1 ; These are tests for SSE3 codegen.
3 ; RUN: llc < %s -march=x86-64 -mcpu=nocona -mtriple=i686-apple-darwin9 -O3 | FileCheck %s --check-prefix=X64
5 ; Test for v8xi16 lowering where we extract the first element of the vector and
6 ; placed it in the second element of the result.
8 define void @t0(<8 x i16>* %dest, <8 x i16>* %old) nounwind {
10 ; X64: ## BB#0: ## %entry
11 ; X64-NEXT: pxor %xmm0, %xmm0
12 ; X64-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
13 ; X64-NEXT: movdqa %xmm0, (%rdi)
16 %tmp3 = load <8 x i16>* %old
17 %tmp6 = shufflevector <8 x i16> %tmp3,
18 <8 x i16> < i16 0, i16 undef, i16 undef, i16 undef, i16 undef, i16 undef, i16 undef, i16 undef >,
19 <8 x i32> < i32 8, i32 0, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef >
20 store <8 x i16> %tmp6, <8 x i16>* %dest
25 define <8 x i16> @t1(<8 x i16>* %A, <8 x i16>* %B) nounwind {
28 ; X64-NEXT: movdqa (%rdi), %xmm0
29 ; X64-NEXT: pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
30 ; X64-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
31 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[2,1,2,3,4,5,6,7]
32 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
33 ; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
34 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,0,2,3,4,5,6,7]
35 ; X64-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
37 %tmp1 = load <8 x i16>* %A
38 %tmp2 = load <8 x i16>* %B
39 %tmp3 = shufflevector <8 x i16> %tmp1, <8 x i16> %tmp2, <8 x i32> < i32 8, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 >
44 define <8 x i16> @t2(<8 x i16> %A, <8 x i16> %B) nounwind {
47 ; X64-NEXT: pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
48 ; X64-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
49 ; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,1,2,3]
50 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[3,2,0,3,4,5,6,7]
51 ; X64-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
53 %tmp = shufflevector <8 x i16> %A, <8 x i16> %B, <8 x i32> < i32 9, i32 1, i32 2, i32 9, i32 4, i32 5, i32 6, i32 7 >
57 define <8 x i16> @t3(<8 x i16> %A, <8 x i16> %B) nounwind {
60 ; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
61 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6,5]
62 ; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
63 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,3,2,1,4,5,6,7]
64 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,6,5,4]
66 %tmp = shufflevector <8 x i16> %A, <8 x i16> %A, <8 x i32> < i32 8, i32 3, i32 2, i32 13, i32 7, i32 6, i32 5, i32 4 >
70 define <8 x i16> @t4(<8 x i16> %A, <8 x i16> %B) nounwind {
73 ; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
74 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,4,7]
75 ; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[3,1,2,0]
76 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,7,4,7]
78 %tmp = shufflevector <8 x i16> %A, <8 x i16> %B, <8 x i32> < i32 0, i32 7, i32 2, i32 3, i32 1, i32 5, i32 6, i32 5 >
82 define <8 x i16> @t5(<8 x i16> %A, <8 x i16> %B) nounwind {
85 ; X64-NEXT: punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
86 ; X64-NEXT: movdqa %xmm1, %xmm0
88 %tmp = shufflevector <8 x i16> %A, <8 x i16> %B, <8 x i32> < i32 8, i32 9, i32 0, i32 1, i32 10, i32 11, i32 2, i32 3 >
92 define <8 x i16> @t6(<8 x i16> %A, <8 x i16> %B) nounwind {
95 ; X64-NEXT: movss %xmm1, %xmm0
97 %tmp = shufflevector <8 x i16> %A, <8 x i16> %B, <8 x i32> < i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 >
101 define <8 x i16> @t7(<8 x i16> %A, <8 x i16> %B) nounwind {
