rr : SS42AI<0x63, MRMSrcReg, (outs),
5072 (ins VR128:$src1, VR128:$src2, i8imm:$src3),
5073 !strconcat(asm, "\t{$src3, $src2, $src1|$src1, $src2, $src3}"),
5074 [(set ECX, (IntId128 VR128:$src1, VR128:$src2, imm:$src3)),
5075 (implicit EFLAGS)]>, OpSize;
5076 def rm : SS42AI<0x63, MRMSrcMem, (outs),
5077 (ins VR128:$src1, i128mem:$src2, i8imm:$src3),
5078 !strconcat(asm, "\t{$src3, $src2, $src1|$src1, $src2, $src3}"),
5079 [(set ECX, (IntId128 VR128:$src1, (load addr:$src2), imm:$src3)),
5080 (implicit EFLAGS)]>, OpSize;
5084 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
5085 defm VPCMPISTRI : SS42AI_pcmpistri<int_x86_sse42_pcmpistri128, "vpcmpistri">,
5087 defm VPCMPISTRIA : SS42AI_pcmpistri<int_x86_sse42_pcmpistria128, "vpcmpistri">,
5089 defm VPCMPISTRIC : SS42AI_pcmpistri<int_x86_sse42_pcmpistric128, "vpcmpistri">,
5091 defm VPCMPISTRIO : SS42AI_pcmpistri<int_x86_sse42_pcmpistrio128, "vpcmpistri">,
5093 defm VPCMPISTRIS : SS42AI_pcmpistri<int_x86_sse42_pcmpistris128, "vpcmpistri">,
5095 defm VPCMPISTRIZ : SS42AI_pcmpistri<int_x86_sse42_pcmpistriz128, "vpcmpistri">,
5099 defm PCMPISTRI : SS42AI_pcmpistri<int_x86_sse42_pcmpistri128>;
5100 defm PCMPISTRIA : SS42AI_pcmpistri<int_x86_sse42_pcmpistria128>;
5101 defm PCMPISTRIC : SS42AI_pcmpistri<int_x86_sse42_pcmpistric128>;
5102 defm PCMPISTRIO : SS42AI_pcmpistri<int_x86_sse42_pcmpistrio128>;
5103 defm PCMPISTRIS : SS42AI_pcmpistri<int_x86_sse42_pcmpistris128>;
5104 defm PCMPISTRIZ : SS42AI_pcmpistri<int_x86_sse42_pcmpistriz128>;
5106 // Packed Compare Explicit Length Strings, Return Index
5107 let Defs = [ECX, EFLAGS], Uses = [EAX, EDX] in {
5108 multiclass SS42AI_pcmpestri<Intrinsic IntId128, string asm = "pcmpestri"> {
5109 def rr : SS42AI<0x61, MRMSrcReg, (outs),
5110 (ins VR128:$src1, VR128:$src3, i8imm:$src5),
5111 !strconcat(asm, "\t{$src5, $src3, $src1|$src1, $src3, $src5}"),
5112 [(set ECX, (IntId128 VR128:$src1, EAX, VR128:$src3, EDX, imm:$src5)),
5113 (implicit EFLAGS)]>, OpSize;
5114 def rm : SS42AI<0x61, MRMSrcMem, (outs),
5115 (ins VR128:$src1, i128mem:$src3, i8imm:$src5),
5116 !strconcat(asm, "\t{$src5, $src3, $src1|$src1, $src3, $src5}"),
5118 (IntId128 VR128:$src1, EAX, (load addr:$src3), EDX, imm:$src5)),
5119 (implicit EFLAGS)]>, OpSize;
5123 let isAsmParserOnly = 1, Predicates = [HasAVX] in {
5124 defm VPCMPESTRI : SS42AI_pcmpestri<int_x86_sse42_pcmpestri128, "vpcmpestri">,
5126 defm VPCMPESTRIA : SS42AI_pcmpestri<int_x86_sse42_pcmpestria128, "vpcmpestri">,
5128 defm VPCMPESTRIC : SS42AI_pcmpestri<int_x86_sse42_pcmpestric128, "vpcmpestri">,
5130 defm VPCMPESTRIO : SS42AI_pcmpestri<int_x86_sse42_pcmpestrio128, "vpcmpestri">,
