1 # Rewrite of SVP64 for OpenPower ISA v3.1
5 The plan is to create an encoding for SVP64, then to create an encoding for
6 SVP48, then to reorganize them both to improve field overlap, reducing the
7 amount of decoder hardware necessary.
9 All bit numbers are in MSB0 form (the bits are numbered from 0 at the MSB and
10 counting up as you move to the LSB end). All bit ranges are inclusive (so
11 `4:6` means bits 4, 5, and 6).
13 64-bit instructions are split into two 32-bit words, the prefix and the suffix. The prefix always comes before the suffix in PC order.
15 ## Definition of Reserved in this spec.
17 For the new fields added in SVP64, instructions that have any of their fields set to a reserved value must cause an illegal instruction trap, to allow emulation of future instruction sets.
19 This is unlike OpenPower ISA v3.1, which doesn't require a CPU to trap.
21 ## Remapped Encoding (`RM[0:23]`)
23 To allow relatively easy remapping of which portions of the Prefix Opcode Map
24 are used for SVP64 without needing to rewrite a large portion of the SVP64
25 spec, a mapping is defined from the OpenPower v3.1 prefix bits to a new 24-bit
26 Remapped Encoding denoted `RM[0]` at the MSB to `RM[23]` at the LSB.
28 The mapping from the OpenPower v3.1 prefix bits to the Remapped Encoding is
29 defined in the Prefix Fields section.
31 ## Remapped Encoding Fields
33 Shows all fields in the Remapped Encoding `RM[0:23]` for all instruction variants. There are two categories: Single and Twin Predication.
35 ### Single Predication (N(src) > 1)
38 | Remapped Encoding Field Name | Field bits | Description |
39 |------------------------------|------------|---------------------------------------------------------------------------|
40 | MASK_KIND | `0` | Execution Mask Kind |
41 | MASK | `1:3` | Execution Mask |
42 | ELWIDTH | `4:5` | Element Width |
43 | SUBVL | `6:7` | Sub-vector length |
44 | Rdest_EXTRA | `8:10` | extra bits for Rdest (Uses R\*_EXTRA Encoding) |
45 | Rsrc1_EXTRA | `11:13` | extra bits for Rsrc1 (Uses R\*_EXTRA Encoding) |
46 | Rsrc2_EXTRA | `14:16` | extra bits for Rsrc2 (Uses R\*_EXTRA Encoding) |
47 | Rsrc3_EXTRA | `17:18` | extra bits for Rsrc3 (Uses 2-bit R\*_EXTRA Encoding) |
48 | MODE | `19:23` | see [[discussion]] |
50 ### Twin Predication (src=1, dest=1)
52 | Remapped Encoding Field Name | Field bits | Description |
53 |------------------------------|------------|---------------------------------------------------------------------------|
54 | MASK_KIND | `0` | Execution Mask Kind |
55 | MASK | `1:3` | Execution Mask |
56 | ELWIDTH | `4:5` | Element Width |
57 | SUBVL | `6:7` | Sub-vector length |
58 | Rdest_EXTRA | `8:10` | extra bits for Rdest (Uses R\*_EXTRA Encoding) |
59 | Rsrc1_EXTRA | `11:13` | extra bits for Rsrc1 (Uses R\*_EXTRA Encoding) |
60 | MASK_SRC | `14:16` | Execution Mask for Source (only on instructions with twin-predication) |
61 | ELWIDTH_SRC | `17:18` | Element Width for Source (only on instructions with twin-predication) |
62 | MODE | `19:23` | see [[discussion]] |
64 note in [[discussion]]: TODO, evaluate if 2nd SUBVL should be added. conclusion: no. 2nd SUBVL makes no sense except for mv, and that is covered by [[mv.vec]]
68 (**TODO: 2-bit version of the table, just like in the original SVPrefix. This is important, to save bits on 4-operand instructions such as fmadd**)
