Add NEON VLD1-dup instructions (load 1 element to all lanes).
[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   InstrItinData<IIC_iCMOVix2,[InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
81                               InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [3, 1]>,
82   //
83   // MVN instructions
84   InstrItinData<IIC_iMVNi , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1]>,
85   InstrItinData<IIC_iMVNr , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1, 1]>,
86   InstrItinData<IIC_iMVNsi, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1, 1]>,
87   InstrItinData<IIC_iMVNsr, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [1, 1, 1]>,
88
89   // Integer multiply pipeline
90   // Result written in E5, but that is relative to the last cycle of multicycle,
91   // so we use 6 for those cases
92   //
93   InstrItinData<IIC_iMUL16   , [InstrStage<1, [A8_Pipe0]>], [5, 1, 1]>,
94   InstrItinData<IIC_iMAC16   , [InstrStage<2, [A8_Pipe0]>], [6, 1, 1, 4]>,
95   InstrItinData<IIC_iMUL32   , [InstrStage<2, [A8_Pipe0]>], [6, 1, 1]>,
96   InstrItinData<IIC_iMAC32   , [InstrStage<2, [A8_Pipe0]>], [6, 1, 1, 4]>,
97   InstrItinData<IIC_iMUL64   , [InstrStage<3, [A8_Pipe0]>], [6, 6, 1, 1]>,
98   InstrItinData<IIC_iMAC64   , [InstrStage<3, [A8_Pipe0]>], [6, 6, 1, 1]>,
99
100   // Integer load pipeline
101   //
102   // Immediate offset
103   InstrItinData<IIC_iLoad_i   , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
104                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
105   InstrItinData<IIC_iLoad_bh_i, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
106                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
107   InstrItinData<IIC_iLoad_d_i,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
108                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
109   //
110   // Register offset
111   InstrItinData<IIC_iLoad_r   , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
112                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
113   InstrItinData<IIC_iLoad_bh_r, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
114                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
115   InstrItinData<IIC_iLoad_d_r , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
116                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
117   //
118   // Scaled register offset, issues over 2 cycles
119   // FIXME: lsl by 2 takes 1 cycle.
120   InstrItinData<IIC_iLoad_si  , [InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
121                                  InstrStage<1, [A8_LSPipe]>], [4, 1, 1]>,
122   InstrItinData<IIC_iLoad_bh_si,[InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
123                                  InstrStage<1, [A8_LSPipe]>], [4, 1, 1]>,
124   //
125   // Immediate offset with update
126   InstrItinData<IIC_iLoad_iu  , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
127                                  InstrStage<1, [A8_LSPipe]>], [3, 2, 1]>,
128   InstrItinData<IIC_iLoad_bh_iu,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
129                                  InstrStage<1, [A8_LSPipe]>], [3, 2, 1]>,
130   //
131   // Register offset with update
132   InstrItinData<IIC_iLoad_ru  , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
133                                  InstrStage<1, [A8_LSPipe]>], [3, 2, 1, 1]>,
134   InstrItinData<IIC_iLoad_bh_ru,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
135                                  InstrStage<1, [A8_LSPipe]>], [3, 2, 1, 1]>,
136   InstrItinData<IIC_iLoad_d_ru, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
137                                  InstrStage<1, [A8_LSPipe]>], [3, 2, 1, 1]>,
138   //
139   // Scaled register offset with update, issues over 2 cycles
140   InstrItinData<IIC_iLoad_siu , [InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
141                                  InstrStage<2, [A8_LSPipe]>], [4, 3, 1, 1]>,
142   InstrItinData<IIC_iLoad_bh_siu,[InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
143                                   InstrStage<2, [A8_LSPipe]>], [4, 3, 1, 1]>,
144   //
145   // Load multiple, def is the 5th operand. Pipeline 0 only.
146   // FIXME: A8_LSPipe cycle time is dynamic, this assumes 3 to 4 registers.
147   InstrItinData<IIC_iLoad_m  , [InstrStage<2, [A8_Pipe0], 0>,
148                                 InstrStage<2, [A8_LSPipe]>], [1, 1, 1, 1, 3]>,
149   //
150   // Load multiple + update, defs are the 1st and 5th operands.
151   InstrItinData<IIC_iLoad_mu , [InstrStage<3, [A8_Pipe0], 0>,
152                                 InstrStage<3, [A8_LSPipe]>], [2, 1, 1, 1, 3]>,
153   //
154   // Load multiple plus branch
155   InstrItinData<IIC_iLoad_mBr, [InstrStage<3, [A8_Pipe0], 0>,
156                                 InstrStage<3, [A8_LSPipe]>,
157                                 InstrStage<1, [A8_Pipe0, A8_Pipe1]>],
158                                [1, 2, 1, 1, 3]>,
159   //
160   // Pop, def is the 3rd operand.
161   InstrItinData<IIC_iPop  ,    [InstrStage<3, [A8_Pipe0], 0>,
162                                 InstrStage<3, [A8_LSPipe]>], [1, 1, 3]>,
163   //
164   // Push, def is the 3th operand.
165   InstrItinData<IIC_iPop_Br,   [InstrStage<3, [A8_Pipe0], 0>,
166                                 InstrStage<3, [A8_LSPipe]>,
167                                 InstrStage<1, [A8_Pipe0, A8_Pipe1]>],
168                                [1, 1, 3]>,
169
170   //
171   // iLoadi + iALUr for t2LDRpci_pic.
