1 //=- ARMScheduleA8.td - ARM Cortex-A8 Scheduling Definitions -*- tablegen -*-=//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This file defines the itinerary class data for the ARM Cortex A8 processors.
12 //===----------------------------------------------------------------------===//
15 // Scheduling information derived from "Cortex-A8 Technical Reference Manual".
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
23 // Dual issue pipeline represented by A8_Pipe0 | A8_Pipe1
25 def CortexA8Itineraries : ProcessorItineraries<
26 [A8_Pipe0, A8_Pipe1, A8_LSPipe, A8_NPipe, A8_NLSPipe],
28 // Two fully-pipelined integer ALU pipelines
31 InstrItinData<IIC_iALUx , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>]>,
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]>,
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]>,
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]>,
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]>,
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]>,
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]>,
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]>,
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]>,
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]>,
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
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]>,
98 // Integer load pipeline
101 InstrItinData<IIC_iLoad_i , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
102 InstrStage<1, [A8_LSPipe]>], [3, 1]>,
103 InstrItinData<IIC_iLoad_bh_i, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
104 InstrStage<1, [A8_LSPipe]>], [3, 1]>,
105 InstrItinData<IIC_iLoad_d_i, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
106 InstrStage<1, [A8_LSPipe]>], [3, 1]>,
109 InstrItinData<IIC_iLoad_r , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
110 InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
111 InstrItinData<IIC_iLoad_bh_r, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
112 InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
113 InstrItinData<IIC_iLoad_d_r , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
114 InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
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], 0>,
119 InstrStage<1, [A8_LSPipe]>], [4, 1, 1]>,
120 InstrItinData<IIC_iLoad_bh_si,[InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
121 InstrStage<1, [A8_LSPipe]>], [4, 1, 1]>,
123 // Immediate offset with update
124 InstrItinData<IIC_iLoad_iu , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
125 InstrStage<1, [A8_LSPipe]>], [3, 2, 1]>,
126 InstrItinData<IIC_iLoad_bh_iu,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
127 InstrStage<1, [A8_LSPipe]>], [3, 2, 1]>,
129 // Register offset with update
130 InstrItinData<IIC_iLoad_ru , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
131 InstrStage<1, [A8_LSPipe]>], [3, 2, 1, 1]>,
132 InstrItinData<IIC_iLoad_bh_ru,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
133 InstrStage<1, [A8_LSPipe]>], [3, 2, 1, 1]>,
134 InstrItinData<IIC_iLoad_d_ru, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
135 InstrStage<1, [A8_LSPipe]>], [3, 2, 1, 1]>,
137 // Scaled register offset with update, issues over 2 cycles
138 InstrItinData<IIC_iLoad_siu , [InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
139 InstrStage<2, [A8_LSPipe]>], [4, 3, 1, 1]>,
140 InstrItinData<IIC_iLoad_bh_siu,[InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
141 InstrStage<2, [A8_LSPipe]>], [4, 3, 1, 1]>,
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<2, [A8_Pipe0], 0>,
146 InstrStage<2, [A8_LSPipe]>], [1, 1, 1, 1, 3]>,
148 // Load multiple + update, defs are the 1st and 5th operands.
149 InstrItinData<IIC_iLoad_mu , [InstrStage<3, [A8_Pipe0], 0>,
150 InstrStage<3, [A8_LSPipe]>], [2, 1, 1, 1, 3]>,
152 // Load multiple plus branch
153 InstrItinData<IIC_iLoad_mBr, [InstrStage<3, [A8_Pipe0], 0>,
154 InstrStage<3, [A8_LSPipe]>,
155 InstrStage<1, [A8_Pipe0, A8_Pipe1]>],
158 // Pop, def is the 3rd operand.
159 InstrItinData<IIC_iPop , [InstrStage<3, [A8_Pipe0], 0>,
160 InstrStage<3, [A8_LSPipe]>], [1, 1, 3]>,
162 // Push, def is the 3th operand.
163 InstrItinData<IIC_iPop_Br, [InstrStage<3, [A8_Pipe0], 0>,
164 InstrStage<3, [A8_LSPipe]>,
165 InstrStage<1, [A8_Pipe0, A8_Pipe1]>],
169 // iLoadi + iALUr for t2LDRpci_pic.
