Sort bit assignments. Cosmetic change only.
[oota-llvm.git] / lib / Target / ARM / ARMScheduleA8.td
1 //=- ARMScheduleA8.td - ARM Cortex-A8 Scheduling Definitions -*- tablegen -*-=//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 // This file defines the itinerary class data for the ARM Cortex A8 processors.
11 //
12 //===----------------------------------------------------------------------===//
13
14 //
15 // Scheduling information derived from "Cortex-A8 Technical Reference Manual".
16 // Functional Units.
17 def A8_Pipe0   : FuncUnit; // pipeline 0
18 def A8_Pipe1   : FuncUnit; // pipeline 1
19 def A8_LSPipe  : FuncUnit; // Load / store pipeline
20 def A8_NPipe   : FuncUnit; // NEON ALU/MUL pipe
21 def A8_NLSPipe : FuncUnit; // NEON LS pipe
22 //
23 // Dual issue pipeline represented by A8_Pipe0 | A8_Pipe1
24 //
25 def CortexA8Itineraries : ProcessorItineraries<
26   [A8_Pipe0, A8_Pipe1, A8_LSPipe, A8_NPipe, A8_NLSPipe],
27   [], [
28   // Two fully-pipelined integer ALU pipelines
29   //
30   // No operand cycles
31   InstrItinData<IIC_iALUx    , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>]>,
32   //
33   // Binary Instructions that produce a result
34   InstrItinData<IIC_iALUi ,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2]>,
35   InstrItinData<IIC_iALUr ,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2, 2]>,
36   InstrItinData<IIC_iALUsi,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2, 1]>,
37   InstrItinData<IIC_iALUsir,[InstrStage<1,[A8_Pipe0, A8_Pipe1]>], [2, 1, 2]>,
38   InstrItinData<IIC_iALUsr,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2, 1, 1]>,
39   //
40   // Bitwise Instructions that produce a result
41   InstrItinData<IIC_iBITi ,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2]>,
42   InstrItinData<IIC_iBITr ,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2, 2]>,
43   InstrItinData<IIC_iBITsi,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2, 1]>,
44   InstrItinData<IIC_iBITsr,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2, 1, 1]>,
45   //
46   // Unary Instructions that produce a result
47   InstrItinData<IIC_iUNAr , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2]>,
48   InstrItinData<IIC_iUNAsi, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 1]>,
49   //
50   // Zero and sign extension instructions
51   InstrItinData<IIC_iEXTr , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1, 1]>,
52   InstrItinData<IIC_iEXTAr, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2, 1]>,
53   InstrItinData<IIC_iEXTAsr,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>],[2, 2, 1, 1]>,
54   //
55   // Compare instructions
56   InstrItinData<IIC_iCMPi , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2]>,
57   InstrItinData<IIC_iCMPr , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2]>,
58   InstrItinData<IIC_iCMPsi, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 1]>,
59   InstrItinData<IIC_iCMPsr, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 1, 1]>,
60   //
61   // Test instructions
62   InstrItinData<IIC_iTSTi , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2]>,
63   InstrItinData<IIC_iTSTr , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2]>,
64   InstrItinData<IIC_iTSTsi, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 1]>,
65   InstrItinData<IIC_iTSTsr, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 1, 1]>,
66   //
67   // Move instructions, unconditional
68   InstrItinData<IIC_iMOVi , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1]>,
69   InstrItinData<IIC_iMOVr , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1, 1]>,
70   InstrItinData<IIC_iMOVsi, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1, 1]>,
71   InstrItinData<IIC_iMOVsr, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1, 1, 1]>,
72   InstrItinData<IIC_iMOVix2,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
73                              InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2]>,
74   //
75   // Move instructions, conditional
76   InstrItinData<IIC_iCMOVi , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2]>,
77   InstrItinData<IIC_iCMOVr , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 1]>,
78   InstrItinData<IIC_iCMOVsi, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 1]>,
79   InstrItinData<IIC_iCMOVsr, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 1, 1]>,
80   //
81   // MVN instructions
82   InstrItinData<IIC_iMVNi , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1]>,
83   InstrItinData<IIC_iMVNr , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1, 1]>,
84   InstrItinData<IIC_iMVNsi, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1, 1]>,
85   InstrItinData<IIC_iMVNsr, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1, 1, 1]>,
86
87   // Integer multiply pipeline
88   // Result written in E5, but that is relative to the last cycle of multicycle,
89   // so we use 6 for those cases
90   //
91   InstrItinData<IIC_iMUL16   , [InstrStage<1, [A8_Pipe0]>], [5, 1, 1]>,
92   InstrItinData<IIC_iMAC16   , [InstrStage<2, [A8_Pipe0]>], [6, 1, 1, 4]>,
93   InstrItinData<IIC_iMUL32   , [InstrStage<2, [A8_Pipe0]>], [6, 1, 1]>,
94   InstrItinData<IIC_iMAC32   , [InstrStage<2, [A8_Pipe0]>], [6, 1, 1, 4]>,
95   InstrItinData<IIC_iMUL64   , [InstrStage<3, [A8_Pipe0]>], [6, 6, 1, 1]>,
96   InstrItinData<IIC_iMAC64   , [InstrStage<3, [A8_Pipe0]>], [6, 6, 1, 1]>,
97
98   // Integer load pipeline
99   //
100   // Immediate offset
101   InstrItinData<IIC_iLoad_i   , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
102                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
103   InstrItinData<IIC_iLoad_bh_i, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
104                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
105   InstrItinData<IIC_iLoad_d_i,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
106                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
107   //
108   // Register offset
109   InstrItinData<IIC_iLoad_r   , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
110                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
111   InstrItinData<IIC_iLoad_bh_r, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
112                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
113   InstrItinData<IIC_iLoad_d_r , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
114                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
115   //
116   // Scaled register offset, issues over 2 cycles
117   // FIXME: lsl by 2 takes 1 cycle.