5132 defm VPCMPESTRIS : SS42AI_pcmpestri<int_x86_sse42_pcmpestris128, "vpcmpestri">,
5134 defm VPCMPESTRIZ : SS42AI_pcmpestri<int_x86_sse42_pcmpestriz128, "vpcmpestri">,
5138 defm PCMPESTRI : SS42AI_pcmpestri<int_x86_sse42_pcmpestri128>;
5139 defm PCMPESTRIA : SS42AI_pcmpestri<int_x86_sse42_pcmpestria128>;
5140 defm PCMPESTRIC : SS42AI_pcmpestri<int_x86_sse42_pcmpestric128>;
5141 defm PCMPESTRIO : SS42AI_pcmpestri<int_x86_sse42_pcmpestrio128>;
5142 defm PCMPESTRIS : SS42AI_pcmpestri<int_x86_sse42_pcmpestris128>;
5143 defm PCMPESTRIZ : SS42AI_pcmpestri<int_x86_sse42_pcmpestriz128>;
5145 //===----------------------------------------------------------------------===//
5146 // SSE4.2 - CRC Instructions
5147 //===----------------------------------------------------------------------===//
5149 // No CRC instructions have AVX equivalents
5151 // crc intrinsic instruction
5152 // This set of instructions are only rm, the only difference is the size
5154 let Constraints = "$src1 = $dst" in {
5155 def CRC32m8 : SS42FI<0xF0, MRMSrcMem, (outs GR32:$dst),
5156 (ins GR32:$src1, i8mem:$src2),
5157 "crc32{b} \t{$src2, $src1|$src1, $src2}",
5159 (int_x86_sse42_crc32_8 GR32:$src1,
5160 (load addr:$src2)))]>;
5161 def CRC32r8 : SS42FI<0xF0, MRMSrcReg, (outs GR32:$dst),
5162 (ins GR32:$src1, GR8:$src2),
5163 "crc32{b} \t{$src2, $src1|$src1, $src2}",
5165 (int_x86_sse42_crc32_8 GR32:$src1, GR8:$src2))]>;
5166 def CRC32m16 : SS42FI<0xF1, MRMSrcMem, (outs GR32:$dst),
5167 (ins GR32:$src1, i16mem:$src2),
5168 "crc32{w} \t{$src2, $src1|$src1, $src2}",
5170 (int_x86_sse42_crc32_16 GR32:$src1,
5171 (load addr:$src2)))]>,
5173 def CRC32r16 : SS42FI<0xF1, MRMSrcReg, (outs GR32:$dst),
5174 (ins GR32:$src1, GR16:$src2),
5175 "crc32{w} \t{$src2, $src1|$src1, $src2}",
5177 (int_x86_sse42_crc32_16 GR32:$src1, GR16:$src2))]>,
5179 def CRC32m32 : SS42FI<0xF1, MRMSrcMem, (outs GR32:$dst),
5180 (ins GR32:$src1, i32mem:$src2),
5181 "crc32{l} \t{$src2, $src1|$src1, $src2}",
5183 (int_x86_sse42_crc32_32 GR32:$src1,
5184 (load addr:$src2)))]>;
5185 def CRC32r32 : SS42FI<0xF1, MRMSrcReg, (outs GR32:$dst),
5186 (ins GR32:$src1, GR32:$src2),
5187 "crc32{l} \t{$src2, $src1|$src1, $src2}",
5189 (int_x86_sse42_crc32_32 GR32:$src1, GR32:$src2))]>;
5190 def CRC64m8 : SS42FI<0xF0, MRMSrcMem, (outs GR64:$dst),
5191 (ins GR64:$src1, i8mem:$src2),
5192 "crc32{b} \t{$src2, $src1|$src1, $src2}",
5194 (int_x86_sse42_crc64_8 GR64:$src1,
5195 (load addr:$src2)))]>,
5197 def CRC64r8 : SS42FI<0xF0, MRMSrcReg, (outs GR64:$dst),
5198 (ins GR64:$src1, GR8:$src2),
5199 "crc32{b} \t{$src2, $src1|$src1, $src2}",
5201 (int_x86_sse42_crc64_8 GR64:$src1, GR8:$src2))]>,
5203 def CRC64m64 : SS42FI<0xF1, MRMSrcMem, (outs GR64:$dst),