170 InstrItinData<IIC_iLoadiALU, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
171 InstrStage<1, [A8_LSPipe]>,
172 InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [4, 1]>,
175 // Integer store pipeline
178 InstrItinData<IIC_iStore_i , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
179 InstrStage<1, [A8_LSPipe]>], [3, 1]>,
180 InstrItinData<IIC_iStore_bh_i,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
181 InstrStage<1, [A8_LSPipe]>], [3, 1]>,
182 InstrItinData<IIC_iStore_d_i, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
183 InstrStage<1, [A8_LSPipe]>], [3, 1]>,
186 InstrItinData<IIC_iStore_r , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
187 InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
188 InstrItinData<IIC_iStore_bh_r,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
189 InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
190 InstrItinData<IIC_iStore_d_r, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
191 InstrStage<1, [A8_LSPipe]>], [3, 1, 1]>,
193 // Scaled register offset, issues over 2 cycles
194 InstrItinData<IIC_iStore_si , [InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
195 InstrStage<2, [A8_LSPipe]>], [3, 1, 1]>,
196 InstrItinData<IIC_iStore_bh_si,[InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
197 InstrStage<2, [A8_LSPipe]>], [3, 1, 1]>,
199 // Immediate offset with update
200 InstrItinData<IIC_iStore_iu , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
201 InstrStage<1, [A8_LSPipe]>], [2, 3, 1]>,
202 InstrItinData<IIC_iStore_bh_iu,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
203 InstrStage<1, [A8_LSPipe]>], [2, 3, 1]>,
205 // Register offset with update
206 InstrItinData<IIC_iStore_ru , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
207 InstrStage<1, [A8_LSPipe]>], [2, 3, 1, 1]>,
208 InstrItinData<IIC_iStore_bh_ru,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
209 InstrStage<1, [A8_LSPipe]>], [2, 3, 1, 1]>,
210 InstrItinData<IIC_iStore_d_ru, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
211 InstrStage<1, [A8_LSPipe]>], [2, 3, 1, 1]>,
213 // Scaled register offset with update, issues over 2 cycles
214 InstrItinData<IIC_iStore_siu, [InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
215 InstrStage<2, [A8_LSPipe]>], [3, 3, 1, 1]>,
216 InstrItinData<IIC_iStore_bh_siu,[InstrStage<2, [A8_Pipe0, A8_Pipe1], 0>,
217 InstrStage<2, [A8_LSPipe]>], [3, 3, 1, 1]>,
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<2, [A8_Pipe0], 0>,
222 InstrStage<2, [A8_LSPipe]>]>,
224 // Store multiple + update
225 InstrItinData<IIC_iStore_mu, [InstrStage<2, [A8_Pipe0], 0>,
226 InstrStage<2, [A8_LSPipe]>], [2]>,
230 InstrItinData<IIC_Preload, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>], [2, 2]>,
234 // no delay slots, so the latency of a branch is unimportant
235 InstrItinData<IIC_Br , [InstrStage<1, [A8_Pipe0, A8_Pipe1]>]>,
238 // Issue through integer pipeline, and execute in NEON unit. We assume
239 // RunFast mode so that NFP pipeline is used for single-precision when
242 // FP Special Register to Integer Register File Move
243 InstrItinData<IIC_fpSTAT , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
244 InstrStage<1, [A8_NLSPipe]>], [20]>,
246 // Single-precision FP Unary
247 InstrItinData<IIC_fpUNA32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
248 InstrStage<1, [A8_NPipe]>], [7, 1]>,
250 // Double-precision FP Unary
251 InstrItinData<IIC_fpUNA64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
252 InstrStage<4, [A8_NPipe], 0>,
253 InstrStage<4, [A8_NLSPipe]>], [4, 1]>,
255 // Single-precision FP Compare
256 InstrItinData<IIC_fpCMP32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
257 InstrStage<1, [A8_NPipe]>], [1, 1]>,
259 // Double-precision FP Compare
260 InstrItinData<IIC_fpCMP64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
261 InstrStage<4, [A8_NPipe], 0>,
262 InstrStage<4, [A8_NLSPipe]>], [4, 1]>,
264 // Single to Double FP Convert
265 InstrItinData<IIC_fpCVTSD , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
