Skip to content

Commit

Permalink
Merging formats SVE_AB_3A and SVE_AD_3A into SVE_AA_3A (dotnet#101754)
Browse files Browse the repository at this point in the history
  • Loading branch information
TIHan authored and Ruihan-Yin committed May 30, 2024
1 parent 9df54e3 commit dd596e5
Show file tree
Hide file tree
Showing 4 changed files with 60 additions and 55 deletions.
28 changes: 12 additions & 16 deletions src/coreclr/jit/codegenarm64test.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -4572,14 +4572,24 @@ void CodeGen::genArm64EmitterUnitTestsSve()
INS_OPTS_SCALABLE_S); // EOR <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_orr, EA_SCALABLE, REG_V29, REG_P7, REG_V31,
INS_OPTS_SCALABLE_D); // ORR <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>

// IF_SVE_AB_3A
theEmitter->emitIns_R_R_R(INS_sve_add, EA_SCALABLE, REG_V5, REG_P6, REG_V7,
INS_OPTS_SCALABLE_B); // ADD <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_sub, EA_SCALABLE, REG_V15, REG_P7, REG_V29,
INS_OPTS_SCALABLE_H); // SUB <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_subr, EA_SCALABLE, REG_V2, REG_P0, REG_V13,
INS_OPTS_SCALABLE_S); // SUBR <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_smax, EA_SCALABLE, REG_V24, REG_P0, REG_V2,
INS_OPTS_SCALABLE_B); // SMAX <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_smin, EA_SCALABLE, REG_V9, REG_P1, REG_V27,
INS_OPTS_SCALABLE_H); // SMIN <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_sabd, EA_SCALABLE, REG_V5, REG_P2, REG_V6,
INS_OPTS_SCALABLE_B); // SABD <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_uabd, EA_SCALABLE, REG_V23, REG_P3, REG_V9,
INS_OPTS_SCALABLE_S); // UABD <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_umax, EA_SCALABLE, REG_V15, REG_P4, REG_V2,
INS_OPTS_SCALABLE_S); // UMAX <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_umin, EA_SCALABLE, REG_V12, REG_P7, REG_V0,
INS_OPTS_SCALABLE_D); // UMIN <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>

#ifdef ALL_ARM64_EMITTER_UNIT_TESTS_SVE_UNSUPPORTED
// IF_SVE_AB_3B
Expand All @@ -4599,20 +4609,6 @@ void CodeGen::genArm64EmitterUnitTestsSve()
theEmitter->emitIns_R_R_R(INS_sve_udivr, EA_SCALABLE, REG_V13, REG_P7, REG_V15,
INS_OPTS_SCALABLE_D); // UDIVR <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>

// IF_SVE_AD_3A
theEmitter->emitIns_R_R_R(INS_sve_smax, EA_SCALABLE, REG_V24, REG_P0, REG_V2,
INS_OPTS_SCALABLE_B); // SMAX <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_smin, EA_SCALABLE, REG_V9, REG_P1, REG_V27,
INS_OPTS_SCALABLE_H); // SMIN <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_sabd, EA_SCALABLE, REG_V5, REG_P2, REG_V6,
INS_OPTS_SCALABLE_B); // SABD <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_uabd, EA_SCALABLE, REG_V23, REG_P3, REG_V9,
INS_OPTS_SCALABLE_S); // UABD <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_umax, EA_SCALABLE, REG_V15, REG_P4, REG_V2,
INS_OPTS_SCALABLE_S); // UMAX <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
theEmitter->emitIns_R_R_R(INS_sve_umin, EA_SCALABLE, REG_V12, REG_P7, REG_V0,
INS_OPTS_SCALABLE_D); // UMIN <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>