70 In the following table, `<N>` denotes the value of the corresponding register field in the SVP64 suffix word.
72 (**Jacob: these tables are not in the slightest bit understandable due to the use of register names that are impossible to interpret clearly**)
74 | R\*_EXTRA | Vector/Scalar<br/>Mode | CR Register | Int/FP<br/>Register |
75 |-----------|------------------------|---------------|---------------------|
76 | 000 | Scalar | `SVCR<N>_000` | `SV[F]R<N>_00` |
77 | 001 | Scalar | `SVCR<N>_010` | `SV[F]R<N>_01` |
78 | 010 | Scalar | `SVCR<N>_100` | `SV[F]R<N>_10` |
79 | 011 | Scalar | `SVCR<N>_110` | `SV[F]R<N>_11` |
80 | 100 | Vector | `SVCR<N>_000` | `SV[F]R<N>_00` |
81 | 101 | Vector | `SVCR<N>_010` | `SV[F]R<N>_01` |
82 | 110 | Vector | `SVCR<N>_100` | `SV[F]R<N>_10` |
83 | 111 | Vector | `SVCR<N>_110` | `SV[F]R<N>_11` |
87 | Instruction Kind | ELWIDTH Value | Mnemonic | Description |
88 |------------------|---------------|---------------------------|-------------------------------------------------------------------------------------|
89 | Integer | 00 | `ELWIDTH=b` | Byte: 8-bit integer |
90 | Integer | 01 | `ELWIDTH=h` | Halfword: 16-bit integer |
91 | Integer | 10 | `ELWIDTH=w` | Word: 32-bit integer |
92 | Integer | 11 | `ELWIDTH=d` | Doubleword: 64-bit integer |
93 | FP | 00 | `ELWIDTH=bf16` (Reserved) | Reserved for [`bf16`](https://en.wikipedia.org/wiki/Bfloat16_floating-point_format) |
94 | FP | 01 | `ELWIDTH=f16` | 16-bit IEEE 754 Half floating-point |
95 | FP | 10 | `ELWIDTH=f32` | 32-bit IEEE 754 Single floating-point |
96 | FP | 11 | `ELWIDTH=f64` | 64-bit IEEE 754 Double floating-point |
100 | SUBVL Value | Mnemonic | Description |
101 |-------------|---------------------|------------------------|
102 | 00 | `SUBVL=4` | Sub-vector length of 4 |
103 | 01 | `SUBVL=1` (default) | Sub-vector length of 1 |
104 | 10 | `SUBVL=2` | Sub-vector length of 2 |
105 | 11 | `SUBVL=3` | Sub-vector length of 3 |
107 ## MASK/MASK_SRC & MASK_KIND Encoding
109 One bit (`MASKMODE`) indicates the mode: CR or Int predication. The two types may not be mixed.
111 | MASK_KIND Value | Description |
112 |-----------------|------------------------------------------------------|
113 | 0 | MASK/MASK_SRC are encoded using Integer Predication |
114 | 1 | MASK/MASK_SRC are encoded using CR-based Predication |
116 Integer Twin predication has a second set if 3 bits that uses the same encoding thus allowing either the same register (r3 or r10) to be used for both src and dest, or different regs (one for src, one for dest).
118 Likewise CR based twin predication has a second set of 3 bits, allowing a different test to be applied.
120 ### Integer Predication (MASK_KIND=0)
122 When the predicate mode bit is zero the 3 bits are interpreted as below.
123 Twin predication has an identical 3 bit field similarly encoded.
125 | MASK/MASK_SRC<br/>Value | Mnemonic | Description |
126 |-------------------------|----------|--------------------------------------------------------|
127 | 000 | ALWAYS | Operation is not masked (mask set to all 1s) |
128 | 001 | 1 << R3 | Element `i` is enabled if `i == R3` |
129 | 010 | R3 | Element `i` is enabled if `R3 & (1 << i)` is non-zero |
130 | 011 | ~R3 | Element `i` is enabled if `R3 & (1 << i)` is zero |
131 | 100 | R10 | Element `i` is enabled if `R10 & (1 << i)` is non-zero |
132 | 101 | ~R10 | Element `i` is enabled if `R10 & (1 << i)` is zero |
133 | 110 | R30 | Element `i` is enabled if `R30 & (1 << i)` is non-zero |
134 | 111 | ~R30 | Element `i` is enabled if `R30 & (1 << i)` is zero |
136 ### CR-based Predication (MASK_KIND=1)
138 When the predicate mode bit is one the 3 bits are interpreted as below. Twin predication has an identical 3 bit field similarly encoded
140 | MASK/MASK_SRC<br/>Value | Mnemonic | Description |
141 |-------------------------|----------|-------------------------------------------------|
142 | 000 | lt | Element `i` is enabled if `CR[6+i].LT` is set |
143 | 001 | nl/ge | Element `i` is enabled if `CR[6+i].LT` is clear |
144 | 010 | gt | Element `i` is enabled if `CR[6+i].GT` is set |
145 | 011 | ng/le | Element `i` is enabled if `CR[6+i].GT` is clear |
146 | 100 | eq | Element `i` is enabled if `CR[6+i].EQ` is set |
147 | 101 | ne | Element `i` is enabled if `CR[6+i].EQ` is clear |
148 | 110 | so/un | Element `i` is enabled if `CR[6+i].FU` is set |
149 | 111 | ns/nu | Element `i` is enabled if `CR[6+i].FU` is clear |