172   InstrItinData<IIC_iLoadiALU, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
173                                 InstrStage<1, [A8_LSPipe]>,
174                                 InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [4, 1]>,
175
176
177   // Integer store pipeline
178   //
179   // Immediate offset
180   InstrItinData<IIC_iStore_i  , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
181                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
182   InstrItinData<IIC_iStore_bh_i,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
183                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
184   InstrItinData<IIC_iStore_d_i, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
185                                  InstrStage<1, [A8_LSPipe]>], [3, 1]>,
186   //
187   // Register offset
188   InstrItinData<IIC_iStore_r  , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
189                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
190   InstrItinData<IIC_iStore_bh_r,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
191                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
192   InstrItinData<IIC_iStore_d_r, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
193                                  InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
194   //
195   // Scaled register offset, issues over 2 cycles
196   InstrItinData<IIC_iStore_si , [InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
197                                  InstrStage<2, [A8_LSPipe]>], [3, 1, 1]>,
198   InstrItinData<IIC_iStore_bh_si,[InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
199                                   InstrStage<2, [A8_LSPipe]>], [3, 1, 1]>,
200   //
201   // Immediate offset with update
202   InstrItinData<IIC_iStore_iu , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
203                                  InstrStage<1, [A8_LSPipe]>], [2, 3, 1]>,
204   InstrItinData<IIC_iStore_bh_iu,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
205                                  InstrStage<1, [A8_LSPipe]>], [2, 3, 1]>,
206   //
207   // Register offset with update
208   InstrItinData<IIC_iStore_ru  , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
209                                   InstrStage<1, [A8_LSPipe]>], [2, 3, 1, 1]>,
210   InstrItinData<IIC_iStore_bh_ru,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
211                                   InstrStage<1, [A8_LSPipe]>], [2, 3, 1, 1]>,
212   InstrItinData<IIC_iStore_d_ru, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
213                                   InstrStage<1, [A8_LSPipe]>], [2, 3, 1, 1]>,
214   //
215   // Scaled register offset with update, issues over 2 cycles
216   InstrItinData<IIC_iStore_siu, [InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
217                                  InstrStage<2, [A8_LSPipe]>], [3, 3, 1, 1]>,
218   InstrItinData<IIC_iStore_bh_siu,[InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
219                                    InstrStage<2, [A8_LSPipe]>], [3, 3, 1, 1]>,
220   //
221   // Store multiple. Pipeline 0 only.
222   // FIXME: A8_LSPipe cycle time is dynamic, this assumes 3 to 4 registers.
223   InstrItinData<IIC_iStore_m , [InstrStage<2, [A8_Pipe0], 0>,
224                                 InstrStage<2, [A8_LSPipe]>]>,
225   //
226   // Store multiple + update
227   InstrItinData<IIC_iStore_mu, [InstrStage<2, [A8_Pipe0], 0>,
228                                 InstrStage<2, [A8_LSPipe]>], [2]>,
229
230   //
231   // Preload
232   InstrItinData<IIC_Preload, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2]>,
233
234   // Branch
235   //
236   // no delay slots, so the latency of a branch is unimportant
237   InstrItinData<IIC_Br      , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>]>,
238
239   // VFP
240   // Issue through integer pipeline, and execute in NEON unit. We assume
241   // RunFast mode so that NFP pipeline is used for single-precision when
242   // possible.
243   //
244   // FP Special Register to Integer Register File Move
245   InstrItinData<IIC_fpSTAT , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
246                               InstrStage<1, [A8_NLSPipe]>], [20]>,
247   //
248   // Single-precision FP Unary
249   InstrItinData<IIC_fpUNA32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
250                                InstrStage<1, [A8_NPipe]>], [7, 1]>,
251   //
252   // Double-precision FP Unary
253   InstrItinData<IIC_fpUNA64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
254                                InstrStage<4, [A8_NPipe], 0>,
255                                InstrStage<4, [A8_NLSPipe]>], [4, 1]>,
256   //
257   // Single-precision FP Compare
258   InstrItinData<IIC_fpCMP32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
259                                InstrStage<1, [A8_NPipe]>], [1, 1]>,
260   //
261   // Double-precision FP Compare
262   InstrItinData<IIC_fpCMP64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
263                                InstrStage<4, [A8_NPipe], 0>,
264                                InstrStage<4, [A8_NLSPipe]>], [4, 1]>,
265   //
266   // Single to Double FP Convert
267   InstrItinData<IIC_fpCVTSD , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
268                                InstrStage<7, [A8_NPipe], 0>,
269                                InstrStage<7, [A8_NLSPipe]>], [7, 1]>,
270   //
271   // Double to Single FP Convert
272   InstrItinData<IIC_fpCVTDS , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
273                                InstrStage<5, [A8_NPipe], 0>,
274                                InstrStage<5, [A8_NLSPipe]>], [5, 1]>,
275   //
276   // Single-Precision FP to Integer Convert
277   InstrItinData<IIC_fpCVTSI , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
278                                InstrStage<1, [A8_NPipe]>], [7, 1]>,
279   //
280   // Double-Precision FP to Integer Convert
281   InstrItinData<IIC_fpCVTDI , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
282                                InstrStage<8, [A8_NPipe], 0>,
283                                InstrStage<8, [A8_NLSPipe]>], [8, 1]>,
284   //
285   // Integer to Single-Precision FP Convert
286   InstrItinData<IIC_fpCVTIS , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
287                                InstrStage<1, [A8_NPipe]>], [7, 1]>,
288   //
289   // Integer to Double-Precision FP Convert
290   InstrItinData<IIC_fpCVTID , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