266 InstrStage<7, [A8_NPipe], 0>,
267 InstrStage<7, [A8_NLSPipe]>], [7, 1]>,
269 // Double to Single FP Convert
270 InstrItinData<IIC_fpCVTDS , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
271 InstrStage<5, [A8_NPipe], 0>,
272 InstrStage<5, [A8_NLSPipe]>], [5, 1]>,
274 // Single-Precision FP to Integer Convert
275 InstrItinData<IIC_fpCVTSI , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
276 InstrStage<1, [A8_NPipe]>], [7, 1]>,
278 // Double-Precision FP to Integer Convert
279 InstrItinData<IIC_fpCVTDI , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
280 InstrStage<8, [A8_NPipe], 0>,
281 InstrStage<8, [A8_NLSPipe]>], [8, 1]>,
283 // Integer to Single-Precision FP Convert
284 InstrItinData<IIC_fpCVTIS , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
285 InstrStage<1, [A8_NPipe]>], [7, 1]>,
287 // Integer to Double-Precision FP Convert
288 InstrItinData<IIC_fpCVTID , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
289 InstrStage<8, [A8_NPipe], 0>,
290 InstrStage<8, [A8_NLSPipe]>], [8, 1]>,
292 // Single-precision FP ALU
293 InstrItinData<IIC_fpALU32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
294 InstrStage<1, [A8_NPipe]>], [7, 1, 1]>,
296 // Double-precision FP ALU
297 InstrItinData<IIC_fpALU64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
298 InstrStage<9, [A8_NPipe], 0>,
299 InstrStage<9, [A8_NLSPipe]>], [9, 1, 1]>,
301 // Single-precision FP Multiply
302 InstrItinData<IIC_fpMUL32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
303 InstrStage<1, [A8_NPipe]>], [7, 1, 1]>,
305 // Double-precision FP Multiply
306 InstrItinData<IIC_fpMUL64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
307 InstrStage<11, [A8_NPipe], 0>,
308 InstrStage<11, [A8_NLSPipe]>], [11, 1, 1]>,
310 // Single-precision FP MAC
311 InstrItinData<IIC_fpMAC32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
312 InstrStage<1, [A8_NPipe]>], [7, 2, 1, 1]>,
314 // Double-precision FP MAC
315 InstrItinData<IIC_fpMAC64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
316 InstrStage<19, [A8_NPipe], 0>,
317 InstrStage<19, [A8_NLSPipe]>], [19, 2, 1, 1]>,
319 // Single-precision FP DIV
320 InstrItinData<IIC_fpDIV32 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
321 InstrStage<20, [A8_NPipe], 0>,
322 InstrStage<20, [A8_NLSPipe]>], [20, 1, 1]>,
324 // Double-precision FP DIV
325 InstrItinData<IIC_fpDIV64 , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
326 InstrStage<29, [A8_NPipe], 0>,
327 InstrStage<29, [A8_NLSPipe]>], [29, 1, 1]>,
329 // Single-precision FP SQRT
330 InstrItinData<IIC_fpSQRT32, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
331 InstrStage<19, [A8_NPipe], 0>,
332 InstrStage<19, [A8_NLSPipe]>], [19, 1]>,
334 // Double-precision FP SQRT
335 InstrItinData<IIC_fpSQRT64, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
336 InstrStage<29, [A8_NPipe], 0>,
337 InstrStage<29, [A8_NLSPipe]>], [29, 1]>,
340 // Integer to Single-precision Move
341 InstrItinData<IIC_fpMOVIS, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
342 InstrStage<1, [A8_NPipe]>],
345 // Integer to Double-precision Move
346 InstrItinData<IIC_fpMOVID, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
347 InstrStage<1, [A8_NPipe]>],
350 // Single-precision to Integer Move
351 InstrItinData<IIC_fpMOVSI, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
352 InstrStage<1, [A8_NPipe]>],
355 // Double-precision to Integer Move
356 InstrItinData<IIC_fpMOVDI, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
357 InstrStage<1, [A8_NPipe]>],
361 // Single-precision FP Load
362 InstrItinData<IIC_fpLoad32, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
363 InstrStage<1, [A8_NLSPipe], 0>,
364 InstrStage<1, [A8_LSPipe]>],
367 // Double-precision FP Load
368 InstrItinData<IIC_fpLoad64, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
369 InstrStage<1, [A8_NLSPipe], 0>,
370 InstrStage<1, [A8_LSPipe]>],
374 // FIXME: A8_LSPipe cycle time is dynamic, this assumes 3 to 4 registers.