151 CR based predication. TODO: select alternate CR for twin predication? see [[discussion]] Overlap of the two CR based predicates must be taken into account, so the starting point for one of them must be suitably high, or accept that for twin predication VL must not exceed the range where overlap will occur, *or* that they use the same starting point but select different *bits* of the same CRs
154 ## Prefix Opcode Map (64-bit instruction encoding) (prefix bits 6:11)
156 (shows both PowerISA v3.1 instructions as well as new SVP instructions; empty spaces are yet-to-be-allocated Illegal Instructions)
158 | bits 6:11 | ---000 | ---001 | ---010 | ---011 | ---100 | ---101 | ---110 | ---111 |
159 |-----------|----------|------------|----------|----------|----------|----------|----------|----------|
160 | 000--- | 8LS-form | 8LS-form | 8LS-form | 8LS-form | 8LS-form | 8LS-form | 8LS-form | 8LS-form |
161 | 001--- | | | | | | | | |
162 | 010--- | 8RR-form | | | | SVP64 | SVP64 | SVP64 | SVP64 |
163 | 011--- | | | | | SVP64 | SVP64 | SVP64 | SVP64 |
164 | 100--- | MLS-form | MLS-form | MLS-form | MLS-form | MLS-form | MLS-form | MLS-form | MLS-form |
165 | 101--- | | | | | | | | |
166 | 110--- | MRR-form | | | | SVP64 | SVP64 | SVP64 | SVP64 |
167 | 111--- | | MMIRR-form | | | SVP64 | SVP64 | SVP64 | SVP64 |
171 | Prefix Field Name | Field bits | Constant Value | Description |
172 |---------------------|------------|----------------|--------------------------------------------|
173 | PO (Primary Opcode) | `0:5` | `1` | Indicates this is a 64-bit instruction |
174 | `RM[0]` | `6` | | Bit 0 of the Remapped Encoding |
175 | SVP64_7 | `7` | `1` | Indicates this is a SVP64 instruction |
176 | `RM[1]` | `8` | | Bit 1 of the Remapped Encoding |
177 | SVP64_9 | `9` | `1` | Indicates this is a SVP64 instruction |
178 | `RM[2:23]` | `10:31` | | Bits 2 through 23 of the Remapped Encoding |
182 This is a novel concept that allows predication to be applied to a single source and a single dest register. The following types of traditional Vector operations may be encoded with it, *without requiring explicit opcodes to do so*
184 * VSPLAT (a single scalar distributed across a vector)
185 * VEXTRACT (like LLVM IR [`extractelement`](https://releases.llvm.org/11.0.0/docs/LangRef.html#extractelement-instruction))
186 * VINSERT (like LLVM IR [`insertelement`](https://releases.llvm.org/11.0.0/docs/LangRef.html#insertelement-instruction))
187 * VCOMPRESS (like LLVM IR [`llvm.masked.compressstore.*`](https://releases.llvm.org/11.0.0/docs/LangRef.html#llvm-masked-compressstore-intrinsics))
188 * VEXPAND (like LLVM IR [`llvm.masked.expandload.*`](https://releases.llvm.org/11.0.0/docs/LangRef.html#llvm-masked-expandload-intrinsics))
190 Those patterns (and more) may be applied to:
192 * mv (the usual way that V\* operations are created)
193 * exts\* sign-extension
194 * rwlinm and other RS-RA shift operations
195 * LD and ST (treating AGEN as one source)
196 * FP fclass, fsgn, fneg, fabs, fcvt, frecip, fsqrt etc.
197 * Condition Register ops mfcr, mtcr and other similar
199 This is a huge list that creates extremely powerful combinations, particularly given that one of the predicate options is `(1<<r3)`
201 Additional unusual capabilities of Twin Predication include a back-to-back version of VCOMPRESS-VEXPAND which is effectively the ability to do an ordered multiple VINSERT.
205 There are two different encodings: single-predication (typically arithmetic operations i.e. with more than one source register) and twin-predication (one source, one destination). They require different encodings
209 SV Registers are numbered using the notation `SV[F|C]R<N>_<M>` where `<N>` is a decimal integer and `<M>` is a binary integer. Two integers are used to enable future register expansions to add more registers by appending more LSB bits to `<M>`.
211 For all `SV[F|C]R<N>_<M>` registers, the N is the
212 upper bits in decimal and the M is the lower bits in binary, so `SVR5_01` is
213 SV integer register `(5 << 2) + 0b01`, `SVCR6_011` is SV condition register
214 `(6 << 3) + 0b011`, and `SVFR20_10` is SV floating-point register
219 a vectorized 32-bit add:
221 add SVR3_01, SVR6_10, SVR10_00, elwidth=w, subvl=1, mask=lt
225 const size_t start_cr = (6 << 3) + 0b000; // starting at SVCR6_000
226 // pretend for the moment that type-punning actually works in C/C++
227 uint32_t *rt = (uint32_t *)®s[(3 << 2) + 0b01]; // SVR3_01
228 uint32_t *ra = (uint32_t *)®s[(6 << 2) + 0b10]; // SVR6_10
229 uint32_t *rb = (uint32_t *)®s[(10 << 2) + 0b00]; // SVR10_00
230 for(size_t i = 0; i < VL; i++) {
231 if(CRs[(start_cr + i) % 64].lt) {
232 rt[i] = ra[i] + rb[i];
239 add r20, r25, r30, elwidth=64, subvl=1
241 where `r20`, `r25`, and `r30` are standard OpenPower register names.