118   InstrItinData<IIC_iLoad_si  , [InstrStage<2, [A8_Pipe0, A8_Pipe1]>,
119                                  InstrStage<1, [A8_LSPipe]>], [4, 1, 1]>,
120   InstrItinData<IIC_iLoad_bh_si,[InstrStage<2, [A8_Pipe0, A8_Pipe1]>,
121                                  InstrStage<1, [A8_LSPipe]>], [4, 1, 1]>,
122   //
123   // Immediate offset with update
124   InstrItinData<IIC_iLoad_iu  , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
125                                  InstrStage<1, [A8_LSPipe]>], [3, 2, 1]>,
126   InstrItinData<IIC_iLoad_bh_iu,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
127                                  InstrStage<1, [A8_LSPipe]>], [3, 2, 1]>,
128   //
129   // Register offset with update
130   InstrItinData<IIC_iLoad_ru  , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
131                                  InstrStage<1, [A8_LSPipe]>], [3, 2, 1, 1]>,
132   InstrItinData<IIC_iLoad_bh_ru,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
133                                  InstrStage<1, [A8_LSPipe]>], [3, 2, 1, 1]>,
134   InstrItinData<IIC_iLoad_d_ru, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
135                                  InstrStage<1, [A8_LSPipe]>], [3, 2, 1, 1]>,
136   //
137   // Scaled register offset with update, issues over 2 cycles
138   InstrItinData<IIC_iLoad_siu , [InstrStage<2, [A8_Pipe0, A8_Pipe1]>,
139                                  InstrStage<1, [A8_LSPipe]>], [4, 3, 1, 1]>,
140   InstrItinData<IIC_iLoad_bh_siu,[InstrStage<2, [A8_Pipe0, A8_Pipe1]>,
141                                   InstrStage<1, [A8_LSPipe]>], [4, 3, 1, 1]>,
142   //
143   // Load multiple, def is the 5th operand. Pipeline 0 only.
144   // FIXME: A8_LSPipe cycle time is dynamic, this assumes 3 to 4 registers.
145   InstrItinData<IIC_iLoad_m  , [InstrStage<1, [A8_Pipe0]>,
146                                 InstrStage<2, [A8_LSPipe]>], [1, 1, 1, 1, 3]>,
147   //
148   // Load multiple + update, defs are the 1st and 5th operands.
149   InstrItinData<IIC_iLoad_mu , [InstrStage<1, [A8_Pipe0]>,
150                                 InstrStage<3, [A8_LSPipe]>], [2, 1, 1, 1, 3]>,
151   //
152   // Load multiple plus branch
153   InstrItinData<IIC_iLoad_mBr, [InstrStage<1, [A8_Pipe0]>,
154                                 InstrStage<3, [A8_LSPipe]>,
155                                 InstrStage<1, [A8_Pipe0, A8_Pipe1]>],
156                                [1, 2, 1, 1, 3]>,
157   //
158   // Pop, def is the 3rd operand.
159   InstrItinData<IIC_iPop  ,    [InstrStage<1, [A8_Pipe0]>,
160                                 InstrStage<3, [A8_LSPipe]>], [1, 1, 3]>,
161   //
162   // Push, def is the 3th operand.
163   InstrItinData<IIC_iPop_Br,   [InstrStage<1, [A8_Pipe0]>,
164                                 InstrStage<3, [A8_LSPipe]>,
165                                 InstrStage<1, [A8_Pipe0, A8_Pipe1]>],
166                                [1, 1, 3]>,
167
168   //
169   // iLoadi + iALUr for t2LDRpci_pic.
170   InstrItinData<IIC_iLoadiALU, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
171                                 InstrStage<1, [A8_LSPipe]>,
172                                 InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [4, 1]>,
173
174
175   // Integer store pipeline
176   //
177   // Immediate offset
178   InstrItinData<IIC_iStore_i  , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
179                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
180   InstrItinData<IIC_iStore_bh_i,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
181                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
182   InstrItinData<IIC_iStore_d_i, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
183                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
184   //
185   // Register offset
186   InstrItinData<IIC_iStore_r  , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
187                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
188   InstrItinData<IIC_iStore_bh_r,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
189                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
190   InstrItinData<IIC_iStore_d_r, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
191                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
192   //
193   // Scaled register offset, issues over 2 cycles
194   InstrItinData<IIC_iStore_si , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
195                                  InstrStage<2, [A8_LSPipe]>], [3, 1, 1]>,
196   InstrItinData<IIC_iStore_bh_si,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
197                                   InstrStage<2, [A8_LSPipe]>], [3, 1, 1]>,
198   //
199   // Immediate offset with update
200   InstrItinData<IIC_iStore_iu , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
201                                  InstrStage<1, [A8_LSPipe]>], [2, 3, 1]>,
202   InstrItinData<IIC_iStore_bh_iu,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
203                                  InstrStage<1, [A8_LSPipe]>], [2, 3, 1]>,
204   //
205   // Register offset with update
206   InstrItinData<IIC_iStore_ru  , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
207                                   InstrStage<1, [A8_LSPipe]>], [2, 3, 1, 1]>,
208   InstrItinData<IIC_iStore_bh_ru,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
209                                   InstrStage<1, [A8_LSPipe]>], [2, 3, 1, 1]>,
210   InstrItinData<IIC_iStore_d_ru, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
211                                   InstrStage<1, [A8_LSPipe]>], [2, 3, 1, 1]>,
212   //
213   // Scaled register offset with update, issues over 2 cycles
214   InstrItinData<IIC_iStore_siu, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
215                                  InstrStage<2, [A8_LSPipe]>], [3, 3, 1, 1]>,
216   InstrItinData<IIC_iStore_bh_siu,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
217                                    InstrStage<2, [A8_LSPipe]>], [3, 3, 1, 1]>,
218   //
219   // Store multiple. Pipeline 0 only.