375 InstrItinData<IIC_fpLoad_m, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
376 InstrStage<1, [A8_NLSPipe], 0>,
377 InstrStage<1, [A8_LSPipe]>,
378 InstrStage<1, [A8_NLSPipe], 0>,
379 InstrStage<1, [A8_LSPipe]>], [1, 1, 1, 2]>,
381 // FP Load Multiple + update
382 InstrItinData<IIC_fpLoad_mu,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
383 InstrStage<1, [A8_NLSPipe], 0>,
384 InstrStage<1, [A8_LSPipe]>,
385 InstrStage<1, [A8_NLSPipe], 0>,
386 InstrStage<1, [A8_LSPipe]>], [2, 1, 1, 1, 2]>,
388 // Single-precision FP Store
389 InstrItinData<IIC_fpStore32,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
390 InstrStage<1, [A8_NLSPipe], 0>,
391 InstrStage<1, [A8_LSPipe]>],
394 // Double-precision FP Store
395 InstrItinData<IIC_fpStore64,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
396 InstrStage<1, [A8_NLSPipe], 0>,
397 InstrStage<1, [A8_LSPipe]>],
401 InstrItinData<IIC_fpStore_m,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
402 InstrStage<1, [A8_NLSPipe], 0>,
403 InstrStage<1, [A8_LSPipe]>,
404 InstrStage<1, [A8_NLSPipe], 0>,
405 InstrStage<1, [A8_LSPipe]>], [1, 1, 1, 1]>,
407 // FP Store Multiple + update
408 InstrItinData<IIC_fpStore_mu,[InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
409 InstrStage<1, [A8_NLSPipe], 0>,
410 InstrStage<1, [A8_LSPipe]>,
411 InstrStage<1, [A8_NLSPipe], 0>,
412 InstrStage<1, [A8_LSPipe]>], [2, 1, 1, 1, 1]>,
415 // Issue through integer pipeline, and execute in NEON unit.
418 InstrItinData<IIC_VLD1, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
419 InstrStage<2, [A8_NLSPipe], 0>,
420 InstrStage<2, [A8_LSPipe]>],
423 InstrItinData<IIC_VLD1x2, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
424 InstrStage<2, [A8_NLSPipe], 0>,
425 InstrStage<2, [A8_LSPipe]>],
429 InstrItinData<IIC_VLD1x3, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
430 InstrStage<3, [A8_NLSPipe], 0>,
431 InstrStage<3, [A8_LSPipe]>],
435 InstrItinData<IIC_VLD1x4, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
436 InstrStage<3, [A8_NLSPipe], 0>,
437 InstrStage<3, [A8_LSPipe]>],
441 InstrItinData<IIC_VLD1u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
442 InstrStage<2, [A8_NLSPipe], 0>,
443 InstrStage<2, [A8_LSPipe]>],
447 InstrItinData<IIC_VLD1x2u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
448 InstrStage<2, [A8_NLSPipe], 0>,
449 InstrStage<2, [A8_LSPipe]>],
453 InstrItinData<IIC_VLD1x3u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
454 InstrStage<3, [A8_NLSPipe], 0>,
455 InstrStage<3, [A8_LSPipe]>],
459 InstrItinData<IIC_VLD1x4u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
460 InstrStage<3, [A8_NLSPipe], 0>,
461 InstrStage<3, [A8_LSPipe]>],
465 InstrItinData<IIC_VLD1ln, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
466 InstrStage<3, [A8_NLSPipe], 1>,
467 InstrStage<3, [A8_LSPipe]>],
471 InstrItinData<IIC_VLD1lnu, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
472 InstrStage<3, [A8_NLSPipe], 1>,
473 InstrStage<3, [A8_LSPipe]>],
477 InstrItinData<IIC_VLD2, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
478 InstrStage<2, [A8_NLSPipe], 0>,
479 InstrStage<2, [A8_LSPipe]>],
483 InstrItinData<IIC_VLD2x2, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
484 InstrStage<3, [A8_NLSPipe], 0>,
485 InstrStage<3, [A8_LSPipe]>],
489 InstrItinData<IIC_VLD2ln, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
490 InstrStage<3, [A8_NLSPipe], 0>,
491 InstrStage<3, [A8_LSPipe]>],
495 InstrItinData<IIC_VLD2u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
496 InstrStage<2, [A8_NLSPipe], 0>,
497 InstrStage<2, [A8_LSPipe]>],
501 InstrItinData<IIC_VLD2x2u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
502 InstrStage<3, [A8_NLSPipe], 0>,
503 InstrStage<3, [A8_LSPipe]>],
507 InstrItinData<IIC_VLD2lnu, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
508 InstrStage<3, [A8_NLSPipe], 0>,
509 InstrStage<3, [A8_LSPipe]>],
510 [3, 3, 2, 1, 1, 1, 1, 1]>,
513 InstrItinData<IIC_VLD3, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
514 InstrStage<4, [A8_NLSPipe], 0>,
515 InstrStage<4, [A8_LSPipe]>],
519 InstrItinData<IIC_VLD3ln, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