242 Those names correspond to `SVR20_00`, `SVR25_00`, and `SVR30_00`.
246 const size_t STD_TO_SV_SHIFT = 2; // gets bigger as reg files expand to 256, 512, ... registers
248 VL = 7; // setvli (omitting maxvl here)
250 for(size_t i = 0; i < VL; i++) {
251 regs[(20 << STD_TO_SV_SHIFT) + i] = regs[(25 << STD_TO_SV_SHIFT) + i]
252 + regs[(30 << STD_TO_SV_SHIFT) + i];
255 Standard PowerISA Integer registers are aliased to some of the SV integer registers:
257 (**Jacob these names are impossible to interpret due to them not being sequential numbering and there being no compact algorithm given that shows how they're created. the original SVPrefix was dead easy to understand**)
259 | Integer<br/>Register | SV Integer<br/>Register | Integer<br/>Register | SV Integer<br/>Register | Integer<br/>Register | SV Integer<br/>Register | Integer<br/>Register | SV Integer<br/>Register |
260 |----------------------|-------------------------|----------------------|-------------------------|----------------------|-------------------------|----------------------|-------------------------|
261 | R0 | SVR0_00 | R8 | SVR8_00 | R16 | SVR16_00 | R24 | SVR24_00 |
262 | | SVR0_01 | | SVR8_01 | | SVR16_01 | | SVR24_01 |
263 | | SVR0_10 | | SVR8_10 | | SVR16_10 | | SVR24_10 |
264 | | SVR0_11 | | SVR8_11 | | SVR16_11 | | SVR24_11 |
265 | R1 | SVR1_00 | R9 | SVR9_00 | R17 | SVR17_00 | R25 | SVR25_00 |
266 | | SVR1_01 | | SVR9_01 | | SVR17_01 | | SVR25_01 |
267 | | SVR1_10 | | SVR9_10 | | SVR17_10 | | SVR25_10 |
268 | | SVR1_11 | | SVR9_11 | | SVR17_11 | | SVR25_11 |
269 | R2 | SVR2_00 | R10 | SVR10_00 | R18 | SVR18_00 | R26 | SVR26_00 |
270 | | SVR2_01 | | SVR10_01 | | SVR18_01 | | SVR26_01 |
271 | | SVR2_10 | | SVR10_10 | | SVR18_10 | | SVR26_10 |
272 | | SVR2_11 | | SVR10_11 | | SVR18_11 | | SVR26_11 |
273 | R3 | SVR3_00 | R11 | SVR11_00 | R19 | SVR19_00 | R27 | SVR27_00 |
274 | | SVR3_01 | | SVR11_01 | | SVR19_01 | | SVR27_01 |
275 | | SVR3_10 | | SVR11_10 | | SVR19_10 | | SVR27_10 |
276 | | SVR3_11 | | SVR11_11 | | SVR19_11 | | SVR27_11 |
277 | R4 | SVR4_00 | R12 | SVR12_00 | R20 | SVR20_00 | R28 | SVR28_00 |
278 | | SVR4_01 | | SVR12_01 | | SVR20_01 | | SVR28_01 |
279 | | SVR4_10 | | SVR12_10 | | SVR20_10 | | SVR28_10 |
280 | | SVR4_11 | | SVR12_11 | | SVR20_11 | | SVR28_11 |
281 | R5 | SVR5_00 | R13 | SVR13_00 | R21 | SVR21_00 | R29 | SVR29_00 |
282 | | SVR5_01 | | SVR13_01 | | SVR21_01 | | SVR29_01 |
283 | | SVR5_10 | | SVR13_10 | | SVR21_10 | | SVR29_10 |
284 | | SVR5_11 | | SVR13_11 | | SVR21_11 | | SVR29_11 |
285 | R6 | SVR6_00 | R14 | SVR14_00 | R22 | SVR22_00 | R30 | SVR30_00 |
286 | | SVR6_01 | | SVR14_01 | | SVR22_01 | | SVR30_01 |
287 | | SVR6_10 | | SVR14_10 | | SVR22_10 | | SVR30_10 |
288 | | SVR6_11 | | SVR14_11 | | SVR22_11 | | SVR30_11 |
289 | R7 | SVR7_00 | R15 | SVR15_00 | R23 | SVR23_00 | R31 | SVR31_00 |
290 | | SVR7_01 | | SVR15_01 | | SVR23_01 | | SVR31_01 |
291 | | SVR7_10 | | SVR15_10 | | SVR23_10 | | SVR31_10 |
292 | | SVR7_11 | | SVR15_11 | | SVR23_11 | | SVR31_11 |
294 ## Floating-Point Registers
296 Standard PowerISA floating-point and VSX registers are aliased to some of the SV floating-point registers:
298 (**Jacob these names are impossible to interpret due to them not being sequential numbering and there being no compact algorithm given that shows how they're created. the original SVPrefix was dead easy to understand**)
300 | FP<br/>Register | VSX Register | SV FP<br/>Register | FP<br/>Register | VSX Register | SV FP<br/>Register |
301 |-----------------|-----------------------|--------------------|-----------------|-----------------------|--------------------|
302 | FPR\[0\] | VSR\[0\]\.dword\[0\] | SVFR0\_00 | FPR\[16\] | VSR\[16\]\.dword\[0\] | SVFR16\_00 |
303 | | VSR\[0\]\.dword\[1\] | SVFR0\_01 | | VSR\[16\]\.dword\[1\] | SVFR16\_01 |
304 | | VSR\[32\]\.dword\[0\] | SVFR0\_10 | | VSR\[48\]\.dword\[0\] | SVFR16\_10 |
305 | | VSR\[32\]\.dword\[1\] | SVFR0\_11 | | VSR\[48\]\.dword\[1\] | SVFR16\_11 |
306 | FPR\[1\] | VSR\[1\]\.dword\[0\] | SVFR1\_00 | FPR\[17\] | VSR\[17\]\.dword\[0\] | SVFR17\_00 |
307 | | VSR\[1\]\.dword\[1\] | SVFR1\_01 | | VSR\[17\]\.dword\[1\] | SVFR17\_01 |
308 | | VSR\[33\]\.dword\[0\] | SVFR1\_10 | | VSR\[49\]\.dword\[0\] | SVFR17\_10 |
309 | | VSR\[33\]\.dword\[1\] | SVFR1\_11 | | VSR\[49\]\.dword\[1\] | SVFR17\_11 |
310 | FPR\[2\] | VSR\[2\]\.dword\[0\] | SVFR2\_00 | FPR\[18\] | VSR\[18\]\.dword\[0\] | SVFR18\_00 |
311 | | VSR\[2\]\.dword\[1\] | SVFR2\_01 | | VSR\[18\]\.dword\[1\] | SVFR18\_01 |
312 | | VSR\[34\]\.dword\[0\] | SVFR2\_10 | | VSR\[50\]\.dword\[0\] | SVFR18\_10 |
313 | | VSR\[34\]\.dword\[1\] | SVFR2\_11 | | VSR\[50\]\.dword\[1\] | SVFR18\_11 |
314 | FPR\[3\] | VSR\[3\]\.dword\[0\] | SVFR3\_00 | FPR\[19\] | VSR\[19\]\.dword\[0\] | SVFR19\_00 |
315 | | VSR\[3\]\.dword\[1\] | SVFR3\_01 | | VSR\[19\]\.dword\[1\] | SVFR19\_01 |
316 | | VSR\[35\]\.dword\[0\] | SVFR3\_10 | | VSR\[51\]\.dword\[0\] | SVFR19\_10 |
317 | | VSR\[35\]\.dword\[1\] | SVFR3\_11 | | VSR\[51\]\.dword\[1\] | SVFR19\_11 |
318 | FPR\[4\] | VSR\[4\]\.dword\[0\] | SVFR4\_00 | FPR\[20\] | VSR\[20\]\.dword\[0\] | SVFR20\_00 |
319 | | VSR\[4\]\.dword\[1\] | SVFR4\_01 | | VSR\[20\]\.dword\[1\] | SVFR20\_01 |
320 | | VSR\[36\]\.dword\[0\] | SVFR4\_10 | | VSR\[52\]\.dword\[0\] | SVFR20\_10 |
321 | | VSR\[36\]\.dword\[1\] | SVFR4\_11 | | VSR\[52\]\.dword\[1\] | SVFR20\_11 |
322 | FPR\[5\] | VSR\[5\]\.dword\[0\] | SVFR5\_00 | FPR\[21\] | VSR\[21\]\.dword\[0\] | SVFR21\_00 |
323 | | VSR\[5\]\.dword\[1\] | SVFR5\_01 | | VSR\[21\]\.dword\[1\] | SVFR21\_01 |
324 | | VSR\[37\]\.dword\[0\] | SVFR5\_10 | | VSR\[53\]\.dword\[0\] | SVFR21\_10 |
325 | | VSR\[37\]\.dword\[1\] | SVFR5\_11 | | VSR\[53\]\.dword\[1\] | SVFR21\_11 |
326 | FPR\[6\] | VSR\[6\]\.dword\[0\] | SVFR6\_00 | FPR\[22\] | VSR\[22\]\.dword\[0\] | SVFR22\_00 |
327 | | VSR\[6\]\.dword\[1\] | SVFR6\_01 | | VSR\[22\]\.dword\[1\] | SVFR22\_01 |
328 | | VSR\[38\]\.dword\[0\] | SVFR6\_10 | | VSR\[54\]\.dword\[0\] | SVFR22\_10 |
329 | | VSR\[38\]\.dword\[1\] | SVFR6\_11 | | VSR\[54\]\.dword\[1\] | SVFR22\_11 |
330 | FPR\[7\] | VSR\[7\]\.dword\[0\] | SVFR7\_00 | FPR\[23\] | VSR\[23\]\.dword\[0\] | SVFR23\_00 |
331 | | VSR\[7\]\.dword\[1\] | SVFR7\_01 | | VSR\[23\]\.dword\[1\] | SVFR23\_01 |
332 | | VSR\[39\]\.dword\[0\] | SVFR7\_10 | | VSR\[55\]\.dword\[0\] | SVFR23\_10 |