5204 (ins GR64:$src1, i64mem:$src2),
5205 "crc32{q} \t{$src2, $src1|$src1, $src2}",
5207 (int_x86_sse42_crc64_64 GR64:$src1,
5208 (load addr:$src2)))]>,
5210 def CRC64r64 : SS42FI<0xF1, MRMSrcReg, (outs GR64:$dst),
5211 (ins GR64:$src1, GR64:$src2),
5212 "crc32{q} \t{$src2, $src1|$src1, $src2}",
5214 (int_x86_sse42_crc64_64 GR64:$src1, GR64:$src2))]>,
5218 //===----------------------------------------------------------------------===//
5219 // AES-NI Instructions
5220 //===----------------------------------------------------------------------===//
5222 multiclass AESI_binop_rm_int<bits<8> opc, string OpcodeStr,
5223 Intrinsic IntId128, bit Is2Addr = 1> {
5224 def rr : AES8I<opc, MRMSrcReg, (outs VR128:$dst),
5225 (ins VR128:$src1, VR128:$src2),
5227 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
5228 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
5229 [(set VR128:$dst, (IntId128 VR128:$src1, VR128:$src2))]>,
5231 def rm : AES8I<opc, MRMSrcMem, (outs VR128:$dst),
5232 (ins VR128:$src1, i128mem:$src2),
5234 !strconcat(OpcodeStr, "\t{$src2, $dst|$dst, $src2}"),
5235 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}")),
5237 (IntId128 VR128:$src1,
5238 (bitconvert (memopv16i8 addr:$src2))))]>, OpSize;
5241 // Perform One Round of an AES Encryption/Decryption Flow
5242 let isAsmParserOnly = 1, Predicates = [HasAVX, HasAES] in {
5243 defm VAESENC : AESI_binop_rm_int<0xDC, "vaesenc",
5244 int_x86_aesni_aesenc, 0>, VEX_4V;
5245 defm VAESENCLAST : AESI_binop_rm_int<0xDD, "vaesenclast",
5246 int_x86_aesni_aesenclast, 0>, VEX_4V;
5247 defm VAESDEC : AESI_binop_rm_int<0xDE, "vaesdec",
5248 int_x86_aesni_aesdec, 0>, VEX_4V;
5249 defm VAESDECLAST : AESI_binop_rm_int<0xDF, "vaesdeclast",
5250 int_x86_aesni_aesdeclast, 0>, VEX_4V;
5253 let Constraints = "$src1 = $dst" in {
5254 defm AESENC : AESI_binop_rm_int<0xDC, "aesenc",
5255 int_x86_aesni_aesenc>;
5256 defm AESENCLAST : AESI_binop_rm_int<0xDD, "aesenclast",
5257 int_x86_aesni_aesenclast>;
5258 defm AESDEC : AESI_binop_rm_int<0xDE, "aesdec",
5259 int_x86_aesni_aesdec>;
5260 defm AESDECLAST : AESI_binop_rm_int<0xDF, "aesdeclast",
5261 int_x86_aesni_aesdeclast>;
5264 def : Pat<(v2i64 (int_x86_aesni_aesenc VR128:$src1, VR128:$src2)),
5265 (AESENCrr VR128:$src1, VR128:$src2)>;
5266 def : Pat<(v2i64 (int_x86_aesni_aesenc VR128:$src1, (memop addr:$src2))),
5267 (AESENCrm VR128:$src1, addr:$src2)>;
5268 def : Pat<(v2i64 (int_x86_aesni_aesenclast VR128:$src1, VR128:$src2)),
5269 (AESENCLASTrr VR128:$src1, VR128:$src2)>;
5270 def : Pat<(v2i64 (int_x86_aesni_aesenclast VR128:$src1, (memop addr:$src2))),
5271 (AESENCLASTrm VR128:$src1, addr:$src2)>;