291                                InstrStage<8, [A8_NPipe], 0>,
292                                InstrStage<8, [A8_NLSPipe]>], [8, 1]>,
293   //
294   // Single-precision FP ALU
295   InstrItinData<IIC_fpALU32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
296                                InstrStage<1, [A8_NPipe]>], [7, 1, 1]>,
297   //
298   // Double-precision FP ALU
299   InstrItinData<IIC_fpALU64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
300                                InstrStage<9, [A8_NPipe], 0>,
301                                InstrStage<9, [A8_NLSPipe]>], [9, 1, 1]>,
302   //
303   // Single-precision FP Multiply
304   InstrItinData<IIC_fpMUL32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
305                                InstrStage<1, [A8_NPipe]>], [7, 1, 1]>,
306   //
307   // Double-precision FP Multiply
308   InstrItinData<IIC_fpMUL64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
309                                InstrStage<11, [A8_NPipe], 0>,
310                                InstrStage<11, [A8_NLSPipe]>], [11, 1, 1]>,
311   //
312   // Single-precision FP MAC
313   InstrItinData<IIC_fpMAC32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
314                                InstrStage<1, [A8_NPipe]>], [7, 2, 1, 1]>,
315   //
316   // Double-precision FP MAC
317   InstrItinData<IIC_fpMAC64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
318                                InstrStage<19, [A8_NPipe], 0>,
319                                InstrStage<19, [A8_NLSPipe]>], [19, 2, 1, 1]>,
320   //
321   // Single-precision FP DIV
322   InstrItinData<IIC_fpDIV32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
323                                InstrStage<20, [A8_NPipe], 0>,
324                                InstrStage<20, [A8_NLSPipe]>], [20, 1, 1]>,
325   //
326   // Double-precision FP DIV
327   InstrItinData<IIC_fpDIV64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
328                                InstrStage<29, [A8_NPipe], 0>,
329                                InstrStage<29, [A8_NLSPipe]>], [29, 1, 1]>,
330   //
331   // Single-precision FP SQRT
332   InstrItinData<IIC_fpSQRT32, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
333                                InstrStage<19, [A8_NPipe], 0>,
334                                InstrStage<19, [A8_NLSPipe]>], [19, 1]>,
335   //
336   // Double-precision FP SQRT
337   InstrItinData<IIC_fpSQRT64, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
338                                InstrStage<29, [A8_NPipe], 0>,
339                                InstrStage<29, [A8_NLSPipe]>], [29, 1]>,
340
341   //
342   // Integer to Single-precision Move
343   InstrItinData<IIC_fpMOVIS,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
344                                InstrStage<1, [A8_NPipe]>],
345                               [2, 1]>,
346   //
347   // Integer to Double-precision Move
348   InstrItinData<IIC_fpMOVID,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
349                                InstrStage<1, [A8_NPipe]>],
350                               [2, 1, 1]>,
351   //
352   // Single-precision to Integer Move
353   InstrItinData<IIC_fpMOVSI,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
354                                InstrStage<1, [A8_NPipe]>],
355                               [20, 1]>,
356   //
357   // Double-precision to Integer Move
358   InstrItinData<IIC_fpMOVDI,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
359                                InstrStage<1, [A8_NPipe]>],
360                               [20, 20, 1]>,
361
362   //
363   // Single-precision FP Load
364   InstrItinData<IIC_fpLoad32, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
365                                InstrStage<1, [A8_NLSPipe], 0>,
366                                InstrStage<1, [A8_LSPipe]>],
367                               [2, 1]>,
368   //
369   // Double-precision FP Load
370   InstrItinData<IIC_fpLoad64, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
371                                InstrStage<1, [A8_NLSPipe], 0>,
372                                InstrStage<1, [A8_LSPipe]>],
373                               [2, 1]>,
374   //
375   // FP Load Multiple
376   // FIXME: A8_LSPipe cycle time is dynamic, this assumes 3 to 4 registers.
377   InstrItinData<IIC_fpLoad_m, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
378                                InstrStage<1, [A8_NLSPipe], 0>,
379                                InstrStage<1, [A8_LSPipe]>,
380                                InstrStage<1, [A8_NLSPipe], 0>,
381                                InstrStage<1, [A8_LSPipe]>], [1, 1, 1, 2]>,
382   //
383   // FP Load Multiple + update
384   InstrItinData<IIC_fpLoad_mu,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
385                                InstrStage<1, [A8_NLSPipe], 0>,
386                                InstrStage<1, [A8_LSPipe]>,
387                                InstrStage<1, [A8_NLSPipe], 0>,
388                                InstrStage<1, [A8_LSPipe]>], [2, 1, 1, 1, 2]>,
389   //
390   // Single-precision FP Store
391   InstrItinData<IIC_fpStore32,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
392                                InstrStage<1, [A8_NLSPipe], 0>,
393                                InstrStage<1, [A8_LSPipe]>],
394                               [1, 1]>,
395   //
396   // Double-precision FP Store
397   InstrItinData<IIC_fpStore64,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
398                                InstrStage<1, [A8_NLSPipe], 0>,
399                                InstrStage<1, [A8_LSPipe]>],
400                               [1, 1]>,
401   //
402   // FP Store Multiple
403   InstrItinData<IIC_fpStore_m,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
404                                InstrStage<1, [A8_NLSPipe], 0>,
405                                InstrStage<1, [A8_LSPipe]>,
406                                InstrStage<1, [A8_NLSPipe], 0>,
407                                InstrStage<1, [A8_LSPipe]>], [1, 1, 1, 1]>,
408   //
409   // FP Store Multiple + update
410   InstrItinData<IIC_fpStore_mu,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
411                                 InstrStage<1, [A8_NLSPipe], 0>,
412                                 InstrStage<1, [A8_LSPipe]>,
413                                 InstrStage<1, [A8_NLSPipe], 0>,
414                                 InstrStage<1, [A8_LSPipe]>], [2, 1, 1, 1, 1]>,
415
416   // NEON
417   // Issue through integer pipeline, and execute in NEON unit.
