Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

AVX10.1 API introduction in JIT #101938

Merged
merged 22 commits into from
Jun 9, 2024
Merged

AVX10.1 API introduction in JIT #101938

Show file tree
Hide file tree
Changes from 21 commits
Commits
Show all changes
22 commits
Select commit Hold shift + click to select a range
8e7745f
Add AVX10v1 API surface
khushal1996 May 8, 2024
41ab759
Define HWINTRINSIC for AVX10v1, AVX10v1_V256 and AVX10v1_V512
khushal1996 May 8, 2024
445be70
Setup template testing for AVX10v1 APIs
khushal1996 May 8, 2024
cad5824
Handle AVX10v1 APIs in JIT where equivalent AVX512* APIs are handled
khushal1996 May 8, 2024
4079f41
Merge Avx10v1 and Avx10v1.V256. Rename Avx10.cs to Avx10v1.cs
khushal1996 May 9, 2024
8054bdf
Add Avx10v1 to relevant places
khushal1996 May 9, 2024
3386ed4
Fix CI errors. Add missing API in Avx10v1.PlatofrmNotSupported ad end…
khushal1996 May 9, 2024
2b56317
Changes to be made with latest changes on main. Make appropriate comm…
khushal1996 May 10, 2024
43d9d80
Lower AVX10v1 hwintrinsic in lowering and gentree.cpp for simdSize 32/16
khushal1996 May 11, 2024
a7bbd75
Fix failures on GNR for AVX10v1
khushal1996 May 12, 2024
6fc9d38
Disable template tests disabled for Avx512
khushal1996 May 14, 2024
ebd3ee3
Distinguish between Avx10v1 and Avx10v1/512, Add appropriate comments…
khushal1996 May 14, 2024
f426baa
Remove duplicate code and rather use a single if condition
khushal1996 May 15, 2024
455d754
Use bool instead of compIsa checks where possible
khushal1996 May 16, 2024
52a5aa0
remove duplication of code in shuffle
khushal1996 May 16, 2024
20e022b
resolve review comments. Make evex encoding checks clear to read and …
khushal1996 May 16, 2024
2d8fc4c
Add FMA and Avx512F.X64 instructions to AVX10v1. Restructure code and…
khushal1996 May 22, 2024
79b2a52
Combine compOpportunistic checks with Avx10 check using IsAvx10OrIsaS…
khushal1996 May 22, 2024
c38b62f
Introduce a new internal ISA InstructionSet_EVEX and remove Instructi…
khushal1996 Jun 3, 2024
7a51e1e
Addressing review comments. resolving errors introduced when merged w…
khushal1996 Jun 3, 2024
b1509c4
fix formatting
khushal1996 Jun 4, 2024
e21bcff
Reorder declaration of InstructionSet_EVEX to proper position. Run fo…
khushal1996 Jun 6, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
72 changes: 38 additions & 34 deletions src/coreclr/inc/corinfoinstructionset.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,11 +76,11 @@ enum CORINFO_InstructionSet
InstructionSet_AVX512VBMI=31,
InstructionSet_AVX512VBMI_VL=32,
InstructionSet_AVX10v1=33,
InstructionSet_AVX10v1_V256=34,
InstructionSet_AVX10v1_V512=35,
InstructionSet_VectorT128=36,
InstructionSet_VectorT256=37,
InstructionSet_VectorT512=38,
InstructionSet_AVX10v1_V512=34,
InstructionSet_VectorT128=35,
InstructionSet_VectorT256=36,
InstructionSet_VectorT512=37,
InstructionSet_EVEX=38,
InstructionSet_X86Base_X64=39,
InstructionSet_SSE_X64=40,
InstructionSet_SSE2_X64=41,
Expand Down Expand Up @@ -111,8 +111,8 @@ enum CORINFO_InstructionSet
InstructionSet_AVX512VBMI_X64=66,
InstructionSet_AVX512VBMI_VL_X64=67,
InstructionSet_AVX10v1_X64=68,
InstructionSet_AVX10v1_V256_X64=69,
InstructionSet_AVX10v1_V512_X64=70,
InstructionSet_AVX10v1_V512_X64=69,
InstructionSet_EVEX_X64=70,
#endif // TARGET_AMD64
#ifdef TARGET_X86
InstructionSet_X86Base=1,
Expand Down Expand Up @@ -148,11 +148,11 @@ enum CORINFO_InstructionSet
InstructionSet_AVX512VBMI=31,
InstructionSet_AVX512VBMI_VL=32,
InstructionSet_AVX10v1=33,
InstructionSet_AVX10v1_V256=34,
InstructionSet_AVX10v1_V512=35,
InstructionSet_VectorT128=36,
InstructionSet_VectorT256=37,
InstructionSet_VectorT512=38,
InstructionSet_AVX10v1_V512=34,
InstructionSet_VectorT128=35,
InstructionSet_VectorT256=36,
InstructionSet_VectorT512=37,
InstructionSet_EVEX=38,
InstructionSet_X86Base_X64=39,
InstructionSet_SSE_X64=40,
InstructionSet_SSE2_X64=41,
Expand Down Expand Up @@ -183,8 +183,8 @@ enum CORINFO_InstructionSet
InstructionSet_AVX512VBMI_X64=66,
InstructionSet_AVX512VBMI_VL_X64=67,
InstructionSet_AVX10v1_X64=68,
InstructionSet_AVX10v1_V256_X64=69,
InstructionSet_AVX10v1_V512_X64=70,
InstructionSet_AVX10v1_V512_X64=69,
InstructionSet_EVEX_X64=70,
#endif // TARGET_X86