220   // FIXME: A8_LSPipe cycle time is dynamic, this assumes 3 to 4 registers.
221   InstrItinData<IIC_iStore_m , [InstrStage<1, [A8_Pipe0]>,
222                                 InstrStage<2, [A8_LSPipe]>]>,
223   //
224   // Store multiple + update
225   InstrItinData<IIC_iStore_mu, [InstrStage<1, [A8_Pipe0]>,
226                                 InstrStage<2, [A8_LSPipe]>], [2]>,
227
228   // Branch
229   //
230   // no delay slots, so the latency of a branch is unimportant
231   InstrItinData<IIC_Br      , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>]>,
232
233   // VFP
234   // Issue through integer pipeline, and execute in NEON unit. We assume
235   // RunFast mode so that NFP pipeline is used for single-precision when
236   // possible.
237   //
238   // FP Special Register to Integer Register File Move
239   InstrItinData<IIC_fpSTAT , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
240                               InstrStage<1, [A8_NLSPipe]>], [20]>,
241   //
242   // Single-precision FP Unary
243   InstrItinData<IIC_fpUNA32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
244                                InstrStage<1, [A8_NPipe]>], [7, 1]>,
245   //
246   // Double-precision FP Unary
247   InstrItinData<IIC_fpUNA64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
248                                InstrStage<4, [A8_NPipe], 0>,
249                                InstrStage<4, [A8_NLSPipe]>], [4, 1]>,
250   //
251   // Single-precision FP Compare
252   InstrItinData<IIC_fpCMP32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
253                                InstrStage<1, [A8_NPipe]>], [1, 1]>,
254   //
255   // Double-precision FP Compare
256   InstrItinData<IIC_fpCMP64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
257                                InstrStage<4, [A8_NPipe], 0>,
258                                InstrStage<4, [A8_NLSPipe]>], [4, 1]>,
259   //
260   // Single to Double FP Convert
261   InstrItinData<IIC_fpCVTSD , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
262                                InstrStage<7, [A8_NPipe], 0>,
263                                InstrStage<7, [A8_NLSPipe]>], [7, 1]>,
264   //
265   // Double to Single FP Convert
266   InstrItinData<IIC_fpCVTDS , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
267                                InstrStage<5, [A8_NPipe], 0>,
268                                InstrStage<5, [A8_NLSPipe]>], [5, 1]>,
269   //
270   // Single-Precision FP to Integer Convert
271   InstrItinData<IIC_fpCVTSI , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
272                                InstrStage<1, [A8_NPipe]>], [7, 1]>,
273   //
274   // Double-Precision FP to Integer Convert
275   InstrItinData<IIC_fpCVTDI , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
276                                InstrStage<8, [A8_NPipe], 0>,
277                                InstrStage<8, [A8_NLSPipe]>], [8, 1]>,
278   //
279   // Integer to Single-Precision FP Convert
280   InstrItinData<IIC_fpCVTIS , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
281                                InstrStage<1, [A8_NPipe]>], [7, 1]>,
282   //
283   // Integer to Double-Precision FP Convert
284   InstrItinData<IIC_fpCVTID , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
285                                InstrStage<8, [A8_NPipe], 0>,
286                                InstrStage<8, [A8_NLSPipe]>], [8, 1]>,
287   //
288   // Single-precision FP ALU
289   InstrItinData<IIC_fpALU32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
290                                InstrStage<1, [A8_NPipe]>], [7, 1, 1]>,
291   //
292   // Double-precision FP ALU
293   InstrItinData<IIC_fpALU64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
294                                InstrStage<9, [A8_NPipe], 0>,
295                                InstrStage<9, [A8_NLSPipe]>], [9, 1, 1]>,
296   //
297   // Single-precision FP Multiply
298   InstrItinData<IIC_fpMUL32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
299                                InstrStage<1, [A8_NPipe]>], [7, 1, 1]>,
300   //
301   // Double-precision FP Multiply
302   InstrItinData<IIC_fpMUL64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
303                                InstrStage<11, [A8_NPipe], 0>,
304                                InstrStage<11, [A8_NLSPipe]>], [11, 1, 1]>,
305   //
306   // Single-precision FP MAC
307   InstrItinData<IIC_fpMAC32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
308                                InstrStage<1, [A8_NPipe]>], [7, 2, 1, 1]>,
309   //
310   // Double-precision FP MAC
311   InstrItinData<IIC_fpMAC64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
312                                InstrStage<19, [A8_NPipe], 0>,
313                                InstrStage<19, [A8_NLSPipe]>], [19, 2, 1, 1]>,
314   //
315   // Single-precision FP DIV
316   InstrItinData<IIC_fpDIV32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
317                                InstrStage<20, [A8_NPipe], 0>,
318                                InstrStage<20, [A8_NLSPipe]>], [20, 1, 1]>,
319   //
320   // Double-precision FP DIV
321   InstrItinData<IIC_fpDIV64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
322                                InstrStage<29, [A8_NPipe], 0>,
323                                InstrStage<29, [A8_NLSPipe]>], [29, 1, 1]>,
324   //
325   // Single-precision FP SQRT
326   InstrItinData<IIC_fpSQRT32, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
327                                InstrStage<19, [A8_NPipe], 0>,
328                                InstrStage<19, [A8_NLSPipe]>], [19, 1]>,
329   //
330   // Double-precision FP SQRT
331   InstrItinData<IIC_fpSQRT64, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
332                                InstrStage<29, [A8_NPipe], 0>,
333                                InstrStage<29, [A8_NLSPipe]>], [29, 1]>,
334
335   //
336   // Integer to Single-precision Move
337   InstrItinData<IIC_fpMOVIS,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
338                                InstrStage<1, [A8_NPipe]>],
339                               [2, 1]>,
340   //
341   // Integer to Double-precision Move
342   InstrItinData<IIC_fpMOVID,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
343                                InstrStage<1, [A8_NPipe]>],
344                               [2, 1, 1]>,
345   //
346   // Single-precision to Integer Move
347   InstrItinData<IIC_fpMOVSI,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
348                                InstrStage<1, [A8_NPipe]>],
349                               [20, 1]>,
350   //
351   // Double-precision to Integer Move
352   InstrItinData<IIC_fpMOVDI,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
353                                InstrStage<1, [A8_NPipe]>],
354                               [20, 20, 1]>,
355
356   //
357   // Single-precision FP Load
358   InstrItinData<IIC_fpLoad32, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
359                                InstrStage<1, [A8_NLSPipe]>,
360                                InstrStage<1, [A8_LSPipe]>],
361                               [2, 1]>,
362   //
363   // Double-precision FP Load
364   InstrItinData<IIC_fpLoad64, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
365                                InstrStage<1, [A8_NLSPipe]>,
366                                InstrStage<1, [A8_LSPipe]>],
367                               [2, 1]>,
368   //
369   // FP Load Multiple
370   // FIXME: A8_LSPipe cycle time is dynamic, this assumes 3 to 4 registers.