5272 def : Pat<(v2i64 (int_x86_aesni_aesdec VR128:$src1, VR128:$src2)),
5273 (AESDECrr VR128:$src1, VR128:$src2)>;
5274 def : Pat<(v2i64 (int_x86_aesni_aesdec VR128:$src1, (memop addr:$src2))),
5275 (AESDECrm VR128:$src1, addr:$src2)>;
5276 def : Pat<(v2i64 (int_x86_aesni_aesdeclast VR128:$src1, VR128:$src2)),
5277 (AESDECLASTrr VR128:$src1, VR128:$src2)>;
5278 def : Pat<(v2i64 (int_x86_aesni_aesdeclast VR128:$src1, (memop addr:$src2))),
5279 (AESDECLASTrm VR128:$src1, addr:$src2)>;
5281 // Perform the AES InvMixColumn Transformation
5282 let isAsmParserOnly = 1, Predicates = [HasAVX, HasAES] in {
5283 def VAESIMCrr : AES8I<0xDB, MRMSrcReg, (outs VR128:$dst),
5285 "vaesimc\t{$src1, $dst|$dst, $src1}",
5287 (int_x86_aesni_aesimc VR128:$src1))]>,
5289 def VAESIMCrm : AES8I<0xDB, MRMSrcMem, (outs VR128:$dst),
5290 (ins i128mem:$src1),
5291 "vaesimc\t{$src1, $dst|$dst, $src1}",
5293 (int_x86_aesni_aesimc (bitconvert (memopv2i64 addr:$src1))))]>,
5296 def AESIMCrr : AES8I<0xDB, MRMSrcReg, (outs VR128:$dst),
5298 "aesimc\t{$src1, $dst|$dst, $src1}",
5300 (int_x86_aesni_aesimc VR128:$src1))]>,
5302 def AESIMCrm : AES8I<0xDB, MRMSrcMem, (outs VR128:$dst),
5303 (ins i128mem:$src1),
5304 "aesimc\t{$src1, $dst|$dst, $src1}",
5306 (int_x86_aesni_aesimc (bitconvert (memopv2i64 addr:$src1))))]>,
5309 // AES Round Key Generation Assist
5310 let isAsmParserOnly = 1, Predicates = [HasAVX, HasAES] in {
5311 def VAESKEYGENASSIST128rr : AESAI<0xDF, MRMSrcReg, (outs VR128:$dst),
5312 (ins VR128:$src1, i8imm:$src2),
5313 "vaeskeygenassist\t{$src2, $src1, $dst|$dst, $src1, $src2}",
5315 (int_x86_aesni_aeskeygenassist VR128:$src1, imm:$src2))]>,
5317 def VAESKEYGENASSIST128rm : AESAI<0xDF, MRMSrcMem, (outs VR128:$dst),
5318 (ins i128mem:$src1, i8imm:$src2),
5319 "vaeskeygenassist\t{$src2, $src1, $dst|$dst, $src1, $src2}",
5321 (int_x86_aesni_aeskeygenassist (bitconvert (memopv2i64 addr:$src1)),
5325 def AESKEYGENASSIST128rr : AESAI<0xDF, MRMSrcReg, (outs VR128:$dst),
5326 (ins VR128:$src1, i8imm:$src2),
5327 "aeskeygenassist\t{$src2, $src1, $dst|$dst, $src1, $src2}",
5329 (int_x86_aesni_aeskeygenassist VR128:$src1, imm:$src2))]>,
5331 def AESKEYGENASSIST128rm : AESAI<0xDF, MRMSrcMem, (outs VR128:$dst),
5332 (ins i128mem:$src1, i8imm:$src2),
5333 "aeskeygenassist\t{$src2, $src1, $dst|$dst, $src1, $src2}",
5335 (int_x86_aesni_aeskeygenassist (bitconvert (memopv2i64 addr:$src1)),
5339 //===----------------------------------------------------------------------===//
5340 // CLMUL Instructions
5341 //===----------------------------------------------------------------------===//
5343 // Only the AVX version of CLMUL instructions are described here.