418   //
419   // VLD1
420   InstrItinData<IIC_VLD1,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
421                                InstrStage<2, [A8_NLSPipe], 0>,
422                                InstrStage<2, [A8_LSPipe]>],
423                               [2, 1]>,
424   // VLD1x2
425   InstrItinData<IIC_VLD1x2,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
426                                InstrStage<2, [A8_NLSPipe], 0>,
427                                InstrStage<2, [A8_LSPipe]>],
428                               [2, 2, 1]>,
429   //
430   // VLD1x3
431   InstrItinData<IIC_VLD1x3,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
432                                InstrStage<3, [A8_NLSPipe], 0>,
433                                InstrStage<3, [A8_LSPipe]>],
434                               [2, 2, 3, 1]>,
435   //
436   // VLD1x4
437   InstrItinData<IIC_VLD1x4,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
438                                InstrStage<3, [A8_NLSPipe], 0>,
439                                InstrStage<3, [A8_LSPipe]>],
440                               [2, 2, 3, 3, 1]>,
441   //
442   // VLD1u
443   InstrItinData<IIC_VLD1u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
444                                InstrStage<2, [A8_NLSPipe], 0>,
445                                InstrStage<2, [A8_LSPipe]>],
446                               [2, 2, 1]>,
447   //
448   // VLD1x2u
449   InstrItinData<IIC_VLD1x2u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
450                                InstrStage<2, [A8_NLSPipe], 0>,
451                                InstrStage<2, [A8_LSPipe]>],
452                               [2, 2, 2, 1]>,
453   //
454   // VLD1x3u
455   InstrItinData<IIC_VLD1x3u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
456                                InstrStage<3, [A8_NLSPipe], 0>,
457                                InstrStage<3, [A8_LSPipe]>],
458                               [2, 2, 3, 2, 1]>,
459   //
460   // VLD1x4u
461   InstrItinData<IIC_VLD1x4u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
462                                InstrStage<3, [A8_NLSPipe], 0>,
463                                InstrStage<3, [A8_LSPipe]>],
464                               [2, 2, 3, 3, 2, 1]>,
465   //
466   // VLD1ln
467   InstrItinData<IIC_VLD1ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
468                                InstrStage<3, [A8_NLSPipe], 0>,
469                                InstrStage<3, [A8_LSPipe]>],
470                               [3, 1, 1, 1]>,
471   //
472   // VLD1lnu
473   InstrItinData<IIC_VLD1lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
474                                InstrStage<3, [A8_NLSPipe], 0>,
475                                InstrStage<3, [A8_LSPipe]>],
476                               [3, 2, 1, 1, 1, 1]>,
477   //
478   // VLD1dup
479   InstrItinData<IIC_VLD1dup,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
480                                InstrStage<2, [A8_NLSPipe], 0>,
481                                InstrStage<2, [A8_LSPipe]>],
482                               [2, 1]>,
483   //
484   // VLD1dupu
485   InstrItinData<IIC_VLD1dupu, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
486                                InstrStage<2, [A8_NLSPipe], 0>,
487                                InstrStage<2, [A8_LSPipe]>],
488                               [2, 2, 1, 1]>,
489   //
490   // VLD2
491   InstrItinData<IIC_VLD2,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
492                                InstrStage<2, [A8_NLSPipe], 0>,
493                                InstrStage<2, [A8_LSPipe]>],
494                               [2, 2, 1]>,
495   //
496   // VLD2x2
497   InstrItinData<IIC_VLD2x2,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
498                                InstrStage<3, [A8_NLSPipe], 0>,
499                                InstrStage<3, [A8_LSPipe]>],
500                               [2, 2, 3, 3, 1]>,
501   //
502   // VLD2ln
503   InstrItinData<IIC_VLD2ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
504                                InstrStage<3, [A8_NLSPipe], 0>,
505                                InstrStage<3, [A8_LSPipe]>],
506                               [3, 3, 1, 1, 1, 1]>,
507   //
508   // VLD2u
509   InstrItinData<IIC_VLD2u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
510                                InstrStage<2, [A8_NLSPipe], 0>,
511                                InstrStage<2, [A8_LSPipe]>],
512                               [2, 2, 2, 1, 1, 1]>,
513   //
514   // VLD2x2u
515   InstrItinData<IIC_VLD2x2u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
516                                InstrStage<3, [A8_NLSPipe], 0>,
517                                InstrStage<3, [A8_LSPipe]>],
518                               [2, 2, 3, 3, 2, 1]>,
519   //
520   // VLD2lnu
521   InstrItinData<IIC_VLD2lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
522                                InstrStage<3, [A8_NLSPipe], 0>,
523                                InstrStage<3, [A8_LSPipe]>],
524                               [3, 3, 2, 1, 1, 1, 1, 1]>,
525   //
526   // VLD3
527   InstrItinData<IIC_VLD3,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
528                                InstrStage<4, [A8_NLSPipe], 0>,
529                                InstrStage<4, [A8_LSPipe]>],
530                               [3, 3, 4, 1]>,
531   //
532   // VLD3ln
533   InstrItinData<IIC_VLD3ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
534                                InstrStage<5, [A8_NLSPipe], 0>,
535                                InstrStage<5, [A8_LSPipe]>],
536                               [4, 4, 5, 1, 1, 1, 1, 2]>,
537   //
538   // VLD3u
539   InstrItinData<IIC_VLD3u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