333 | | VSR\[39\]\.dword\[1\] | SVFR7\_11 | | VSR\[55\]\.dword\[1\] | SVFR23\_11 |
334 | FPR\[8\] | VSR\[8\]\.dword\[0\] | SVFR8\_00 | FPR\[24\] | VSR\[24\]\.dword\[0\] | SVFR24\_00 |
335 | | VSR\[8\]\.dword\[1\] | SVFR8\_01 | | VSR\[24\]\.dword\[1\] | SVFR24\_01 |
336 | | VSR\[40\]\.dword\[0\] | SVFR8\_10 | | VSR\[56\]\.dword\[0\] | SVFR24\_10 |
337 | | VSR\[40\]\.dword\[1\] | SVFR8\_11 | | VSR\[56\]\.dword\[1\] | SVFR24\_11 |
338 | FPR\[9\] | VSR\[9\]\.dword\[0\] | SVFR9\_00 | FPR\[25\] | VSR\[25\]\.dword\[0\] | SVFR25\_00 |
339 | | VSR\[9\]\.dword\[1\] | SVFR9\_01 | | VSR\[25\]\.dword\[1\] | SVFR25\_01 |
340 | | VSR\[41\]\.dword\[0\] | SVFR9\_10 | | VSR\[57\]\.dword\[0\] | SVFR25\_10 |
341 | | VSR\[41\]\.dword\[1\] | SVFR9\_11 | | VSR\[57\]\.dword\[1\] | SVFR25\_11 |
342 | FPR\[10\] | VSR\[10\]\.dword\[0\] | SVFR10\_00 | FPR\[26\] | VSR\[26\]\.dword\[0\] | SVFR26\_00 |
343 | | VSR\[10\]\.dword\[1\] | SVFR10\_01 | | VSR\[26\]\.dword\[1\] | SVFR26\_01 |
344 | | VSR\[42\]\.dword\[0\] | SVFR10\_10 | | VSR\[58\]\.dword\[0\] | SVFR26\_10 |
345 | | VSR\[42\]\.dword\[1\] | SVFR10\_11 | | VSR\[58\]\.dword\[1\] | SVFR26\_11 |
346 | FPR\[11\] | VSR\[11\]\.dword\[0\] | SVFR11\_00 | FPR\[27\] | VSR\[27\]\.dword\[0\] | SVFR27\_00 |
347 | | VSR\[11\]\.dword\[1\] | SVFR11\_01 | | VSR\[27\]\.dword\[1\] | SVFR27\_01 |
348 | | VSR\[43\]\.dword\[0\] | SVFR11\_10 | | VSR\[59\]\.dword\[0\] | SVFR27\_10 |
349 | | VSR\[43\]\.dword\[1\] | SVFR11\_11 | | VSR\[59\]\.dword\[1\] | SVFR27\_11 |
350 | FPR\[12\] | VSR\[12\]\.dword\[0\] | SVFR12\_00 | FPR\[28\] | VSR\[28\]\.dword\[0\] | SVFR28\_00 |
351 | | VSR\[12\]\.dword\[1\] | SVFR12\_01 | | VSR\[28\]\.dword\[1\] | SVFR28\_01 |
352 | | VSR\[44\]\.dword\[0\] | SVFR12\_10 | | VSR\[60\]\.dword\[0\] | SVFR28\_10 |
353 | | VSR\[44\]\.dword\[1\] | SVFR12\_11 | | VSR\[60\]\.dword\[1\] | SVFR28\_11 |
354 | FPR\[13\] | VSR\[13\]\.dword\[0\] | SVFR13\_00 | FPR\[29\] | VSR\[29\]\.dword\[0\] | SVFR29\_00 |
355 | | VSR\[13\]\.dword\[1\] | SVFR13\_01 | | VSR\[29\]\.dword\[1\] | SVFR29\_01 |
356 | | VSR\[45\]\.dword\[0\] | SVFR13\_10 | | VSR\[61\]\.dword\[0\] | SVFR29\_10 |
357 | | VSR\[45\]\.dword\[1\] | SVFR13\_11 | | VSR\[61\]\.dword\[1\] | SVFR29\_11 |
358 | FPR\[14\] | VSR\[14\]\.dword\[0\] | SVFR14\_00 | FPR\[30\] | VSR\[30\]\.dword\[0\] | SVFR30\_00 |
359 | | VSR\[14\]\.dword\[1\] | SVFR14\_01 | | VSR\[30\]\.dword\[1\] | SVFR30\_01 |
360 | | VSR\[46\]\.dword\[0\] | SVFR14\_10 | | VSR\[62\]\.dword\[0\] | SVFR30\_10 |
361 | | VSR\[46\]\.dword\[1\] | SVFR14\_11 | | VSR\[62\]\.dword\[1\] | SVFR30\_11 |
362 | FPR\[15\] | VSR\[15\]\.dword\[0\] | SVFR15\_00 | FPR\[31\] | VSR\[31\]\.dword\[0\] | SVFR31\_00 |
363 | | VSR\[15\]\.dword\[1\] | SVFR15\_01 | | VSR\[31\]\.dword\[1\] | SVFR31\_01 |
364 | | VSR\[47\]\.dword\[0\] | SVFR15\_10 | | VSR\[63\]\.dword\[0\] | SVFR31\_10 |
365 | | VSR\[47\]\.dword\[1\] | SVFR15\_11 | | VSR\[63\]\.dword\[1\] | SVFR31\_11 |
369 ## CR fields as inputs/outputs of vector operations
371 When vectorized, the CR inputs/outputs are read/written to 4-bit CR fields
372 starting from SVCR6_000 and incrementing from there. If SVCR7_111 is reached, the next CR
373 field used wraps around to SVCR0_000, then incrementing from there.