104 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,0,3,2,4,5,6,7]
105 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,4,7]
107 %tmp = shufflevector <8 x i16> %A, <8 x i16> %B, <8 x i32> < i32 0, i32 0, i32 3, i32 2, i32 4, i32 6, i32 4, i32 7 >
111 define void @t8(<2 x i64>* %res, <2 x i64>* %A) nounwind {
114 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = mem[2,1,0,3,4,5,6,7]
115 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,5,4,7]
116 ; X64-NEXT: movdqa %xmm0, (%rdi)
118 %tmp = load <2 x i64>* %A
119 %tmp.upgrd.1 = bitcast <2 x i64> %tmp to <8 x i16>
120 %tmp0 = extractelement <8 x i16> %tmp.upgrd.1, i32 0
121 %tmp1 = extractelement <8 x i16> %tmp.upgrd.1, i32 1
122 %tmp2 = extractelement <8 x i16> %tmp.upgrd.1, i32 2
123 %tmp3 = extractelement <8 x i16> %tmp.upgrd.1, i32 3
124 %tmp4 = extractelement <8 x i16> %tmp.upgrd.1, i32 4
125 %tmp5 = extractelement <8 x i16> %tmp.upgrd.1, i32 5
126 %tmp6 = extractelement <8 x i16> %tmp.upgrd.1, i32 6
127 %tmp7 = extractelement <8 x i16> %tmp.upgrd.1, i32 7
128 %tmp8 = insertelement <8 x i16> undef, i16 %tmp2, i32 0
129 %tmp9 = insertelement <8 x i16> %tmp8, i16 %tmp1, i32 1
130 %tmp10 = insertelement <8 x i16> %tmp9, i16 %tmp0, i32 2
131 %tmp11 = insertelement <8 x i16> %tmp10, i16 %tmp3, i32 3
132 %tmp12 = insertelement <8 x i16> %tmp11, i16 %tmp6, i32 4
133 %tmp13 = insertelement <8 x i16> %tmp12, i16 %tmp5, i32 5
134 %tmp14 = insertelement <8 x i16> %tmp13, i16 %tmp4, i32 6
135 %tmp15 = insertelement <8 x i16> %tmp14, i16 %tmp7, i32 7
136 %tmp15.upgrd.2 = bitcast <8 x i16> %tmp15 to <2 x i64>
137 store <2 x i64> %tmp15.upgrd.2, <2 x i64>* %res
141 define void @t9(<4 x float>* %r, <2 x i32>* %A) nounwind {
144 ; X64-NEXT: movapd (%rdi), %xmm0
145 ; X64-NEXT: movhpd (%rsi), %xmm0
146 ; X64-NEXT: movapd %xmm0, (%rdi)
148 %tmp = load <4 x float>* %r
149 %tmp.upgrd.3 = bitcast <2 x i32>* %A to double*
150 %tmp.upgrd.4 = load double* %tmp.upgrd.3
151 %tmp.upgrd.5 = insertelement <2 x double> undef, double %tmp.upgrd.4, i32 0
152 %tmp5 = insertelement <2 x double> %tmp.upgrd.5, double undef, i32 1
153 %tmp6 = bitcast <2 x double> %tmp5 to <4 x float>
154 %tmp.upgrd.6 = extractelement <4 x float> %tmp, i32 0
155 %tmp7 = extractelement <4 x float> %tmp, i32 1
156 %tmp8 = extractelement <4 x float> %tmp6, i32 0
157 %tmp9 = extractelement <4 x float> %tmp6, i32 1
158 %tmp10 = insertelement <4 x float> undef, float %tmp.upgrd.6, i32 0
159 %tmp11 = insertelement <4 x float> %tmp10, float %tmp7, i32 1
160 %tmp12 = insertelement <4 x float> %tmp11, float %tmp8, i32 2
161 %tmp13 = insertelement <4 x float> %tmp12, float %tmp9, i32 3
162 store <4 x float> %tmp13, <4 x float>* %r
168 ; FIXME: This testcase produces icky code. It can be made much better!
171 @g1 = external constant <4 x i32>
172 @g2 = external constant <4 x i16>
174 define void @t10() nounwind {
177 ; X64-NEXT: movq _g1@{{.*}}(%rip), %rax
178 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = mem[0,2,2,3,4,5,6,7]
179 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
180 ; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
181 ; X64-NEXT: movq _g2@{{.*}}(%rip), %rax
182 ; X64-NEXT: movq %xmm0, (%rax)
184 load <4 x i32>* @g1, align 16
185 bitcast <4 x i32> %1 to <8 x i16>
186 shufflevector <8 x i16> %2, <8 x i16> undef, <8 x i32> < i32 0, i32 2, i32 4, i32 6, i32 undef, i32 undef, i32 undef, i32 undef >
187 bitcast <8 x i16> %3 to <2 x i64>
188 extractelement <2 x i64> %4, i32 0
189 bitcast i64 %5 to <4 x i16>
190 store <4 x i16> %6, <4 x i16>* @g2, align 8
194 ; Pack various elements via shuffles.