371   InstrItinData<IIC_fpLoad_m, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
372                                InstrStage<1, [A8_NLSPipe]>,
373                                InstrStage<1, [A8_LSPipe]>,
374                                InstrStage<1, [A8_NLSPipe]>,
375                                InstrStage<1, [A8_LSPipe]>], [1, 1, 1, 2]>,
376   //
377   // FP Load Multiple + update
378   InstrItinData<IIC_fpLoad_mu,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
379                                InstrStage<1, [A8_NLSPipe]>,
380                                InstrStage<1, [A8_LSPipe]>,
381                                InstrStage<1, [A8_NLSPipe]>,
382                                InstrStage<1, [A8_LSPipe]>], [2, 1, 1, 1, 2]>,
383   //
384   // Single-precision FP Store
385   InstrItinData<IIC_fpStore32,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
386                                InstrStage<1, [A8_NLSPipe]>,
387                                InstrStage<1, [A8_LSPipe]>],
388                               [1, 1]>,
389   //
390   // Double-precision FP Store
391   InstrItinData<IIC_fpStore64,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
392                                InstrStage<1, [A8_NLSPipe]>,
393                                InstrStage<1, [A8_LSPipe]>],
394                               [1, 1]>,
395   //
396   // FP Store Multiple
397   InstrItinData<IIC_fpStore_m,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
398                                InstrStage<1, [A8_NLSPipe]>,
399                                InstrStage<1, [A8_LSPipe]>,
400                                InstrStage<1, [A8_NLSPipe]>,
401                                InstrStage<1, [A8_LSPipe]>], [1, 1, 1, 1]>,
402   //
403   // FP Store Multiple + update
404   InstrItinData<IIC_fpStore_mu,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
405                                 InstrStage<1, [A8_NLSPipe]>,
406                                 InstrStage<1, [A8_LSPipe]>,
407                                 InstrStage<1, [A8_NLSPipe]>,
408                                 InstrStage<1, [A8_LSPipe]>], [2, 1, 1, 1, 1]>,
409
410   // NEON
411   // Issue through integer pipeline, and execute in NEON unit.
412   //
413   // VLD1
414   InstrItinData<IIC_VLD1,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
415                                InstrStage<2, [A8_NLSPipe], 1>,
416                                InstrStage<2, [A8_LSPipe]>],
417                               [2, 1]>,
418   // VLD1x2
419   InstrItinData<IIC_VLD1x2,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
420                                InstrStage<2, [A8_NLSPipe], 1>,
421                                InstrStage<2, [A8_LSPipe]>],
422                               [2, 2, 1]>,
423   //
424   // VLD1x3
425   InstrItinData<IIC_VLD1x3,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
426                                InstrStage<3, [A8_NLSPipe], 1>,
427                                InstrStage<3, [A8_LSPipe]>],
428                               [2, 2, 3, 1]>,
429   //
430   // VLD1x4
431   InstrItinData<IIC_VLD1x4,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
432                                InstrStage<3, [A8_NLSPipe], 1>,
433                                InstrStage<3, [A8_LSPipe]>],
434                               [2, 2, 3, 3, 1]>,
435   //
436   // VLD1u
437   InstrItinData<IIC_VLD1u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
438                                InstrStage<2, [A8_NLSPipe], 1>,
439                                InstrStage<2, [A8_LSPipe]>],
440                               [2, 2, 1]>,
441   //
442   // VLD1x2u
443   InstrItinData<IIC_VLD1x2u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
444                                InstrStage<2, [A8_NLSPipe], 1>,
445                                InstrStage<2, [A8_LSPipe]>],
446                               [2, 2, 2, 1]>,
447   //
448   // VLD1x3u
449   InstrItinData<IIC_VLD1x3u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
450                                InstrStage<3, [A8_NLSPipe], 1>,
451                                InstrStage<3, [A8_LSPipe]>],
452                               [2, 2, 3, 2, 1]>,
453   //
454   // VLD1x4u
455   InstrItinData<IIC_VLD1x4u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
456                                InstrStage<3, [A8_NLSPipe], 1>,
457                                InstrStage<3, [A8_LSPipe]>],
458                               [2, 2, 3, 3, 2, 1]>,
459   //
460   // VLD1ln
461   InstrItinData<IIC_VLD1ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
462                                InstrStage<3, [A8_NLSPipe], 1>,
463                                InstrStage<3, [A8_LSPipe]>],
464                               [3, 1, 1, 1]>,
465   //
466   // VLD1lnu
467   InstrItinData<IIC_VLD1lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
468                                InstrStage<3, [A8_NLSPipe], 1>,
469                                InstrStage<3, [A8_LSPipe]>],
470                               [3, 2, 1, 1, 1, 1]>,
471   //
472   // VLD2
473   InstrItinData<IIC_VLD2,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
474                                InstrStage<2, [A8_NLSPipe], 1>,
475                                InstrStage<2, [A8_LSPipe]>],
476                               [2, 2, 1]>,
477   //
478   // VLD2x2
479   InstrItinData<IIC_VLD2x2,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