374 (see [[discussion]]. some alternative schemes are described there)
376 SVCR6_000 was chosen to balance avoiding needing to save CR2-CR4 (which are
377 callee-saved) just to use SV vectors with VL <= 61 as well as having the first
378 vector CR field readily accessible to standard CR instructions and branches.
379 Additionally, SVCR6_000 is used as the implicit result of a OpenPower ISA v3.1
380 standard vector (SIMD) instruction with Rc=1.
382 ## Table of CR fields
384 CR[i] is the notation used by the OpenPower spec to refer to CR field #i,
385 so FP instructions with Rc=1 write to CR[1] aka SVCR1_000.
387 There are 3 new SPRs for holding CRs: CR_EXT1, CR_EXT2, and CR_EXT3.
389 The 64 SV CRs are arranged similarly to the way the 128 integer registers are arranged:
391 (**Jacob these names are impossible to interpret due to them not being sequential numbering and there being no compact algorithm given that shows how they're created. the original SVPrefix was dead easy to understand**)
393 | CR<br/>Register | SPR<br/>Field | SV CR<br/>Register | CR<br/>Register | SPR<br/>Field | SV CR<br/>Register |
394 |-----------------|----------------|--------------------|-----------------|----------------|--------------------|
395 | CR[0] | CR[32:35] | SVCR0_000 | CR[4] | CR[48:51] | SVCR4_000 |
396 | | CR_EXT1[32:35] | SVCR0_001 | | CR_EXT1[48:51] | SVCR4_001 |
397 | | CR_EXT2[32:35] | SVCR0_010 | | CR_EXT2[48:51] | SVCR4_010 |
398 | | CR_EXT3[32:35] | SVCR0_011 | | CR_EXT3[48:51] | SVCR4_011 |
399 | *CR[-8]* | CR[0:3] | SVCR0_100 | *CR[-4]* | CR[16:19] | SVCR4_100 |
400 | | CR_EXT1[0:3] | SVCR0_101 | | CR_EXT1[16:19] | SVCR4_101 |
401 | | CR_EXT2[0:3] | SVCR0_110 | | CR_EXT2[16:19] | SVCR4_110 |
402 | | CR_EXT3[0:3] | SVCR0_111 | | CR_EXT3[16:19] | SVCR4_111 |
403 | CR[1] | CR[36:39] | SVCR1_000 | CR[5] | CR[52:55] | SVCR5_000 |
404 | | CR_EXT1[36:39] | SVCR1_001 | | CR_EXT1[52:55] | SVCR5_001 |
405 | | CR_EXT2[36:39] | SVCR1_010 | | CR_EXT2[52:55] | SVCR5_010 |
406 | | CR_EXT3[36:39] | SVCR1_011 | | CR_EXT3[52:55] | SVCR5_011 |
407 | *CR[-7]* | CR[4:7] | SVCR1_100 | *CR[-3]* | CR[20:23] | SVCR5_100 |
408 | | CR_EXT1[4:7] | SVCR1_101 | | CR_EXT1[20:23] | SVCR5_101 |
409 | | CR_EXT2[4:7] | SVCR1_110 | | CR_EXT2[20:23] | SVCR5_110 |
410 | | CR_EXT3[4:7] | SVCR1_111 | | CR_EXT3[20:23] | SVCR5_111 |
411 | CR[2] | CR[40:43] | SVCR2_000 | CR[6] | CR[56:59] | SVCR6_000 |
412 | | CR_EXT1[40:43] | SVCR2_001 | | CR_EXT1[56:59] | SVCR6_001 |
413 | | CR_EXT2[40:43] | SVCR2_010 | | CR_EXT2[56:59] | SVCR6_010 |
414 | | CR_EXT3[40:43] | SVCR2_011 | | CR_EXT3[56:59] | SVCR6_011 |
415 | *CR[-6]* | CR[8:11] | SVCR2_100 | *CR[-2]* | CR[24:27] | SVCR6_100 |
416 | | CR_EXT1[8:11] | SVCR2_101 | | CR_EXT1[24:27] | SVCR6_101 |
417 | | CR_EXT2[8:11] | SVCR2_110 | | CR_EXT2[24:27] | SVCR6_110 |
418 | | CR_EXT3[8:11] | SVCR2_111 | | CR_EXT3[24:27] | SVCR6_111 |
419 | CR[3] | CR[44:47] | SVCR3_000 | CR[7] | CR[60:63] | SVCR7_000 |
420 | | CR_EXT1[44:47] | SVCR3_001 | | CR_EXT1[60:63] | SVCR7_001 |
421 | | CR_EXT2[44:47] | SVCR3_010 | | CR_EXT2[60:63] | SVCR7_010 |
422 | | CR_EXT3[44:47] | SVCR3_011 | | CR_EXT3[60:63] | SVCR7_011 |
423 | *CR[-5]* | CR[12:15] | SVCR3_100 | *CR[-1]* | CR[28:31] | SVCR7_100 |
424 | | CR_EXT1[12:15] | SVCR3_101 | | CR_EXT1[28:31] | SVCR7_101 |
425 | | CR_EXT2[12:15] | SVCR3_110 | | CR_EXT2[28:31] | SVCR7_110 |
426 | | CR_EXT3[12:15] | SVCR3_111 | | CR_EXT3[28:31] | SVCR7_111 |
428 Note: CR[-8] through CR[-1] are not part of OpenPower v3.1, they are the MSB half of the 64-bit CR SPR.
432 Instructions are broken down by Register Profiles as listed in the following auto-generated page:
433 [[opcode_regs_deduped]]. "Non-SV" indicates that the operations with this Register Profile cannot be Vectorised (mtspr, bc, dcbz, twi)
457 Remapped Encoding Fields:
459 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:23` |
460 |-----------|-------|---------|-------|-------------|-------------|-------------|---------|
461 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | Rsrc2_EXTRA | TBD |
465 Remapped Encoding Fields:
467 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:18` | `19:20` | `21:23` |