// IF_SVE_AE_3A
theEmitter->emitIns_R_R_R(INS_sve_mul, EA_SCALABLE, REG_V5, REG_P1, REG_V3,
INS_OPTS_SCALABLE_D); // MUL <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
Expand Down
55 changes: 33 additions & 22 deletions src/coreclr/jit/emitarm64sve.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -347,7 +347,7 @@ emitter::code_t emitter::emitInsCodeSve(instruction ins, insFormat fmt)
const static insFormat formatEncode4J[4] = {IF_SVE_BV_2A, IF_SVE_BV_2A_A, IF_SVE_CP_3A, IF_SVE_CQ_3A};
const static insFormat formatEncode4K[4] = {IF_SVE_IF_4A, IF_SVE_IF_4A_A, IF_SVE_IM_3A, IF_SVE_IN_4A};
const static insFormat formatEncode4L[4] = {IF_SVE_IZ_4A, IF_SVE_IZ_4A_A, IF_SVE_JB_4A, IF_SVE_JM_3A};
const static insFormat formatEncode3A[3] = {IF_SVE_AB_3A, IF_SVE_AT_3A, IF_SVE_EC_1A};
const static insFormat formatEncode3A[3] = {IF_SVE_AA_3A, IF_SVE_AT_3A, IF_SVE_EC_1A};
const static insFormat formatEncode3B[3] = {IF_SVE_BH_3A, IF_SVE_BH_3B, IF_SVE_BH_3B_A};
const static insFormat formatEncode3C[3] = {IF_SVE_BW_2A, IF_SVE_CB_2A, IF_SVE_EB_1A};
const static insFormat formatEncode3D[3] = {IF_SVE_BR_3A, IF_SVE_BR_3B, IF_SVE_CI_3A};
Expand All @@ -371,7 +371,7 @@ emitter::code_t emitter::emitInsCodeSve(instruction ins, insFormat fmt)
const static insFormat formatEncode3V[3] = {IF_SVE_JA_4A, IF_SVE_JB_4A, IF_SVE_JM_3A};
const static insFormat formatEncode2AA[2] = {IF_SVE_ID_2A, IF_SVE_IE_2A};
const static insFormat formatEncode2AB[2] = {IF_SVE_JG_2A, IF_SVE_JH_2A};
const static insFormat formatEncode2AC[2] = {IF_SVE_AD_3A, IF_SVE_ED_1A};
const static insFormat formatEncode2AC[2] = {IF_SVE_AA_3A, IF_SVE_ED_1A};
const static insFormat formatEncode2AD[2] = {IF_SVE_AB_3B, IF_SVE_AT_3B};
const static insFormat formatEncode2AE[2] = {IF_SVE_CG_2A, IF_SVE_CJ_2A};
const static insFormat formatEncode2AF[2] = {IF_SVE_AE_3A, IF_SVE_BD_3A};
Expand All @@ -387,7 +387,7 @@ emitter::code_t emitter::emitInsCodeSve(instruction ins, insFormat fmt)
const static insFormat formatEncode2AP[2] = {IF_SVE_GY_3B, IF_SVE_HA_3A};
const static insFormat formatEncode2AQ[2] = {IF_SVE_GO_3A, IF_SVE_HC_3A};
const static insFormat formatEncode2AR[2] = {IF_SVE_AP_3A, IF_SVE_CZ_4A};
const static insFormat formatEncode2AT[2] = {IF_SVE_AB_3A, IF_SVE_EC_1A};
const static insFormat formatEncode2AT[2] = {IF_SVE_AA_3A, IF_SVE_EC_1A};
const static insFormat formatEncode2AU[2] = {IF_SVE_AH_3A, IF_SVE_BI_2A};
const static insFormat formatEncode2AV[2] = {IF_SVE_BM_1A, IF_SVE_BN_1A};
const static insFormat formatEncode2AW[2] = {IF_SVE_BO_1A, IF_SVE_BP_1A};
Expand Down Expand Up @@ -2947,7 +2947,7 @@ void emitter::emitInsSve_R_R_R(instruction ins,
{
assert(isLowPredicateRegister(reg2));
assert(insScalableOptsNone(sopt));
fmt = IF_SVE_AB_3A;
fmt = IF_SVE_AA_3A;
}
break;