195 define <8 x i16> @t11(<8 x i16> %T0, <8 x i16> %T1) nounwind readnone {
197 ; X64: ## BB#0: ## %entry
198 ; X64-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
199 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[2,1,2,3,4,5,6,7]
202 %tmp7 = shufflevector <8 x i16> %T0, <8 x i16> %T1, <8 x i32> < i32 1, i32 8, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef , i32 undef >
207 define <8 x i16> @t12(<8 x i16> %T0, <8 x i16> %T1) nounwind readnone {
209 ; X64: ## BB#0: ## %entry
210 ; X64-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
211 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
212 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,7,6,7]
215 %tmp9 = shufflevector <8 x i16> %T0, <8 x i16> %T1, <8 x i32> < i32 0, i32 1, i32 undef, i32 undef, i32 3, i32 11, i32 undef , i32 undef >
220 define <8 x i16> @t13(<8 x i16> %T0, <8 x i16> %T1) nounwind readnone {
222 ; X64: ## BB#0: ## %entry
223 ; X64-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
224 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm1[0,2,2,3,4,5,6,7]
225 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,7,6,7]
228 %tmp9 = shufflevector <8 x i16> %T0, <8 x i16> %T1, <8 x i32> < i32 8, i32 9, i32 undef, i32 undef, i32 11, i32 3, i32 undef , i32 undef >
232 define <8 x i16> @t14(<8 x i16> %T0, <8 x i16> %T1) nounwind readnone {
234 ; X64: ## BB#0: ## %entry
235 ; X64-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
236 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm1[0,2,2,3,4,5,6,7]
239 %tmp9 = shufflevector <8 x i16> %T0, <8 x i16> %T1, <8 x i32> < i32 8, i32 9, i32 undef, i32 undef, i32 undef, i32 2, i32 undef , i32 undef >
243 ; FIXME: t15 is worse off from disabling of scheduler 2-address hack.
244 define <8 x i16> @t15(<8 x i16> %T0, <8 x i16> %T1) nounwind readnone {
246 ; X64: ## BB#0: ## %entry
247 ; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
248 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[1,1,2,3,4,5,6,7]
249 ; X64-NEXT: punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
250 ; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,0,3]
251 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm0[0,1,0,2,4,5,6,7]
252 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,6,7]
255 %tmp8 = shufflevector <8 x i16> %T0, <8 x i16> %T1, <8 x i32> < i32 undef, i32 undef, i32 7, i32 2, i32 8, i32 undef, i32 undef , i32 undef >
259 ; Test yonah where we convert a shuffle to pextrw and pinrsw
260 define <16 x i8> @t16(<16 x i8> %T0) nounwind readnone {
262 ; X64: ## BB#0: ## %entry
263 ; X64-NEXT: movdqa {{.*#+}} xmm1 = [0,0,0,0,1,1,1,1,0,0,0,0,0,0,0,0]
264 ; X64-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
265 ; X64-NEXT: pxor %xmm2, %xmm2
266 ; X64-NEXT: punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
267 ; X64-NEXT: punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
268 ; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
269 ; X64-NEXT: punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
270 ; X64-NEXT: pshuflw {{.*#+}} xmm0 = xmm1[0,2,2,3,4,5,6,7]
271 ; X64-NEXT: pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,7,6,7]
272 ; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
273 ; X64-NEXT: packuswb %xmm0, %xmm0
276 %tmp8 = shufflevector <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 1, i8 1, i8 1, i8 1, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, <16 x i8> %T0, <16 x i32> < i32 0, i32 1, i32 16, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef , i32 undef >
277 %tmp9 = shufflevector <16 x i8> %tmp8, <16 x i8> %T0, <16 x i32> < i32 0, i32 1, i32 2, i32 17, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef , i32 undef >
282 define <4 x i32> @t17() nounwind {
284 ; X64: ## BB#0: ## %entry
285 ; X64-NEXT: movddup (%rax), %xmm0
286 ; X64-NEXT: andpd {{.*}}(%rip), %xmm0
289 %tmp1 = load <4 x float>* undef, align 16
290 %tmp2 = shufflevector <4 x float> %tmp1, <4 x float> undef, <4 x i32> <i32 4, i32 1, i32 2, i32 3>
291 %tmp3 = load <4 x float>* undef, align 16
292 %tmp4 = shufflevector <4 x float> %tmp2, <4 x float> undef, <4 x i32> <i32 undef, i32 undef, i32 0, i32 1>
293 %tmp5 = bitcast <4 x float> %tmp3 to <4 x i32>
294 %tmp6 = shufflevector <4 x i32> %tmp5, <4 x i32> undef, <4 x i32> <i32 undef, i32 undef, i32 0, i32 1>
295 %tmp7 = and <4 x i32> %tmp6, <i32 undef, i32 undef, i32 -1, i32 0>