540                                InstrStage<4, [A8_NLSPipe], 0>,
541                                InstrStage<4, [A8_LSPipe]>],
542                               [3, 3, 4, 2, 1]>,
543   //
544   // VLD3lnu
545   InstrItinData<IIC_VLD3lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
546                                InstrStage<5, [A8_NLSPipe], 0>,
547                                InstrStage<5, [A8_LSPipe]>],
548                               [4, 4, 5, 2, 1, 1, 1, 1, 1, 2]>,
549   //
550   // VLD4
551   InstrItinData<IIC_VLD4,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
552                                InstrStage<4, [A8_NLSPipe], 0>,
553                                InstrStage<4, [A8_LSPipe]>],
554                               [3, 3, 4, 4, 1]>,
555   //
556   // VLD4ln
557   InstrItinData<IIC_VLD4ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
558                                InstrStage<5, [A8_NLSPipe], 0>,
559                                InstrStage<5, [A8_LSPipe]>],
560                               [4, 4, 5, 5, 1, 1, 1, 1, 2, 2]>,
561   //
562   // VLD4u
563   InstrItinData<IIC_VLD4u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
564                                InstrStage<4, [A8_NLSPipe], 0>,
565                                InstrStage<4, [A8_LSPipe]>],
566                               [3, 3, 4, 4, 2, 1]>,
567   //
568   // VLD4lnu
569   InstrItinData<IIC_VLD4lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
570                                InstrStage<5, [A8_NLSPipe], 0>,
571                                InstrStage<5, [A8_LSPipe]>],
572                               [4, 4, 5, 5, 2, 1, 1, 1, 1, 1, 2, 2]>,
573   //
574   // VST1
575   InstrItinData<IIC_VST1,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
576                                InstrStage<2, [A8_NLSPipe], 0>,
577                                InstrStage<2, [A8_LSPipe]>],
578                               [1, 1, 1]>,
579   //
580   // VST1x2
581   InstrItinData<IIC_VST1x2,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
582                                InstrStage<2, [A8_NLSPipe], 0>,
583                                InstrStage<2, [A8_LSPipe]>],
584                               [1, 1, 1, 1]>,
585   //
586   // VST1x3
587   InstrItinData<IIC_VST1x3,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
588                                InstrStage<3, [A8_NLSPipe], 0>,
589                                InstrStage<3, [A8_LSPipe]>],
590                               [1, 1, 1, 1, 2]>,
591   //
592   // VST1x4
593   InstrItinData<IIC_VST1x4,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
594                                InstrStage<3, [A8_NLSPipe], 0>,
595                                InstrStage<3, [A8_LSPipe]>],
596                               [1, 1, 1, 1, 2, 2]>,
597   //
598   // VST1u
599   InstrItinData<IIC_VST1u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
600                                InstrStage<2, [A8_NLSPipe], 0>,
601                                InstrStage<2, [A8_LSPipe]>],
602                               [2, 1, 1, 1, 1]>,
603   //
604   // VST1x2u
605   InstrItinData<IIC_VST1x2u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
606                                InstrStage<2, [A8_NLSPipe], 0>,
607                                InstrStage<2, [A8_LSPipe]>],
608                               [2, 1, 1, 1, 1, 1]>,
609   //
610   // VST1x3u
611   InstrItinData<IIC_VST1x3u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
612                                InstrStage<3, [A8_NLSPipe], 0>,
613                                InstrStage<3, [A8_LSPipe]>],
614                               [2, 1, 1, 1, 1, 1, 2]>,
615   //
616   // VST1x4u
617   InstrItinData<IIC_VST1x4u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
618                                InstrStage<3, [A8_NLSPipe], 0>,
619                                InstrStage<3, [A8_LSPipe]>],
620                               [2, 1, 1, 1, 1, 1, 2, 2]>,
621   //
622   // VST1ln
623   InstrItinData<IIC_VST1ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
624                                InstrStage<2, [A8_NLSPipe], 0>,
625                                InstrStage<2, [A8_LSPipe]>],
626                               [1, 1, 1]>,
627   //
628   // VST1lnu
629   InstrItinData<IIC_VST1lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
630                                InstrStage<2, [A8_NLSPipe], 0>,
631                                InstrStage<2, [A8_LSPipe]>],
632                               [2, 1, 1, 1, 1]>,
633   //
634   // VST2
635   InstrItinData<IIC_VST2,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
636                                InstrStage<2, [A8_NLSPipe], 0>,
637                                InstrStage<2, [A8_LSPipe]>],
638                               [1, 1, 1, 1]>,
639   //
640   // VST2x2
641   InstrItinData<IIC_VST2x2,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
642                                InstrStage<4, [A8_NLSPipe], 0>,
643                                InstrStage<4, [A8_LSPipe]>],
644                               [1, 1, 1, 1, 2, 2]>,
645   //
646   // VST2u
647   InstrItinData<IIC_VST2u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
648                                InstrStage<2, [A8_NLSPipe], 0>,
649                                InstrStage<2, [A8_LSPipe]>],
650                               [2, 1, 1, 1, 1, 1]>,
651   //
652   // VST2x2u
653   InstrItinData<IIC_VST2x2u,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
654                                InstrStage<4, [A8_NLSPipe], 0>,
655                                InstrStage<4, [A8_LSPipe]>],
656                               [2, 1, 1, 1, 1, 1, 2, 2]>,
657   //
658   // VST2ln
659   InstrItinData<IIC_VST2ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
660                                InstrStage<2, [A8_NLSPipe], 0>,