520 InstrStage<5, [A8_NLSPipe], 0>,
521 InstrStage<5, [A8_LSPipe]>],
522 [4, 4, 5, 1, 1, 1, 1, 2]>,
525 InstrItinData<IIC_VLD3u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
526 InstrStage<4, [A8_NLSPipe], 0>,
527 InstrStage<4, [A8_LSPipe]>],
531 InstrItinData<IIC_VLD3lnu, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
532 InstrStage<5, [A8_NLSPipe], 0>,
533 InstrStage<5, [A8_LSPipe]>],
534 [4, 4, 5, 2, 1, 1, 1, 1, 1, 2]>,
537 InstrItinData<IIC_VLD4, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
538 InstrStage<4, [A8_NLSPipe], 0>,
539 InstrStage<4, [A8_LSPipe]>],
543 InstrItinData<IIC_VLD4ln, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
544 InstrStage<5, [A8_NLSPipe], 0>,
545 InstrStage<5, [A8_LSPipe]>],
546 [4, 4, 5, 5, 1, 1, 1, 1, 2, 2]>,
549 InstrItinData<IIC_VLD4u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
550 InstrStage<4, [A8_NLSPipe], 0>,
551 InstrStage<4, [A8_LSPipe]>],
555 InstrItinData<IIC_VLD4lnu, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
556 InstrStage<5, [A8_NLSPipe], 0>,
557 InstrStage<5, [A8_LSPipe]>],
558 [4, 4, 5, 5, 2, 1, 1, 1, 1, 1, 2, 2]>,
561 InstrItinData<IIC_VST1, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
562 InstrStage<2, [A8_NLSPipe], 0>,
563 InstrStage<2, [A8_LSPipe]>],
567 InstrItinData<IIC_VST1x2, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
568 InstrStage<2, [A8_NLSPipe], 0>,
569 InstrStage<2, [A8_LSPipe]>],
573 InstrItinData<IIC_VST1x3, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
574 InstrStage<3, [A8_NLSPipe], 0>,
575 InstrStage<3, [A8_LSPipe]>],
579 InstrItinData<IIC_VST1x4, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
580 InstrStage<3, [A8_NLSPipe], 0>,
581 InstrStage<3, [A8_LSPipe]>],
585 InstrItinData<IIC_VST1u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
586 InstrStage<2, [A8_NLSPipe], 0>,
587 InstrStage<2, [A8_LSPipe]>],
591 InstrItinData<IIC_VST1x2u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
592 InstrStage<2, [A8_NLSPipe], 0>,
593 InstrStage<2, [A8_LSPipe]>],
597 InstrItinData<IIC_VST1x3u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
598 InstrStage<3, [A8_NLSPipe], 0>,
599 InstrStage<3, [A8_LSPipe]>],
600 [2, 1, 1, 1, 1, 1, 2]>,
603 InstrItinData<IIC_VST1x4u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
604 InstrStage<3, [A8_NLSPipe], 0>,
605 InstrStage<3, [A8_LSPipe]>],
606 [2, 1, 1, 1, 1, 1, 2, 2]>,
609 InstrItinData<IIC_VST1ln, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
610 InstrStage<2, [A8_NLSPipe], 1>,
611 InstrStage<2, [A8_LSPipe]>],
615 InstrItinData<IIC_VST1lnu, [InstrStage<1, [A8_Pipe0, A8_Pipe1]>,
616 InstrStage<2, [A8_NLSPipe], 1>,
617 InstrStage<2, [A8_LSPipe]>],
621 InstrItinData<IIC_VST2, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
622 InstrStage<2, [A8_NLSPipe], 0>,
623 InstrStage<2, [A8_LSPipe]>],
627 InstrItinData<IIC_VST2x2, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
628 InstrStage<4, [A8_NLSPipe], 0>,
629 InstrStage<4, [A8_LSPipe]>],
633 InstrItinData<IIC_VST2u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
634 InstrStage<2, [A8_NLSPipe], 0>,
635 InstrStage<2, [A8_LSPipe]>],
639 InstrItinData<IIC_VST2x2u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
640 InstrStage<4, [A8_NLSPipe], 0>,
641 InstrStage<4, [A8_LSPipe]>],
642 [2, 1, 1, 1, 1, 1, 2, 2]>,
645 InstrItinData<IIC_VST2ln, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
646 InstrStage<2, [A8_NLSPipe], 0>,
647 InstrStage<2, [A8_LSPipe]>],
651 InstrItinData<IIC_VST2lnu, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
652 InstrStage<2, [A8_NLSPipe], 0>,
653 InstrStage<2, [A8_LSPipe]>],
657 InstrItinData<IIC_VST3, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
658 InstrStage<3, [A8_NLSPipe], 0>,
659 InstrStage<3, [A8_LSPipe]>],
663 InstrItinData<IIC_VST3u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
664 InstrStage<3, [A8_NLSPipe], 0>,
665 InstrStage<3, [A8_LSPipe]>],
666 [2, 1, 1, 1, 1, 1, 2]>,
669 InstrItinData<IIC_VST3ln, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
670 InstrStage<3, [A8_NLSPipe], 0>,
671 InstrStage<3, [A8_LSPipe]>],