5345 // Carry-less Multiplication instructions
5346 let isAsmParserOnly = 1 in {
5347 def VPCLMULQDQrr : CLMULIi8<0x44, MRMSrcReg, (outs VR128:$dst),
5348 (ins VR128:$src1, VR128:$src2, i8imm:$src3),
5349 "vpclmulqdq\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}",
5352 def VPCLMULQDQrm : CLMULIi8<0x44, MRMSrcMem, (outs VR128:$dst),
5353 (ins VR128:$src1, i128mem:$src2, i8imm:$src3),
5354 "vpclmulqdq\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}",
5358 multiclass avx_vpclmul<string asm> {
5359 def rr : I<0, Pseudo, (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
5360 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
5363 def rm : I<0, Pseudo, (outs VR128:$dst), (ins VR128:$src1, i128mem:$src2),
5364 !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
5367 defm VPCLMULHQHQDQ : avx_vpclmul<"vpclmulhqhqdq">;
5368 defm VPCLMULHQLQDQ : avx_vpclmul<"vpclmulhqlqdq">;
5369 defm VPCLMULLQHQDQ : avx_vpclmul<"vpclmullqhqdq">;
5370 defm VPCLMULLQLQDQ : avx_vpclmul<"vpclmullqlqdq">;
5372 } // isAsmParserOnly
5374 //===----------------------------------------------------------------------===//
5376 //===----------------------------------------------------------------------===//
5378 let isAsmParserOnly = 1 in {
5380 // Load from memory and broadcast to all elements of the destination operand
5381 class avx_broadcast<bits<8> opc, string OpcodeStr, RegisterClass RC,
5382 X86MemOperand x86memop, Intrinsic Int> :
5383 AVX8I<opc, MRMSrcMem, (outs RC:$dst), (ins x86memop:$src),
5384 !strconcat(OpcodeStr, "\t{$src, $dst|$dst, $src}"),
5385 [(set RC:$dst, (Int addr:$src))]>, VEX;
5387 def VBROADCASTSS : avx_broadcast<0x18, "vbroadcastss", VR128, f32mem,
5388 int_x86_avx_vbroadcastss>;
5389 def VBROADCASTSSY : avx_broadcast<0x18, "vbroadcastss", VR256, f32mem,
5390 int_x86_avx_vbroadcastss_256>;
5391 def VBROADCASTSD : avx_broadcast<0x19, "vbroadcastsd", VR256, f64mem,
5392 int_x86_avx_vbroadcast_sd_256>;
5393 def VBROADCASTF128 : avx_broadcast<0x1A, "vbroadcastf128", VR256, f128mem,
5394 int_x86_avx_vbroadcastf128_pd_256>;
5396 // Insert packed floating-point values
5397 def VINSERTF128rr : AVXAIi8<0x18, MRMSrcReg, (outs VR256:$dst),
5398 (ins VR256:$src1, VR128:$src2, i8imm:$src3),
5399 "vinsertf128\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}",
5401 def VINSERTF128rm : AVXAIi8<0x18, MRMSrcMem, (outs VR256:$dst),
5402 (ins VR256:$src1, f128mem:$src2, i8imm:$src3),
5403 "vinsertf128\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}",
5406 // Extract packed floating-point values
5407 def VEXTRACTF128rr : AVXAIi8<0x19, MRMDestReg, (outs VR128:$dst),
5408 (ins VR256:$src1, i8imm:$src2),
5409 "vextractf128\t{$src2, $src1, $dst|$dst, $src1, $src2}",
5411 def VEXTRACTF128mr : AVXAIi8<0x19, MRMDestMem, (outs),
5412 (ins f128mem:$dst, VR256:$src1, i8imm:$src2),
5413 "vextractf128\t{$src2, $src1, $dst|$dst, $src1, $src2}",
5416 // Conditional SIMD Packed Loads and Stores