Expand Down Expand Up @@ -2994,7 +2994,7 @@ void emitter::emitInsSve_R_R_R(instruction ins,
assert(isVectorRegister(reg3));
assert(insOptsScalableStandard(opt));
assert(insScalableOptsNone(sopt));
fmt = IF_SVE_AD_3A;
fmt = IF_SVE_AA_3A;
break;

case INS_sve_mul:
Expand Down Expand Up @@ -8017,11 +8017,9 @@ void emitter::emitIns_PRFOP_R_R_I(instruction ins,
case IF_SVE_HP_3B:
case IF_SVE_AR_4A:
case IF_SVE_BV_2A_A:
case IF_SVE_AB_3A:
case IF_SVE_ET_3A:
case IF_SVE_HU_4A:
case IF_SVE_HL_3B:
case IF_SVE_AD_3A:
case IF_SVE_AB_3B:
case IF_SVE_AE_3A:
case IF_SVE_EU_3A:
Expand Down Expand Up @@ -9905,10 +9903,9 @@ BYTE* emitter::emitOutput_InstrSve(BYTE* dst, instrDesc* id)
switch (fmt)
{
// Scalable.
case IF_SVE_AA_3A: // ........xx...... ...gggmmmmmddddd -- SVE bitwise logical operations (predicated)
case IF_SVE_AB_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer add/subtract vectors (predicated)
case IF_SVE_AA_3A: // ........xx...... ...gggmmmmmddddd -- SVE bitwise logical operations + integer add/subtract
// vectors + SVE integer min/max/difference (predicated)
case IF_SVE_AC_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer divide vectors (predicated)
case IF_SVE_AD_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer min/max/difference (predicated)
case IF_SVE_AE_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer multiply vectors (predicated)
case IF_SVE_AF_3A: // ........xx...... ...gggnnnnnddddd -- SVE bitwise logical reduction (predicated)
case IF_SVE_AG_3A: // ........xx...... ...gggnnnnnddddd -- SVE bitwise logical reduction (quadwords)
Expand Down Expand Up @@ -12556,9 +12553,8 @@ void emitter::emitInsSveSanityCheck(instrDesc* id)
break;

// Scalable.
case IF_SVE_AA_3A: // ........xx...... ...gggmmmmmddddd -- SVE bitwise logical operations (predicated)
case IF_SVE_AB_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer add/subtract vectors (predicated)
case IF_SVE_AD_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer min/max/difference (predicated)
case IF_SVE_AA_3A: // ........xx...... ...gggmmmmmddddd -- SVE bitwise logical operations + integer add/subtract
// vectors + SVE integer min/max/difference (predicated)
case IF_SVE_AE_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer multiply vectors (predicated)
case IF_SVE_AN_3A: // ........xx...... ...gggmmmmmddddd -- SVE bitwise shift by vector (predicated)
case IF_SVE_CM_3A: // ........xx...... ...gggmmmmmddddd -- SVE conditionally broadcast element to vector
Expand Down Expand Up @@ -14484,10 +14480,9 @@ void emitter::emitDispInsSveHelp(instrDesc* id)
bitMaskImm bmi;