661                                InstrStage<2, [A8_LSPipe]>],
662                               [1, 1, 1, 1]>,
663   //
664   // VST2lnu
665   InstrItinData<IIC_VST2lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
666                                InstrStage<2, [A8_NLSPipe], 0>,
667                                InstrStage<2, [A8_LSPipe]>],
668                               [2, 1, 1, 1, 1, 1]>,
669   //
670   // VST3
671   InstrItinData<IIC_VST3,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
672                                InstrStage<3, [A8_NLSPipe], 0>,
673                                InstrStage<3, [A8_LSPipe]>],
674                               [1, 1, 1, 1, 2]>,
675   //
676   // VST3u
677   InstrItinData<IIC_VST3u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
678                                InstrStage<3, [A8_NLSPipe], 0>,
679                                InstrStage<3, [A8_LSPipe]>],
680                               [2, 1, 1, 1, 1, 1, 2]>,
681   //
682   // VST3ln
683   InstrItinData<IIC_VST3ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
684                                InstrStage<3, [A8_NLSPipe], 0>,
685                                InstrStage<3, [A8_LSPipe]>],
686                               [1, 1, 1, 1, 2]>,
687   //
688   // VST3lnu
689   InstrItinData<IIC_VST3lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
690                                InstrStage<3, [A8_NLSPipe], 0>,
691                                InstrStage<3, [A8_LSPipe]>],
692                               [2, 1, 1, 1, 1, 1, 2]>,
693   //
694   // VST4
695   InstrItinData<IIC_VST4,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
696                                InstrStage<4, [A8_NLSPipe], 0>,
697                                InstrStage<4, [A8_LSPipe]>],
698                               [1, 1, 1, 1, 2, 2]>,
699   //
700   // VST4u
701   InstrItinData<IIC_VST4u,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
702                                InstrStage<4, [A8_NLSPipe], 0>,
703                                InstrStage<4, [A8_LSPipe]>],
704                               [2, 1, 1, 1, 1, 1, 2, 2]>,
705   //
706   // VST4ln
707   InstrItinData<IIC_VST4ln,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
708                                InstrStage<4, [A8_NLSPipe], 0>,
709                                InstrStage<4, [A8_LSPipe]>],
710                               [1, 1, 1, 1, 2, 2]>,
711   //
712   // VST4lnu
713   InstrItinData<IIC_VST4lnu,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
714                                InstrStage<4, [A8_NLSPipe], 0>,
715                                InstrStage<4, [A8_LSPipe]>],
716                               [2, 1, 1, 1, 1, 1, 2, 2]>,
717   //
718   // Double-register FP Unary
719   InstrItinData<IIC_VUNAD,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
720                                InstrStage<1, [A8_NPipe]>], [5, 2]>,
721   //
722   // Quad-register FP Unary
723   // Result written in N5, but that is relative to the last cycle of multicycle,
724   // so we use 6 for those cases
725   InstrItinData<IIC_VUNAQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
726                                InstrStage<2, [A8_NPipe]>], [6, 2]>,
727   //
728   // Double-register FP Binary
729   InstrItinData<IIC_VBIND,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
730                                InstrStage<1, [A8_NPipe]>], [5, 2, 2]>,
731   //
732   // VPADD, etc.
733   InstrItinData<IIC_VPBIND,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
734                                InstrStage<1, [A8_NPipe]>], [5, 2, 2]>,
735   //
736   // Double-register FP VMUL
737   InstrItinData<IIC_VFMULD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
738                                InstrStage<1, [A8_NPipe]>], [5, 2, 1]>,
739
740   //
741   // Quad-register FP Binary
742   // Result written in N5, but that is relative to the last cycle of multicycle,
743   // so we use 6 for those cases
744   InstrItinData<IIC_VBINQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
745                                InstrStage<2, [A8_NPipe]>], [6, 2, 2]>,
746   //
747   // Quad-register FP VMUL
748   InstrItinData<IIC_VFMULQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
749                                InstrStage<1, [A8_NPipe]>], [6, 2, 1]>,
750   //
751   // Move
752   InstrItinData<IIC_VMOV,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
753                                InstrStage<1, [A8_NPipe]>], [1, 1]>,
754   //
755   // Move Immediate
756   InstrItinData<IIC_VMOVImm,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
757                                InstrStage<1, [A8_NPipe]>], [3]>,
758   //
759   // Double-register Permute Move
760   InstrItinData<IIC_VMOVD,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
761                                InstrStage<1, [A8_NLSPipe]>], [2, 1]>,
762   //
763   // Quad-register Permute Move
764   // Result written in N2, but that is relative to the last cycle of multicycle,
765   // so we use 3 for those cases
766   InstrItinData<IIC_VMOVQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
767                                InstrStage<2, [A8_NLSPipe]>], [3, 1]>,
768   //
769   // Integer to Single-precision Move
770   InstrItinData<IIC_VMOVIS ,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
771                                InstrStage<1, [A8_NLSPipe]>], [2, 1]>,
772   //
773   // Integer to Double-precision Move
774   InstrItinData<IIC_VMOVID ,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
775                                InstrStage<1, [A8_NLSPipe]>], [2, 1, 1]>,
776   //
777   // Single-precision to Integer Move
778   InstrItinData<IIC_VMOVSI ,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