480                                InstrStage<3, [A8_NLSPipe], 1>,
481                                InstrStage<3, [A8_LSPipe]>],
482                               [2, 2, 3, 3, 1]>,
483   //
484   // VLD2ln
485   InstrItinData<IIC_VLD2ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
486                                InstrStage<3, [A8_NLSPipe], 1>,
487                                InstrStage<3, [A8_LSPipe]>],
488                               [3, 3, 1, 1, 1, 1]>,
489   //
490   // VLD2u
491   InstrItinData<IIC_VLD2u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
492                                InstrStage<2, [A8_NLSPipe], 1>,
493                                InstrStage<2, [A8_LSPipe]>],
494                               [2, 2, 2, 1, 1, 1]>,
495   //
496   // VLD2x2u
497   InstrItinData<IIC_VLD2x2u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
498                                InstrStage<3, [A8_NLSPipe], 1>,
499                                InstrStage<3, [A8_LSPipe]>],
500                               [2, 2, 3, 3, 2, 1]>,
501   //
502   // VLD2lnu
503   InstrItinData<IIC_VLD2lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
504                                InstrStage<3, [A8_NLSPipe], 1>,
505                                InstrStage<3, [A8_LSPipe]>],
506                               [3, 3, 2, 1, 1, 1, 1, 1]>,
507   //
508   // VLD3
509   InstrItinData<IIC_VLD3,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
510                                InstrStage<4, [A8_NLSPipe], 1>,
511                                InstrStage<4, [A8_LSPipe]>],
512                               [3, 3, 4, 1]>,
513   //
514   // VLD3ln
515   InstrItinData<IIC_VLD3ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
516                                InstrStage<5, [A8_NLSPipe], 1>,
517                                InstrStage<5, [A8_LSPipe]>],
518                               [4, 4, 5, 1, 1, 1, 1, 2]>,
519   //
520   // VLD3u
521   InstrItinData<IIC_VLD3u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
522                                InstrStage<4, [A8_NLSPipe], 1>,
523                                InstrStage<4, [A8_LSPipe]>],
524                               [3, 3, 4, 2, 1]>,
525   //
526   // VLD3lnu
527   InstrItinData<IIC_VLD3lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
528                                InstrStage<5, [A8_NLSPipe], 1>,
529                                InstrStage<5, [A8_LSPipe]>],
530                               [4, 4, 5, 2, 1, 1, 1, 1, 1, 2]>,
531   //
532   // VLD4
533   InstrItinData<IIC_VLD4,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
534                                InstrStage<4, [A8_NLSPipe], 1>,
535                                InstrStage<4, [A8_LSPipe]>],
536                               [3, 3, 4, 4, 1]>,
537   //
538   // VLD4ln
539   InstrItinData<IIC_VLD4ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
540                                InstrStage<5, [A8_NLSPipe], 1>,
541                                InstrStage<5, [A8_LSPipe]>],
542                               [4, 4, 5, 5, 1, 1, 1, 1, 2, 2]>,
543   //
544   // VLD4u
545   InstrItinData<IIC_VLD4u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
546                                InstrStage<4, [A8_NLSPipe], 1>,
547                                InstrStage<4, [A8_LSPipe]>],
548                               [3, 3, 4, 4, 2, 1]>,
549   //
550   // VLD4lnu
551   InstrItinData<IIC_VLD4lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
552                                InstrStage<5, [A8_NLSPipe], 1>,
553                                InstrStage<5, [A8_LSPipe]>],
554                               [4, 4, 5, 5, 2, 1, 1, 1, 1, 1, 2, 2]>,
555   //
556   // VST1
557   InstrItinData<IIC_VST1,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
558                                InstrStage<2, [A8_NLSPipe], 1>,
559                                InstrStage<2, [A8_LSPipe]>],
560                               [1, 1, 1]>,
561   //
562   // VST1x2
563   InstrItinData<IIC_VST1x2,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
564                                InstrStage<2, [A8_NLSPipe], 1>,
565                                InstrStage<2, [A8_LSPipe]>],
566                               [1, 1, 1, 1]>,
567   //
568   // VST1x3
569   InstrItinData<IIC_VST1x3,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
570                                InstrStage<3, [A8_NLSPipe], 1>,
571                                InstrStage<3, [A8_LSPipe]>],
572                               [1, 1, 1, 1, 2]>,
573   //
574   // VST1x4
575   InstrItinData<IIC_VST1x4,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
576                                InstrStage<3, [A8_NLSPipe], 1>,
577                                InstrStage<3, [A8_LSPipe]>],
578                               [1, 1, 1, 1, 2, 2]>,
579   //
580   // VST1u
581   InstrItinData<IIC_VST1u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
582                                InstrStage<2, [A8_NLSPipe], 1>,
583                                InstrStage<2, [A8_LSPipe]>],
584                               [2, 1, 1, 1, 1]>,
585   //
586   // VST1x2u
587   InstrItinData<IIC_VST1x2u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
588                                InstrStage<2, [A8_NLSPipe], 1>,
589                                InstrStage<2, [A8_LSPipe]>],