// <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T>
case IF_SVE_AA_3A: // ........xx...... ...gggmmmmmddddd -- SVE bitwise logical operations (predicated)
case IF_SVE_AB_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer add/subtract vectors (predicated)
case IF_SVE_AA_3A: // ........xx...... ...gggmmmmmddddd -- SVE bitwise logical operations + integer add/subtract
// vectors + SVE integer min/max/difference (predicated)
case IF_SVE_AC_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer divide vectors (predicated)
case IF_SVE_AD_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer min/max/difference (predicated)
case IF_SVE_AE_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer multiply vectors (predicated)
case IF_SVE_AN_3A: // ........xx...... ...gggmmmmmddddd -- SVE bitwise shift by vector (predicated)
case IF_SVE_EP_3A: // ........xx...... ...gggmmmmmddddd -- SVE2 integer halving add/subtract (predicated)
Expand Down Expand Up @@ -16342,16 +16337,32 @@ void emitter::getInsSveExecutionCharacteristics(instrDesc* id, insExecutionChara
switch (id->idInsFmt())
{
// Predicate logical
case IF_SVE_AA_3A: // ........xx...... ...gggmmmmmddddd -- SVE bitwise logical operations (predicated)
result.insLatency = PERFSCORE_LATENCY_1C;
result.insThroughput = PERFSCORE_THROUGHPUT_1C;
case IF_SVE_AA_3A: // ........xx...... ...gggmmmmmddddd -- SVE bitwise logical operations + integer add/subtract
// vectors + SVE integer min/max/difference (predicated)
switch (ins)
{
case INS_sve_add:
case INS_sve_sub:
case INS_sve_subr:
case INS_sve_sabd:
case INS_sve_smax:
case INS_sve_smin:
case INS_sve_uabd:
case INS_sve_umax:
case INS_sve_umin:
result.insLatency = PERFSCORE_LATENCY_2C;
result.insThroughput = PERFSCORE_THROUGHPUT_2X;
break;

default:
result.insLatency = PERFSCORE_LATENCY_1C;
result.insThroughput = PERFSCORE_THROUGHPUT_1C;
break;
}
break;

// Arithmetic, basic
case IF_SVE_AB_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer add/subtract vectors (predicated)
case IF_SVE_EP_3A: // ........xx...... ...gggmmmmmddddd -- SVE2 integer halving add/subtract (predicated)
// Max/min, basic and pairwise
case IF_SVE_AD_3A: // ........xx...... ...gggmmmmmddddd -- SVE integer min/max/difference (predicated)
result.insLatency = PERFSCORE_LATENCY_2C;
result.insThroughput = PERFSCORE_THROUGHPUT_2X;
break;
Expand Down
2 changes: 0 additions & 2 deletions src/coreclr/jit/emitfmtsarm64sve.h
Original file line number Diff line number Diff line change
Expand Up @@ -138,10 +138,8 @@ IF_DEF(SVE_2BS, IS_NONE, NONE) // Instruction has 2 possible encoding types, ty
*****************************************************************************/

IF_DEF(SVE_AA_3A, IS_NONE, NONE) // SVE_AA_3A ........xx...... ...gggmmmmmddddd -- SVE bitwise logical operations (predicated)
IF_DEF(SVE_AB_3A, IS_NONE, NONE) // SVE_AB_3A ........xx...... ...gggmmmmmddddd -- SVE integer add/subtract vectors (predicated)
IF_DEF(SVE_AB_3B, IS_NONE, NONE) // SVE_AB_3B ................ ...gggmmmmmddddd -- SVE integer add/subtract vectors (predicated)
IF_DEF(SVE_AC_3A, IS_NONE, NONE) // SVE_AC_3A ........xx...... ...gggmmmmmddddd -- SVE integer divide vectors (predicated)
IF_DEF(SVE_AD_3A, IS_NONE, NONE) // SVE_AD_3A ........xx...... ...gggmmmmmddddd -- SVE integer min/max/difference (predicated)
IF_DEF(SVE_AE_3A, IS_NONE, NONE) // SVE_AE_3A ........xx...... ...gggmmmmmddddd -- SVE integer multiply vectors (predicated)
IF_DEF(SVE_AF_3A, IS_NONE, NONE) // SVE_AF_3A ........xx...... ...gggnnnnnddddd -- SVE bitwise logical reduction (predicated)
IF_DEF(SVE_AG_3A, IS_NONE, NONE) // SVE_AG_3A ........xx...... ...gggnnnnnddddd -- SVE bitwise logical reduction (quadwords)
Expand Down
Loading

0 comments on commit dd596e5

Please sign in to comment.