779                                InstrStage<1, [A8_NLSPipe]>], [20, 1]>,
780   //
781   // Double-precision to Integer Move
782   InstrItinData<IIC_VMOVDI ,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
783                                InstrStage<1, [A8_NLSPipe]>], [20, 20, 1]>,
784   //
785   // Integer to Lane Move
786   InstrItinData<IIC_VMOVISL , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
787                                InstrStage<2, [A8_NLSPipe]>], [3, 1, 1]>,
788   //
789   // Vector narrow move
790   InstrItinData<IIC_VMOVN   , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
791                                InstrStage<1, [A8_NPipe]>], [2, 1]>,
792   //
793   // Double-register Permute
794   InstrItinData<IIC_VPERMD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
795                                InstrStage<1, [A8_NLSPipe]>], [2, 2, 1, 1]>,
796   //
797   // Quad-register Permute
798   // Result written in N2, but that is relative to the last cycle of multicycle,
799   // so we use 3 for those cases
800   InstrItinData<IIC_VPERMQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
801                                InstrStage<2, [A8_NLSPipe]>], [3, 3, 1, 1]>,
802   //
803   // Quad-register Permute (3 cycle issue)
804   // Result written in N2, but that is relative to the last cycle of multicycle,
805   // so we use 4 for those cases
806   InstrItinData<IIC_VPERMQ3,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
807                                InstrStage<1, [A8_NLSPipe]>,
808                                InstrStage<1, [A8_NPipe], 0>,
809                                InstrStage<2, [A8_NLSPipe]>], [4, 4, 1, 1]>,
810   //
811   // Double-register FP Multiple-Accumulate
812   InstrItinData<IIC_VMACD,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
813                                InstrStage<1, [A8_NPipe]>], [9, 3, 2, 2]>,
814   //
815   // Quad-register FP Multiple-Accumulate
816   // Result written in N9, but that is relative to the last cycle of multicycle,
817   // so we use 10 for those cases
818   InstrItinData<IIC_VMACQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
819                                InstrStage<2, [A8_NPipe]>], [10, 3, 2, 2]>,
820   //
821   // Double-register Reciprical Step
822   InstrItinData<IIC_VRECSD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
823                                InstrStage<1, [A8_NPipe]>], [9, 2, 2]>,
824   //
825   // Quad-register Reciprical Step
826   InstrItinData<IIC_VRECSQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
827                                InstrStage<2, [A8_NPipe]>], [10, 2, 2]>,
828   //
829   // Double-register Integer Count
830   InstrItinData<IIC_VCNTiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
831                                InstrStage<1, [A8_NPipe]>], [3, 2, 2]>,
832   //
833   // Quad-register Integer Count
834   // Result written in N3, but that is relative to the last cycle of multicycle,
835   // so we use 4 for those cases
836   InstrItinData<IIC_VCNTiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
837                                InstrStage<2, [A8_NPipe]>], [4, 2, 2]>,
838   //
839   // Double-register Integer Unary
840   InstrItinData<IIC_VUNAiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
841                                InstrStage<1, [A8_NPipe]>], [4, 2]>,
842   //
843   // Quad-register Integer Unary
844   InstrItinData<IIC_VUNAiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
845                                InstrStage<1, [A8_NPipe]>], [4, 2]>,
846   //
847   // Double-register Integer Q-Unary
848   InstrItinData<IIC_VQUNAiD,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
849                                InstrStage<1, [A8_NPipe]>], [4, 1]>,
850   //
851   // Quad-register Integer CountQ-Unary
852   InstrItinData<IIC_VQUNAiQ,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
853                                InstrStage<1, [A8_NPipe]>], [4, 1]>,
854   //
855   // Double-register Integer Binary
856   InstrItinData<IIC_VBINiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
857                                InstrStage<1, [A8_NPipe]>], [3, 2, 2]>,
858   //
859   // Quad-register Integer Binary
860   InstrItinData<IIC_VBINiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
861                                InstrStage<1, [A8_NPipe]>], [3, 2, 2]>,
862   //
863   // Double-register Integer Binary (4 cycle)
864   InstrItinData<IIC_VBINi4D,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
865                                InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
866   //
867   // Quad-register Integer Binary (4 cycle)
868   InstrItinData<IIC_VBINi4Q,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
869                                InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
870
871   //
872   // Double-register Integer Subtract
873   InstrItinData<IIC_VSUBiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
874                                InstrStage<1, [A8_NPipe]>], [3, 2, 1]>,
875   //
876   // Quad-register Integer Subtract
877   InstrItinData<IIC_VSUBiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
878                                InstrStage<1, [A8_NPipe]>], [3, 2, 1]>,
879   //
880   // Double-register Integer Subtract
881   InstrItinData<IIC_VSUBi4D,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
882                                InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
883   //
884   // Quad-register Integer Subtract
885   InstrItinData<IIC_VSUBi4Q,  [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
886                                InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
887   //
888   // Double-register Integer Shift
889   InstrItinData<IIC_VSHLiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
890                                InstrStage<1, [A8_NPipe]>], [3, 1, 1]>,