468 |-----------|-------|---------|-------|-------------|-------------|----------|-------------|-----------|---------|
469 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | MASK_SRC | ELWIDTH_SRC | SUBVL_SRC | TBD |
473 Remapped Encoding Fields:
475 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:18` | `19:20` | `21:23` |
476 |-----------|-------|---------|-------|-------------|-------------|----------|-------------|-----------|---------|
477 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | MASK_SRC | ELWIDTH_SRC | SUBVL_SRC | TBD |
481 Remapped Encoding Fields:
483 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:18` | `19:20` | `21:23` |
484 |-----------|-------|---------|-------|-------------|-------------|----------|-------------|-----------|---------|
485 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | MASK_SRC | ELWIDTH_SRC | SUBVL_SRC | TBD |
489 Remapped Encoding Fields:
491 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:18` | `19:20` | `21:23` |
492 |-----------|-------|---------|-------|-------------|-------------|----------|-------------|-----------|---------|
493 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | MASK_SRC | ELWIDTH_SRC | SUBVL_SRC | TBD |
497 Remapped Encoding Fields:
499 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:18` | `19:20` | `21:23` |
500 |-----------|-------|---------|-------|-------------|-------------|----------|-------------|-----------|---------|
501 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | MASK_SRC | ELWIDTH_SRC | SUBVL_SRC | TBD |
505 Remapped Encoding Fields:
507 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:18` | `19:20` | `21:23` |
508 |-----------|-------|---------|-------|-------------|-------------|----------|-------------|-----------|---------|
509 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | MASK_SRC | ELWIDTH_SRC | SUBVL_SRC | TBD |
513 Remapped Encoding Fields:
515 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:18` | `19:20` | `21:23` |
516 |-----------|-------|---------|-------|-------------|-------------|----------|-------------|-----------|---------|
517 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | MASK_SRC | ELWIDTH_SRC | SUBVL_SRC | TBD |
521 Remapped Encoding Fields:
523 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:23` |
524 |-----------|-------|---------|-------|-------------|-------------|-------------|---------|
525 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | Rsrc2_EXTRA | TBD |
529 Remapped Encoding Fields:
531 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:23` |
532 |-----------|-------|---------|-------|-------------|-------------|-------------|---------|
533 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | Rsrc2_EXTRA | TBD |
537 Remapped Encoding Fields:
539 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:23` |
540 |-----------|-------|---------|-------|-------------|-------------|-------------|---------|
541 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | Rsrc2_EXTRA | TBD |
545 Remapped Encoding Fields:
547 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:23` |
548 |-----------|-------|---------|-------|-------------|-------------|-------------|---------|
549 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | Rsrc2_EXTRA | TBD |
551 <!-- comment needed to stop ikiwiki markdown from mis-parsing table -->
553 ## 2R-1W-CRo (rl(w|d)imi)
555 Remapped Encoding Fields:
557 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:23` |
558 |-----------|-------|---------|-------|-------------|-------------|---------|
559 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | TBD |
565 Remapped Encoding Fields:
567 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:23` |
568 |-----------|-------|---------|-------|-------------|-------------|-------------|---------|
569 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | Rsrc2_EXTRA | TBD |
573 Remapped Encoding Fields:
575 | `0` | `1:3` | `4:5` | `6:7` | `8:10` | `11:13` | `14:16` | `17:19` | `20:23` |
576 |-----------|-------|---------|-------|-------------|-------------|-------------|-------------|----------|
577 | MASK_KIND | MASK | ELWIDTH | SUBVL | Rdest_EXTRA | Rsrc1_EXTRA | Rsrc2_EXTRA | Rsrc3_EXTRA | Reserved |