675 InstrItinData<IIC_VST3lnu, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
676 InstrStage<3, [A8_NLSPipe], 0>,
677 InstrStage<3, [A8_LSPipe]>],
678 [2, 1, 1, 1, 1, 1, 2]>,
681 InstrItinData<IIC_VST4, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
682 InstrStage<4, [A8_NLSPipe], 0>,
683 InstrStage<4, [A8_LSPipe]>],
687 InstrItinData<IIC_VST4u, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
688 InstrStage<4, [A8_NLSPipe], 0>,
689 InstrStage<4, [A8_LSPipe]>],
690 [2, 1, 1, 1, 1, 1, 2, 2]>,
693 InstrItinData<IIC_VST4ln, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
694 InstrStage<4, [A8_NLSPipe], 0>,
695 InstrStage<4, [A8_LSPipe]>],
699 InstrItinData<IIC_VST4lnu, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
700 InstrStage<4, [A8_NLSPipe], 0>,
701 InstrStage<4, [A8_LSPipe]>],
702 [2, 1, 1, 1, 1, 1, 2, 2]>,
704 // Double-register FP Unary
705 InstrItinData<IIC_VUNAD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
706 InstrStage<1, [A8_NPipe]>], [5, 2]>,
708 // Quad-register FP Unary
709 // Result written in N5, but that is relative to the last cycle of multicycle,
710 // so we use 6 for those cases
711 InstrItinData<IIC_VUNAQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
712 InstrStage<2, [A8_NPipe]>], [6, 2]>,
714 // Double-register FP Binary
715 InstrItinData<IIC_VBIND, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
716 InstrStage<1, [A8_NPipe]>], [5, 2, 2]>,
719 InstrItinData<IIC_VPBIND, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
720 InstrStage<1, [A8_NPipe]>], [5, 2, 2]>,
722 // Double-register FP VMUL
723 InstrItinData<IIC_VFMULD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
724 InstrStage<1, [A8_NPipe]>], [5, 2, 1]>,
727 // Quad-register FP Binary
728 // Result written in N5, but that is relative to the last cycle of multicycle,
729 // so we use 6 for those cases
730 InstrItinData<IIC_VBINQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
731 InstrStage<2, [A8_NPipe]>], [6, 2, 2]>,
733 // Quad-register FP VMUL
734 InstrItinData<IIC_VFMULQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
735 InstrStage<1, [A8_NPipe]>], [6, 2, 1]>,
738 InstrItinData<IIC_VMOV, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
739 InstrStage<1, [A8_NPipe]>], [1, 1]>,
742 InstrItinData<IIC_VMOVImm, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
743 InstrStage<1, [A8_NPipe]>], [3]>,
745 // Double-register Permute Move
746 InstrItinData<IIC_VMOVD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
747 InstrStage<1, [A8_NLSPipe]>], [2, 1]>,
749 // Quad-register Permute Move
750 // Result written in N2, but that is relative to the last cycle of multicycle,
751 // so we use 3 for those cases
752 InstrItinData<IIC_VMOVQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
753 InstrStage<2, [A8_NLSPipe]>], [3, 1]>,
755 // Integer to Single-precision Move
756 InstrItinData<IIC_VMOVIS , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
757 InstrStage<1, [A8_NLSPipe]>], [2, 1]>,
759 // Integer to Double-precision Move
760 InstrItinData<IIC_VMOVID , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
761 InstrStage<1, [A8_NLSPipe]>], [2, 1, 1]>,
763 // Single-precision to Integer Move
764 InstrItinData<IIC_VMOVSI , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
765 InstrStage<1, [A8_NLSPipe]>], [20, 1]>,
767 // Double-precision to Integer Move
768 InstrItinData<IIC_VMOVDI , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
769 InstrStage<1, [A8_NLSPipe]>], [20, 20, 1]>,
771 // Integer to Lane Move
772 InstrItinData<IIC_VMOVISL , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
773 InstrStage<2, [A8_NLSPipe]>], [3, 1, 1]>,
775 // Vector narrow move
776 InstrItinData<IIC_VMOVN , [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
777 InstrStage<1, [A8_NPipe]>], [2, 1]>,
779 // Double-register Permute
780 InstrItinData<IIC_VPERMD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
781 InstrStage<1, [A8_NLSPipe]>], [2, 2, 1, 1]>,
783 // Quad-register Permute
784 // Result written in N2, but that is relative to the last cycle of multicycle,
785 // so we use 3 for those cases
786 InstrItinData<IIC_VPERMQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