5417 multiclass avx_movmask_rm<bits<8> opc_rm, bits<8> opc_mr, string OpcodeStr,
5418 Intrinsic IntLd, Intrinsic IntLd256,
5419 Intrinsic IntSt, Intrinsic IntSt256,
5420 PatFrag pf128, PatFrag pf256> {
5421 def rm : AVX8I<opc_rm, MRMSrcMem, (outs VR128:$dst),
5422 (ins VR128:$src1, f128mem:$src2),
5423 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
5424 [(set VR128:$dst, (IntLd addr:$src2, VR128:$src1))]>,
5426 def Yrm : AVX8I<opc_rm, MRMSrcMem, (outs VR256:$dst),
5427 (ins VR256:$src1, f256mem:$src2),
5428 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
5429 [(set VR256:$dst, (IntLd256 addr:$src2, VR256:$src1))]>,
5431 def mr : AVX8I<opc_mr, MRMDestMem, (outs),
5432 (ins f128mem:$dst, VR128:$src1, VR128:$src2),
5433 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
5434 [(IntSt addr:$dst, VR128:$src1, VR128:$src2)]>, VEX_4V;
5435 def Ymr : AVX8I<opc_mr, MRMDestMem, (outs),
5436 (ins f256mem:$dst, VR256:$src1, VR256:$src2),
5437 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
5438 [(IntSt256 addr:$dst, VR256:$src1, VR256:$src2)]>, VEX_4V;
5441 defm VMASKMOVPS : avx_movmask_rm<0x2C, 0x2E, "vmaskmovps",
5442 int_x86_avx_maskload_ps,
5443 int_x86_avx_maskload_ps_256,
5444 int_x86_avx_maskstore_ps,
5445 int_x86_avx_maskstore_ps_256,
5446 memopv4f32, memopv8f32>;
5447 defm VMASKMOVPD : avx_movmask_rm<0x2D, 0x2F, "vmaskmovpd",
5448 int_x86_avx_maskload_pd,
5449 int_x86_avx_maskload_pd_256,
5450 int_x86_avx_maskstore_pd,
5451 int_x86_avx_maskstore_pd_256,
5452 memopv2f64, memopv4f64>;
5454 // Permute Floating-Point Values
5455 multiclass avx_permil<bits<8> opc_rm, bits<8> opc_rmi, string OpcodeStr,
5456 RegisterClass RC, X86MemOperand x86memop_f,
5457 X86MemOperand x86memop_i, PatFrag f_frag, PatFrag i_frag,
5458 Intrinsic IntVar, Intrinsic IntImm> {
5459 def rr : AVX8I<opc_rm, MRMSrcReg, (outs RC:$dst),
5460 (ins RC:$src1, RC:$src2),
5461 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
5462 [(set RC:$dst, (IntVar RC:$src1, RC:$src2))]>, VEX_4V;
5463 def rm : AVX8I<opc_rm, MRMSrcMem, (outs RC:$dst),
5464 (ins RC:$src1, x86memop_i:$src2),
5465 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
5466 [(set RC:$dst, (IntVar RC:$src1, (i_frag addr:$src2)))]>, VEX_4V;
5468 def ri : AVXAIi8<opc_rmi, MRMSrcReg, (outs RC:$dst),
5469 (ins RC:$src1, i8imm:$src2),
5470 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
5471 [(set RC:$dst, (IntImm RC:$src1, imm:$src2))]>, VEX;
5472 def mi : AVXAIi8<opc_rmi, MRMSrcMem, (outs RC:$dst),
5473 (ins x86memop_f:$src1, i8imm:$src2),
5474 !strconcat(OpcodeStr, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
5475 [(set RC:$dst, (IntImm (f_frag addr:$src1), imm:$src2))]>, VEX;
5478 defm VPERMILPS : avx_permil<0x0C, 0x04, "vpermilps", VR128, f128mem, i128mem,
5479 memopv4f32, memopv4i32,
5480 int_x86_avx_vpermilvar_ps,
5481 int_x86_avx_vpermil_ps>;
5482 defm VPERMILPSY : avx_permil<0x0C, 0x04, "vpermilps", VR256, f256mem, i256mem,
5483 memopv8f32, memopv8i32,