};
Expand Down Expand Up @@ -360,10 +360,10 @@ struct CORINFO_InstructionSetFlags
AddInstructionSet(InstructionSet_AVX512VBMI_VL_X64);
if (HasInstructionSet(InstructionSet_AVX10v1))
AddInstructionSet(InstructionSet_AVX10v1_X64);
if (HasInstructionSet(InstructionSet_AVX10v1_V256))
AddInstructionSet(InstructionSet_AVX10v1_V256_X64);
if (HasInstructionSet(InstructionSet_AVX10v1_V512))
AddInstructionSet(InstructionSet_AVX10v1_V512_X64);
if (HasInstructionSet(InstructionSet_EVEX))
AddInstructionSet(InstructionSet_EVEX_X64);
#endif // TARGET_AMD64
#ifdef TARGET_X86
#endif // TARGET_X86
Expand Down Expand Up @@ -564,14 +564,14 @@ inline CORINFO_InstructionSetFlags EnsureInstructionSetFlagsAreValid(CORINFO_Ins
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_X64) && !resultflags.HasInstructionSet(InstructionSet_AVX10v1))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1_X64);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V256) && !resultflags.HasInstructionSet(InstructionSet_AVX10v1_V256_X64))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1_V256);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V256_X64) && !resultflags.HasInstructionSet(InstructionSet_AVX10v1_V256))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1_V256_X64);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V512) && !resultflags.HasInstructionSet(InstructionSet_AVX10v1_V512_X64))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1_V512);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V512_X64) && !resultflags.HasInstructionSet(InstructionSet_AVX10v1_V512))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1_V512_X64);
if (resultflags.HasInstructionSet(InstructionSet_EVEX) && !resultflags.HasInstructionSet(InstructionSet_EVEX_X64))
resultflags.RemoveInstructionSet(InstructionSet_EVEX);
if (resultflags.HasInstructionSet(InstructionSet_EVEX_X64) && !resultflags.HasInstructionSet(InstructionSet_EVEX))
resultflags.RemoveInstructionSet(InstructionSet_EVEX_X64);
if (resultflags.HasInstructionSet(InstructionSet_SSE) && !resultflags.HasInstructionSet(InstructionSet_X86Base))
resultflags.RemoveInstructionSet(InstructionSet_SSE);
if (resultflags.HasInstructionSet(InstructionSet_SSE2) && !resultflags.HasInstructionSet(InstructionSet_SSE))
Expand Down Expand Up @@ -618,6 +618,8 @@ inline CORINFO_InstructionSetFlags EnsureInstructionSetFlagsAreValid(CORINFO_Ins
resultflags.RemoveInstructionSet(InstructionSet_AVX512F);
if (resultflags.HasInstructionSet(InstructionSet_AVX512F) && !resultflags.HasInstructionSet(InstructionSet_FMA))
resultflags.RemoveInstructionSet(InstructionSet_AVX512F);
if (resultflags.HasInstructionSet(InstructionSet_AVX512F) && !resultflags.HasInstructionSet(InstructionSet_EVEX))
resultflags.RemoveInstructionSet(InstructionSet_AVX512F);
if (resultflags.HasInstructionSet(InstructionSet_AVX512F_VL) && !resultflags.HasInstructionSet(InstructionSet_AVX512F))
resultflags.RemoveInstructionSet(InstructionSet_AVX512F_VL);
if (resultflags.HasInstructionSet(InstructionSet_AVX512CD) && !resultflags.HasInstructionSet(InstructionSet_AVX512F))
Expand Down Expand Up @@ -648,9 +650,9 @@ inline CORINFO_InstructionSetFlags EnsureInstructionSetFlagsAreValid(CORINFO_Ins
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1) && !resultflags.HasInstructionSet(InstructionSet_FMA))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V256) && !resultflags.HasInstructionSet(InstructionSet_AVX10v1))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1_V256);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V512) && !resultflags.HasInstructionSet(InstructionSet_AVX10v1_V256))