590                               [2, 1, 1, 1, 1, 1]>,
591   //
592   // VST1x3u
593   InstrItinData<IIC_VST1x3u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
594                                InstrStage<3, [A8_NLSPipe], 1>,
595                                InstrStage<3, [A8_LSPipe]>],
596                               [2, 1, 1, 1, 1, 1, 2]>,
597   //
598   // VST1x4u
599   InstrItinData<IIC_VST1x4u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
600                                InstrStage<3, [A8_NLSPipe], 1>,
601                                InstrStage<3, [A8_LSPipe]>],
602                               [2, 1, 1, 1, 1, 1, 2, 2]>,
603   //
604   // VST2
605   InstrItinData<IIC_VST2,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
606                                InstrStage<2, [A8_NLSPipe], 1>,
607                                InstrStage<2, [A8_LSPipe]>],
608                               [1, 1, 1, 1]>,
609   //
610   // VST2x2
611   InstrItinData<IIC_VST2x2,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
612                                InstrStage<4, [A8_NLSPipe], 1>,
613                                InstrStage<4, [A8_LSPipe]>],
614                               [1, 1, 1, 1, 2, 2]>,
615   //
616   // VST2u
617   InstrItinData<IIC_VST2u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
618                                InstrStage<2, [A8_NLSPipe], 1>,
619                                InstrStage<2, [A8_LSPipe]>],
620                               [2, 1, 1, 1, 1, 1]>,
621   //
622   // VST2x2u
623   InstrItinData<IIC_VST2x2u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
624                                InstrStage<4, [A8_NLSPipe], 1>,
625                                InstrStage<4, [A8_LSPipe]>],
626                               [2, 1, 1, 1, 1, 1, 2, 2]>,
627   //
628   // VST2ln
629   InstrItinData<IIC_VST2ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
630                                InstrStage<2, [A8_NLSPipe], 1>,
631                                InstrStage<2, [A8_LSPipe]>],
632                               [1, 1, 1, 1]>,
633   //
634   // VST2lnu
635   InstrItinData<IIC_VST2lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
636                                InstrStage<2, [A8_NLSPipe], 1>,
637                                InstrStage<2, [A8_LSPipe]>],
638                               [2, 1, 1, 1, 1, 1]>,
639   //
640   // VST3
641   InstrItinData<IIC_VST3,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
642                                InstrStage<3, [A8_NLSPipe], 1>,
643                                InstrStage<3, [A8_LSPipe]>],
644                               [1, 1, 1, 1, 2]>,
645   //
646   // VST3u
647   InstrItinData<IIC_VST3u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
648                                InstrStage<3, [A8_NLSPipe], 1>,
649                                InstrStage<3, [A8_LSPipe]>],
650                               [2, 1, 1, 1, 1, 1, 2]>,
651   //
652   // VST3ln
653   InstrItinData<IIC_VST3ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
654                                InstrStage<3, [A8_NLSPipe], 1>,
655                                InstrStage<3, [A8_LSPipe]>],
656                               [1, 1, 1, 1, 2]>,
657   //
658   // VST3lnu
659   InstrItinData<IIC_VST3lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
660                                InstrStage<3, [A8_NLSPipe], 1>,
661                                InstrStage<3, [A8_LSPipe]>],
662                               [2, 1, 1, 1, 1, 1, 2]>,
663   //
664   // VST4
665   InstrItinData<IIC_VST4,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
666                                InstrStage<4, [A8_NLSPipe], 1>,
667                                InstrStage<4, [A8_LSPipe]>],
668                               [1, 1, 1, 1, 2, 2]>,
669   //
670   // VST4u
671   InstrItinData<IIC_VST4u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
672                                InstrStage<4, [A8_NLSPipe], 1>,
673                                InstrStage<4, [A8_LSPipe]>],
674                               [2, 1, 1, 1, 1, 1, 2, 2]>,
675   //
676   // VST4ln
677   InstrItinData<IIC_VST4ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
678                                InstrStage<4, [A8_NLSPipe], 1>,
679                                InstrStage<4, [A8_LSPipe]>],
680                               [1, 1, 1, 1, 2, 2]>,
681   //
682   // VST4lnu
683   InstrItinData<IIC_VST4lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
684                                InstrStage<4, [A8_NLSPipe], 1>,
685                                InstrStage<4, [A8_LSPipe]>],
686                               [2, 1, 1, 1, 1, 1, 2, 2]>,
687   //
688   // Double-register FP Unary
689   InstrItinData<IIC_VUNAD,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
690                                InstrStage<1, [A8_NPipe]>], [5, 2]>,
691   //
692   // Quad-register FP Unary
693   // Result written in N5, but that is relative to the last cycle of multicycle,
694   // so we use 6 for those cases
695   InstrItinData<IIC_VUNAQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
696                                InstrStage<2, [A8_NPipe]>], [6, 2]>,
697   //
698   // Double-register FP Binary
699   InstrItinData<IIC_VBIND,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
700                                InstrStage<1, [A8_NPipe]>], [5, 2, 2]>,
701   //
702   // VPADD, etc.