5484 int_x86_avx_vpermilvar_ps_256,
5485 int_x86_avx_vpermil_ps_256>;
5486 defm VPERMILPD : avx_permil<0x0D, 0x05, "vpermilpd", VR128, f128mem, i128mem,
5487 memopv2f64, memopv2i64,
5488 int_x86_avx_vpermilvar_pd,
5489 int_x86_avx_vpermil_pd>;
5490 defm VPERMILPDY : avx_permil<0x0D, 0x05, "vpermilpd", VR256, f256mem, i256mem,
5491 memopv4f64, memopv4i64,
5492 int_x86_avx_vpermilvar_pd_256,
5493 int_x86_avx_vpermil_pd_256>;
5495 def VPERM2F128rr : AVXAIi8<0x06, MRMSrcReg, (outs VR256:$dst),
5496 (ins VR256:$src1, VR256:$src2, i8imm:$src3),
5497 "vperm2f128\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}",
5499 def VPERM2F128rm : AVXAIi8<0x06, MRMSrcMem, (outs VR256:$dst),
5500 (ins VR256:$src1, f256mem:$src2, i8imm:$src3),
5501 "vperm2f128\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}",
5504 // Zero All YMM registers
5505 def VZEROALL : I<0x77, RawFrm, (outs), (ins), "vzeroall",
5506 [(int_x86_avx_vzeroall)]>, VEX, VEX_L, Requires<[HasAVX]>;
5508 // Zero Upper bits of YMM registers
5509 def VZEROUPPER : I<0x77, RawFrm, (outs), (ins), "vzeroupper",
5510 [(int_x86_avx_vzeroupper)]>, VEX, Requires<[HasAVX]>;
5512 } // isAsmParserOnly
5514 def : Pat<(int_x86_avx_vinsertf128_pd_256 VR256:$src1, VR128:$src2, imm:$src3),
5515 (VINSERTF128rr VR256:$src1, VR128:$src2, imm:$src3)>;
5516 def : Pat<(int_x86_avx_vinsertf128_ps_256 VR256:$src1, VR128:$src2, imm:$src3),
5517 (VINSERTF128rr VR256:$src1, VR128:$src2, imm:$src3)>;
5518 def : Pat<(int_x86_avx_vinsertf128_si_256 VR256:$src1, VR128:$src2, imm:$src3),
5519 (VINSERTF128rr VR256:$src1, VR128:$src2, imm:$src3)>;
5521 def : Pat<(int_x86_avx_vextractf128_pd_256 VR256:$src1, imm:$src2),
5522 (VEXTRACTF128rr VR256:$src1, imm:$src2)>;
5523 def : Pat<(int_x86_avx_vextractf128_ps_256 VR256:$src1, imm:$src2),
5524 (VEXTRACTF128rr VR256:$src1, imm:$src2)>;
5525 def : Pat<(int_x86_avx_vextractf128_si_256 VR256:$src1, imm:$src2),
5526 (VEXTRACTF128rr VR256:$src1, imm:$src2)>;
5528 def : Pat<(int_x86_avx_vbroadcastf128_ps_256 addr:$src),
5529 (VBROADCASTF128 addr:$src)>;
5531 def : Pat<(int_x86_avx_vperm2f128_ps_256 VR256:$src1, VR256:$src2, imm:$src3),
5532 (VPERM2F128rr VR256:$src1, VR256:$src2, imm:$src3)>;
5533 def : Pat<(int_x86_avx_vperm2f128_pd_256 VR256:$src1, VR256:$src2, imm:$src3),
5534 (VPERM2F128rr VR256:$src1, VR256:$src2, imm:$src3)>;
5535 def : Pat<(int_x86_avx_vperm2f128_si_256 VR256:$src1, VR256:$src2, imm:$src3),
5536 (VPERM2F128rr VR256:$src1, VR256:$src2, imm:$src3)>;
5538 def : Pat<(int_x86_avx_vperm2f128_ps_256
5539 VR256:$src1, (memopv8f32 addr:$src2), imm:$src3),
5540 (VPERM2F128rm VR256:$src1, addr:$src2, imm:$src3)>;
5541 def : Pat<(int_x86_avx_vperm2f128_pd_256
5542 VR256:$src1, (memopv4f64 addr:$src2), imm:$src3),
5543 (VPERM2F128rm VR256:$src1, addr:$src2, imm:$src3)>;
5544 def : Pat<(int_x86_avx_vperm2f128_si_256
5545 VR256:$src1, (memopv8i32 addr:$src2), imm:$src3),
5546 (VPERM2F128rm VR256:$src1, addr:$src2, imm:$src3)>;