787 InstrStage<2, [A8_NLSPipe]>], [3, 3, 1, 1]>,
789 // Quad-register Permute (3 cycle issue)
790 // Result written in N2, but that is relative to the last cycle of multicycle,
791 // so we use 4 for those cases
792 InstrItinData<IIC_VPERMQ3, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
793 InstrStage<1, [A8_NLSPipe]>,
794 InstrStage<1, [A8_NPipe], 0>,
795 InstrStage<2, [A8_NLSPipe]>], [4, 4, 1, 1]>,
797 // Double-register FP Multiple-Accumulate
798 InstrItinData<IIC_VMACD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
799 InstrStage<1, [A8_NPipe]>], [9, 3, 2, 2]>,
801 // Quad-register FP Multiple-Accumulate
802 // Result written in N9, but that is relative to the last cycle of multicycle,
803 // so we use 10 for those cases
804 InstrItinData<IIC_VMACQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
805 InstrStage<2, [A8_NPipe]>], [10, 3, 2, 2]>,
807 // Double-register Reciprical Step
808 InstrItinData<IIC_VRECSD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
809 InstrStage<1, [A8_NPipe]>], [9, 2, 2]>,
811 // Quad-register Reciprical Step
812 InstrItinData<IIC_VRECSQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
813 InstrStage<2, [A8_NPipe]>], [10, 2, 2]>,
815 // Double-register Integer Count
816 InstrItinData<IIC_VCNTiD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
817 InstrStage<1, [A8_NPipe]>], [3, 2, 2]>,
819 // Quad-register Integer Count
820 // Result written in N3, but that is relative to the last cycle of multicycle,
821 // so we use 4 for those cases
822 InstrItinData<IIC_VCNTiQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
823 InstrStage<2, [A8_NPipe]>], [4, 2, 2]>,
825 // Double-register Integer Unary
826 InstrItinData<IIC_VUNAiD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
827 InstrStage<1, [A8_NPipe]>], [4, 2]>,
829 // Quad-register Integer Unary
830 InstrItinData<IIC_VUNAiQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
831 InstrStage<1, [A8_NPipe]>], [4, 2]>,
833 // Double-register Integer Q-Unary
834 InstrItinData<IIC_VQUNAiD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
835 InstrStage<1, [A8_NPipe]>], [4, 1]>,
837 // Quad-register Integer CountQ-Unary
838 InstrItinData<IIC_VQUNAiQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
839 InstrStage<1, [A8_NPipe]>], [4, 1]>,
841 // Double-register Integer Binary
842 InstrItinData<IIC_VBINiD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
843 InstrStage<1, [A8_NPipe]>], [3, 2, 2]>,
845 // Quad-register Integer Binary
846 InstrItinData<IIC_VBINiQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
847 InstrStage<1, [A8_NPipe]>], [3, 2, 2]>,
849 // Double-register Integer Binary (4 cycle)
850 InstrItinData<IIC_VBINi4D, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
851 InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
853 // Quad-register Integer Binary (4 cycle)
854 InstrItinData<IIC_VBINi4Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
855 InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
858 // Double-register Integer Subtract
859 InstrItinData<IIC_VSUBiD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
860 InstrStage<1, [A8_NPipe]>], [3, 2, 1]>,
862 // Quad-register Integer Subtract
863 InstrItinData<IIC_VSUBiQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
864 InstrStage<1, [A8_NPipe]>], [3, 2, 1]>,
866 // Double-register Integer Subtract
867 InstrItinData<IIC_VSUBi4D, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
868 InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
870 // Quad-register Integer Subtract
871 InstrItinData<IIC_VSUBi4Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
872 InstrStage<1, [A8_NPipe]>], [4, 2, 1]>,
874 // Double-register Integer Shift
875 InstrItinData<IIC_VSHLiD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
876 InstrStage<1, [A8_NPipe]>], [3, 1, 1]>,
878 // Quad-register Integer Shift
879 InstrItinData<IIC_VSHLiQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
880 InstrStage<2, [A8_NPipe]>], [4, 1, 1]>,
882 // Double-register Integer Shift (4 cycle)
883 InstrItinData<IIC_VSHLi4D, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