891   //
892   // Quad-register Integer Shift
893   InstrItinData<IIC_VSHLiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
894                                InstrStage<2, [A8_NPipe]>], [4, 1, 1]>,
895   //
896   // Double-register Integer Shift (4 cycle)
897   InstrItinData<IIC_VSHLi4D,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
898                                InstrStage<1, [A8_NPipe]>], [4, 1, 1]>,
899   //
900   // Quad-register Integer Shift (4 cycle)
901   InstrItinData<IIC_VSHLi4Q,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
902                                InstrStage<2, [A8_NPipe]>], [5, 1, 1]>,
903   //
904   // Double-register Integer Pair Add Long
905   InstrItinData<IIC_VPALiD,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
906                                InstrStage<1, [A8_NPipe]>], [6, 3, 1]>,
907   //
908   // Quad-register Integer Pair Add Long
909   InstrItinData<IIC_VPALiQ,   [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
910                                InstrStage<2, [A8_NPipe]>], [7, 3, 1]>,
911   //
912   // Double-register Absolute Difference and Accumulate
913   InstrItinData<IIC_VABAD,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
914                                InstrStage<1, [A8_NPipe]>], [6, 3, 2, 1]>,
915   //
916   // Quad-register Absolute Difference and Accumulate
917   InstrItinData<IIC_VABAQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
918                                InstrStage<2, [A8_NPipe]>], [6, 3, 2, 1]>,
919
920   //
921   // Double-register Integer Multiply (.8, .16)
922   InstrItinData<IIC_VMULi16D, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
923                                InstrStage<1, [A8_NPipe]>], [6, 2, 2]>,
924   //
925   // Double-register Integer Multiply (.32)
926   InstrItinData<IIC_VMULi32D, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
927                                InstrStage<2, [A8_NPipe]>], [7, 2, 1]>,
928   //
929   // Quad-register Integer Multiply (.8, .16)
930   InstrItinData<IIC_VMULi16Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
931                                InstrStage<2, [A8_NPipe]>], [7, 2, 2]>,
932   //
933   // Quad-register Integer Multiply (.32)
934   InstrItinData<IIC_VMULi32Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
935                                InstrStage<1, [A8_NPipe]>,
936                                InstrStage<2, [A8_NLSPipe], 0>,
937                                InstrStage<3, [A8_NPipe]>], [9, 2, 1]>,
938   //
939   // Double-register Integer Multiply-Accumulate (.8, .16)
940   InstrItinData<IIC_VMACi16D, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
941                                InstrStage<1, [A8_NPipe]>], [6, 3, 2, 2]>,
942   //
943   // Double-register Integer Multiply-Accumulate (.32)
944   InstrItinData<IIC_VMACi32D, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
945                                InstrStage<2, [A8_NPipe]>], [7, 3, 2, 1]>,
946   //
947   // Quad-register Integer Multiply-Accumulate (.8, .16)
948   InstrItinData<IIC_VMACi16Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
949                                InstrStage<2, [A8_NPipe]>], [7, 3, 2, 2]>,
950   //
951   // Quad-register Integer Multiply-Accumulate (.32)
952   InstrItinData<IIC_VMACi32Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
953                                InstrStage<1, [A8_NPipe]>,
954                                InstrStage<2, [A8_NLSPipe], 0>,
955                                InstrStage<3, [A8_NPipe]>], [9, 3, 2, 1]>,
956   //
957   // Double-register VEXT
958   InstrItinData<IIC_VEXTD,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
959                                InstrStage<1, [A8_NLSPipe]>], [2, 1, 1]>,
960   //
961   // Quad-register VEXT
962   InstrItinData<IIC_VEXTQ,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
963                                InstrStage<2, [A8_NLSPipe]>], [3, 1, 1]>,
964   //
965   // VTB
966   InstrItinData<IIC_VTB1,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
967                                InstrStage<2, [A8_NLSPipe]>], [3, 2, 1]>,
968   InstrItinData<IIC_VTB2,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
969                                InstrStage<2, [A8_NLSPipe]>], [3, 2, 2, 1]>,
970   InstrItinData<IIC_VTB3,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
971                                InstrStage<1, [A8_NLSPipe]>,
972                                InstrStage<1, [A8_NPipe], 0>,
973                                InstrStage<2, [A8_NLSPipe]>], [4, 2, 2, 3, 1]>,
974   InstrItinData<IIC_VTB4,     [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
975                                InstrStage<1, [A8_NLSPipe]>,
976                                InstrStage<1, [A8_NPipe], 0>,
977                                InstrStage<2, [A8_NLSPipe]>],[4, 2, 2, 3, 3, 1]>,
978   //
979   // VTBX
980   InstrItinData<IIC_VTBX1,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
981                                InstrStage<2, [A8_NLSPipe]>], [3, 1, 2, 1]>,
982   InstrItinData<IIC_VTBX2,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
983                                InstrStage<2, [A8_NLSPipe]>], [3, 1, 2, 2, 1]>,
984   InstrItinData<IIC_VTBX3,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
985                                InstrStage<1, [A8_NLSPipe]>,
986                                InstrStage<1, [A8_NPipe], 0>,
987                                InstrStage<2, [A8_NLSPipe]>],[4, 1, 2, 2, 3, 1]>,
988   InstrItinData<IIC_VTBX4,    [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
989                                InstrStage<1, [A8_NLSPipe]>,
990                                InstrStage<1, [A8_NPipe], 0>,
991                             InstrStage<2, [A8_NLSPipe]>], [4, 1, 2, 2, 3, 3, 1]>
992 ]>;