703   InstrItinData<IIC_VPBIND,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
704                                InstrStage<1, [A8_NPipe]>], [5, 2, 2]>,
705   //
706   // Double-register FP VMUL
707   InstrItinData<IIC_VFMULD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
708                                InstrStage<1, [A8_NPipe]>], [5, 2, 1]>,
709
710   //
711   // Quad-register FP Binary
712   // Result written in N5, but that is relative to the last cycle of multicycle,
713   // so we use 6 for those cases
714   InstrItinData<IIC_VBINQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
715                                InstrStage<2, [A8_NPipe]>], [6, 2, 2]>,
716   //
717   // Quad-register FP VMUL
718   InstrItinData<IIC_VFMULQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
719                                InstrStage<1, [A8_NPipe]>], [6, 2, 1]>,
720   //
721   // Move
722   InstrItinData<IIC_VMOV,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
723                                InstrStage<1, [A8_NPipe]>], [1, 1]>,
724   //
725   // Move Immediate
726   InstrItinData<IIC_VMOVImm,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
727                                InstrStage<1, [A8_NPipe]>], [3]>,
728   //
729   // Double-register Permute Move
730   InstrItinData<IIC_VMOVD,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
731                                InstrStage<1, [A8_NLSPipe]>], [2, 1]>,
732   //
733   // Quad-register Permute Move
734   // Result written in N2, but that is relative to the last cycle of multicycle,
735   // so we use 3 for those cases
736   InstrItinData<IIC_VMOVQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
737                                InstrStage<2, [A8_NLSPipe]>], [3, 1]>,
738   //
739   // Integer to Single-precision Move
740   InstrItinData<IIC_VMOVIS ,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
741                                InstrStage<1, [A8_NLSPipe]>], [2, 1]>,
742   //
743   // Integer to Double-precision Move
744   InstrItinData<IIC_VMOVID ,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
745                                InstrStage<1, [A8_NLSPipe]>], [2, 1, 1]>,
746   //
747   // Single-precision to Integer Move
748   InstrItinData<IIC_VMOVSI ,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
749                                InstrStage<1, [A8_NLSPipe]>], [20, 1]>,
750   //
751   // Double-precision to Integer Move
752   InstrItinData<IIC_VMOVDI ,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
753                                InstrStage<1, [A8_NLSPipe]>], [20, 20, 1]>,
754   //
755   // Integer to Lane Move
756   InstrItinData<IIC_VMOVISL , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
757                                InstrStage<2, [A8_NLSPipe]>], [3, 1, 1]>,
758   //
759   // Vector narrow move
760   InstrItinData<IIC_VMOVN   , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
761                                InstrStage<1, [A8_NPipe]>], [2, 1]>,
762   //
763   // Double-register Permute
764   InstrItinData<IIC_VPERMD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
765                                InstrStage<1, [A8_NLSPipe]>], [2, 2, 1, 1]>,
766   //
767   // Quad-register Permute
768   // Result written in N2, but that is relative to the last cycle of multicycle,
769   // so we use 3 for those cases
770   InstrItinData<IIC_VPERMQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
771                                InstrStage<2, [A8_NLSPipe]>], [3, 3, 1, 1]>,
772   //
773   // Quad-register Permute (3 cycle issue)
774   // Result written in N2, but that is relative to the last cycle of multicycle,
775   // so we use 4 for those cases
776   InstrItinData<IIC_VPERMQ3,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
777                                InstrStage<1, [A8_NLSPipe]>,
778                                InstrStage<1, [A8_NPipe], 0>,
779                                InstrStage<2, [A8_NLSPipe]>], [4, 4, 1, 1]>,
780   //
781   // Double-register FP Multiple-Accumulate
782   InstrItinData<IIC_VMACD,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
783                                InstrStage<1, [A8_NPipe]>], [9, 3, 2, 2]>,
784   //
785   // Quad-register FP Multiple-Accumulate
786   // Result written in N9, but that is relative to the last cycle of multicycle,
787   // so we use 10 for those cases
788   InstrItinData<IIC_VMACQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
789                                InstrStage<2, [A8_NPipe]>], [10, 3, 2, 2]>,
790   //
791   // Double-register Reciprical Step
792   InstrItinData<IIC_VRECSD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
793                                InstrStage<1, [A8_NPipe]>], [9, 2, 2]>,
794   //
795   // Quad-register Reciprical Step
796   InstrItinData<IIC_VRECSQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
797                                InstrStage<2, [A8_NPipe]>], [10, 2, 2]>,
798   //
799   // Double-register Integer Count
800   InstrItinData<IIC_VCNTiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
801                                InstrStage<1, [A8_NPipe]>], [3, 2, 2]>,
802   //
803   // Quad-register Integer Count
804   // Result written in N3, but that is relative to the last cycle of multicycle,
805   // so we use 4 for those cases
806   InstrItinData<IIC_VCNTiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
807                                InstrStage<2, [A8_NPipe]>], [4, 2, 2]>,
808   //
809   // Double-register Integer Unary
810   InstrItinData<IIC_VUNAiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
811                                InstrStage<1, [A8_NPipe]>], [4, 2]>,
812   //
813   // Quad-register Integer Unary
814   InstrItinData<IIC_VUNAiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
815                                InstrStage<1, [A8_NPipe]>], [4, 2]>,
816   //
817   // Double-register Integer Q-Unary
818   InstrItinData<IIC_VQUNAiD,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
819                                InstrStage<1, [A8_NPipe]>], [4, 1]>,
820   //
821   // Quad-register Integer CountQ-Unary
822   InstrItinData<IIC_VQUNAiQ,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
823                                InstrStage<1, [A8_NPipe]>], [4, 1]>,
824   //
825   // Double-register Integer Binary
826   InstrItinData<IIC_VBINiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
827                                InstrStage<1, [A8_NPipe]>], [3, 2, 2]>,
828   //
829   // Quad-register Integer Binary
830   InstrItinData<IIC_VBINiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
831                                InstrStage<1, [A8_NPipe]>], [3, 2, 2]>,
832   //
833   // Double-register Integer Binary (4 cycle)
834   InstrItinData<IIC_VBINi4D,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
835                                InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
836   //
837   // Quad-register Integer Binary (4 cycle)