884 InstrStage<1, [A8_NPipe]>], [4, 1, 1]>,
886 // Quad-register Integer Shift (4 cycle)
887 InstrItinData<IIC_VSHLi4Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
888 InstrStage<2, [A8_NPipe]>], [5, 1, 1]>,
890 // Double-register Integer Pair Add Long
891 InstrItinData<IIC_VPALiD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
892 InstrStage<1, [A8_NPipe]>], [6, 3, 1]>,
894 // Quad-register Integer Pair Add Long
895 InstrItinData<IIC_VPALiQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
896 InstrStage<2, [A8_NPipe]>], [7, 3, 1]>,
898 // Double-register Absolute Difference and Accumulate
899 InstrItinData<IIC_VABAD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
900 InstrStage<1, [A8_NPipe]>], [6, 3, 2, 1]>,
902 // Quad-register Absolute Difference and Accumulate
903 InstrItinData<IIC_VABAQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
904 InstrStage<2, [A8_NPipe]>], [6, 3, 2, 1]>,
907 // Double-register Integer Multiply (.8, .16)
908 InstrItinData<IIC_VMULi16D, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
909 InstrStage<1, [A8_NPipe]>], [6, 2, 2]>,
911 // Double-register Integer Multiply (.32)
912 InstrItinData<IIC_VMULi32D, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
913 InstrStage<2, [A8_NPipe]>], [7, 2, 1]>,
915 // Quad-register Integer Multiply (.8, .16)
916 InstrItinData<IIC_VMULi16Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
917 InstrStage<2, [A8_NPipe]>], [7, 2, 2]>,
919 // Quad-register Integer Multiply (.32)
920 InstrItinData<IIC_VMULi32Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
921 InstrStage<1, [A8_NPipe]>,
922 InstrStage<2, [A8_NLSPipe], 0>,
923 InstrStage<3, [A8_NPipe]>], [9, 2, 1]>,
925 // Double-register Integer Multiply-Accumulate (.8, .16)
926 InstrItinData<IIC_VMACi16D, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
927 InstrStage<1, [A8_NPipe]>], [6, 3, 2, 2]>,
929 // Double-register Integer Multiply-Accumulate (.32)
930 InstrItinData<IIC_VMACi32D, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
931 InstrStage<2, [A8_NPipe]>], [7, 3, 2, 1]>,
933 // Quad-register Integer Multiply-Accumulate (.8, .16)
934 InstrItinData<IIC_VMACi16Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
935 InstrStage<2, [A8_NPipe]>], [7, 3, 2, 2]>,
937 // Quad-register Integer Multiply-Accumulate (.32)
938 InstrItinData<IIC_VMACi32Q, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
939 InstrStage<1, [A8_NPipe]>,
940 InstrStage<2, [A8_NLSPipe], 0>,
941 InstrStage<3, [A8_NPipe]>], [9, 3, 2, 1]>,
943 // Double-register VEXT
944 InstrItinData<IIC_VEXTD, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
945 InstrStage<1, [A8_NLSPipe]>], [2, 1, 1]>,
947 // Quad-register VEXT
948 InstrItinData<IIC_VEXTQ, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
949 InstrStage<2, [A8_NLSPipe]>], [3, 1, 1]>,
952 InstrItinData<IIC_VTB1, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
953 InstrStage<2, [A8_NLSPipe]>], [3, 2, 1]>,
954 InstrItinData<IIC_VTB2, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
955 InstrStage<2, [A8_NLSPipe]>], [3, 2, 2, 1]>,
956 InstrItinData<IIC_VTB3, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
957 InstrStage<1, [A8_NLSPipe]>,
958 InstrStage<1, [A8_NPipe], 0>,
959 InstrStage<2, [A8_NLSPipe]>], [4, 2, 2, 3, 1]>,
960 InstrItinData<IIC_VTB4, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
961 InstrStage<1, [A8_NLSPipe]>,
962 InstrStage<1, [A8_NPipe], 0>,
963 InstrStage<2, [A8_NLSPipe]>],[4, 2, 2, 3, 3, 1]>,
966 InstrItinData<IIC_VTBX1, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
967 InstrStage<2, [A8_NLSPipe]>], [3, 1, 2, 1]>,
968 InstrItinData<IIC_VTBX2, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
969 InstrStage<2, [A8_NLSPipe]>], [3, 1, 2, 2, 1]>,
970 InstrItinData<IIC_VTBX3, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
971 InstrStage<1, [A8_NLSPipe]>,
972 InstrStage<1, [A8_NPipe], 0>,
973 InstrStage<2, [A8_NLSPipe]>],[4, 1, 2, 2, 3, 1]>,
974 InstrItinData<IIC_VTBX4, [InstrStage<1, [A8_Pipe0, A8_Pipe1], 0>,
975 InstrStage<1, [A8_NLSPipe]>,
976 InstrStage<1, [A8_NPipe], 0>,
977 InstrStage<2, [A8_NLSPipe]>], [4, 1, 2, 2, 3, 3, 1]>