838   InstrItinData<IIC_VBINi4Q,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
839                                InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
840
841   //
842   // Double-register Integer Subtract
843   InstrItinData<IIC_VSUBiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
844                                InstrStage<1, [A8_NPipe]>], [3, 2, 1]>,
845   //
846   // Quad-register Integer Subtract
847   InstrItinData<IIC_VSUBiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
848                                InstrStage<1, [A8_NPipe]>], [3, 2, 1]>,
849   //
850   // Double-register Integer Subtract
851   InstrItinData<IIC_VSUBi4D,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
852                                InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
853   //
854   // Quad-register Integer Subtract
855   InstrItinData<IIC_VSUBi4Q,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
856                                InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
857   //
858   // Double-register Integer Shift
859   InstrItinData<IIC_VSHLiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
860                                InstrStage<1, [A8_NPipe]>], [3, 1, 1]>,
861   //
862   // Quad-register Integer Shift
863   InstrItinData<IIC_VSHLiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
864                                InstrStage<2, [A8_NPipe]>], [4, 1, 1]>,
865   //
866   // Double-register Integer Shift (4 cycle)
867   InstrItinData<IIC_VSHLi4D,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
868                                InstrStage<1, [A8_NPipe]>], [4, 1, 1]>,
869   //
870   // Quad-register Integer Shift (4 cycle)
871   InstrItinData<IIC_VSHLi4Q,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
872                                InstrStage<2, [A8_NPipe]>], [5, 1, 1]>,
873   //
874   // Double-register Integer Pair Add Long
875   InstrItinData<IIC_VPALiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
876                                InstrStage<1, [A8_NPipe]>], [6, 3, 1]>,
877   //
878   // Quad-register Integer Pair Add Long
879   InstrItinData<IIC_VPALiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
880                                InstrStage<2, [A8_NPipe]>], [7, 3, 1]>,
881   //
882   // Double-register Absolute Difference and Accumulate
883   InstrItinData<IIC_VABAD,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
884                                InstrStage<1, [A8_NPipe]>], [6, 3, 2, 1]>,
885   //
886   // Quad-register Absolute Difference and Accumulate
887   InstrItinData<IIC_VABAQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
888                                InstrStage<2, [A8_NPipe]>], [6, 3, 2, 1]>,
889
890   //
891   // Double-register Integer Multiply (.8, .16)
892   InstrItinData<IIC_VMULi16D, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
893                                InstrStage<1, [A8_NPipe]>], [6, 2, 2]>,
894   //
895   // Double-register Integer Multiply (.32)
896   InstrItinData<IIC_VMULi32D, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
897                                InstrStage<2, [A8_NPipe]>], [7, 2, 1]>,
898   //
899   // Quad-register Integer Multiply (.8, .16)
900   InstrItinData<IIC_VMULi16Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
901                                InstrStage<2, [A8_NPipe]>], [7, 2, 2]>,
902   //
903   // Quad-register Integer Multiply (.32)
904   InstrItinData<IIC_VMULi32Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
905                                InstrStage<1, [A8_NPipe]>,
906                                InstrStage<2, [A8_NLSPipe], 0>,
907                                InstrStage<3, [A8_NPipe]>], [9, 2, 1]>,
908   //
909   // Double-register Integer Multiply-Accumulate (.8, .16)
910   InstrItinData<IIC_VMACi16D, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
911                                InstrStage<1, [A8_NPipe]>], [6, 3, 2, 2]>,
912   //
913   // Double-register Integer Multiply-Accumulate (.32)
914   InstrItinData<IIC_VMACi32D, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
915                                InstrStage<2, [A8_NPipe]>], [7, 3, 2, 1]>,
916   //
917   // Quad-register Integer Multiply-Accumulate (.8, .16)
918   InstrItinData<IIC_VMACi16Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
919                                InstrStage<2, [A8_NPipe]>], [7, 3, 2, 2]>,
920   //
921   // Quad-register Integer Multiply-Accumulate (.32)
922   InstrItinData<IIC_VMACi32Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
923                                InstrStage<1, [A8_NPipe]>,
924                                InstrStage<2, [A8_NLSPipe], 0>,
925                                InstrStage<3, [A8_NPipe]>], [9, 3, 2, 1]>,
926   //
927   // Double-register VEXT
928   InstrItinData<IIC_VEXTD,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
929                                InstrStage<1, [A8_NLSPipe]>], [2, 1, 1]>,
930   //
931   // Quad-register VEXT
932   InstrItinData<IIC_VEXTQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
933                                InstrStage<2, [A8_NLSPipe]>], [3, 1, 1]>,
934   //
935   // VTB
936   InstrItinData<IIC_VTB1,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
937                                InstrStage<2, [A8_NLSPipe]>], [3, 2, 1]>,
938   InstrItinData<IIC_VTB2,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
939                                InstrStage<2, [A8_NLSPipe]>], [3, 2, 2, 1]>,
940   InstrItinData<IIC_VTB3,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
941                                InstrStage<1, [A8_NLSPipe]>,
942                                InstrStage<1, [A8_NPipe], 0>,
943                                InstrStage<2, [A8_NLSPipe]>], [4, 2, 2, 3, 1]>,
944   InstrItinData<IIC_VTB4,     [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
945                                InstrStage<1, [A8_NLSPipe]>,
946                                InstrStage<1, [A8_NPipe], 0>,
947                                InstrStage<2, [A8_NLSPipe]>],[4, 2, 2, 3, 3, 1]>,
948   //
949   // VTBX
950   InstrItinData<IIC_VTBX1,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
951                                InstrStage<2, [A8_NLSPipe]>], [3, 1, 2, 1]>,
952   InstrItinData<IIC_VTBX2,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
953                                InstrStage<2, [A8_NLSPipe]>], [3, 1, 2, 2, 1]>,
954   InstrItinData<IIC_VTBX3,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
955                                InstrStage<1, [A8_NLSPipe]>,
956                                InstrStage<1, [A8_NPipe], 0>,
957                                InstrStage<2, [A8_NLSPipe]>],[4, 1, 2, 2, 3, 1]>,
958   InstrItinData<IIC_VTBX4,    [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
959                                InstrStage<1, [A8_NLSPipe]>,
960                                InstrStage<1, [A8_NPipe], 0>,
961                             InstrStage<2, [A8_NLSPipe]>], [4, 1, 2, 2, 3, 3, 1]>
962 ]>;