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1) && !resultflags.HasInstructionSet(InstructionSet_EVEX))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V512) && !resultflags.HasInstructionSet(InstructionSet_AVX10v1))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1_V512);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V512) && !resultflags.HasInstructionSet(InstructionSet_AVX512F))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1_V512);
Expand Down Expand Up @@ -732,6 +734,8 @@ inline CORINFO_InstructionSetFlags EnsureInstructionSetFlagsAreValid(CORINFO_Ins
resultflags.RemoveInstructionSet(InstructionSet_AVX512F);
if (resultflags.HasInstructionSet(InstructionSet_AVX512F) && !resultflags.HasInstructionSet(InstructionSet_FMA))
resultflags.RemoveInstructionSet(InstructionSet_AVX512F);
if (resultflags.HasInstructionSet(InstructionSet_AVX512F) && !resultflags.HasInstructionSet(InstructionSet_EVEX))
resultflags.RemoveInstructionSet(InstructionSet_AVX512F);
if (resultflags.HasInstructionSet(InstructionSet_AVX512F_VL) && !resultflags.HasInstructionSet(InstructionSet_AVX512F))
resultflags.RemoveInstructionSet(InstructionSet_AVX512F_VL);
if (resultflags.HasInstructionSet(InstructionSet_AVX512CD) && !resultflags.HasInstructionSet(InstructionSet_AVX512F))
Expand Down Expand Up @@ -762,9 +766,9 @@ inline CORINFO_InstructionSetFlags EnsureInstructionSetFlagsAreValid(CORINFO_Ins
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1) && !resultflags.HasInstructionSet(InstructionSet_FMA))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V256) && !resultflags.HasInstructionSet(InstructionSet_AVX10v1))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1_V256);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V512) && !resultflags.HasInstructionSet(InstructionSet_AVX10v1_V256))
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1) && !resultflags.HasInstructionSet(InstructionSet_EVEX))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V512) && !resultflags.HasInstructionSet(InstructionSet_AVX10v1))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1_V512);
if (resultflags.HasInstructionSet(InstructionSet_AVX10v1_V512) && !resultflags.HasInstructionSet(InstructionSet_AVX512F))
resultflags.RemoveInstructionSet(InstructionSet_AVX10v1_V512);
Expand Down Expand Up @@ -992,10 +996,6 @@ inline const char *InstructionSetToString(CORINFO_InstructionSet instructionSet)
return "AVX10v1";
case InstructionSet_AVX10v1_X64 :
return "AVX10v1_X64";
case InstructionSet_AVX10v1_V256 :
return "AVX10v1_V256";
case InstructionSet_AVX10v1_V256_X64 :
return "AVX10v1_V256_X64";
case InstructionSet_AVX10v1_V512 :
return "AVX10v1_V512";
case InstructionSet_AVX10v1_V512_X64 :
Expand All @@ -1006,6 +1006,10 @@ inline const char *InstructionSetToString(CORINFO_InstructionSet instructionSet)
return "VectorT256";
case InstructionSet_VectorT512 :
return "VectorT512";
case InstructionSet_EVEX :
return "EVEX";
case InstructionSet_EVEX_X64 :
return "EVEX_X64";
#endif // TARGET_AMD64
#ifdef TARGET_X86
case InstructionSet_X86Base :
Expand Down Expand Up @@ -1074,8 +1078,6 @@ inline const char *InstructionSetToString(CORINFO_InstructionSet instructionSet)
return "AVX512VBMI_VL";
case InstructionSet_AVX10v1 :
return "AVX10v1";
case InstructionSet_AVX10v1_V256 :
return "AVX10v1_V256";
case InstructionSet_AVX10v1_V512 :
return "AVX10v1_V512";
case InstructionSet_VectorT128 :
Expand All @@ -1084,6 +1086,8 @@ inline const char *InstructionSetToString(CORINFO_InstructionSet instructionSet)
return "VectorT256";
case InstructionSet_VectorT512 :
return "VectorT512";
case InstructionSet_EVEX :
return "EVEX";
#endif // TARGET_X86

default:
Expand Down Expand Up @@ -1149,11 +1153,11 @@ inline CORINFO_InstructionSet InstructionSetFromR2RInstructionSet(ReadyToRunInst
case READYTORUN_INSTRUCTION_Avx512Vbmi: return InstructionSet_AVX512VBMI;
case READYTORUN_INSTRUCTION_Avx512Vbmi_VL: return InstructionSet_AVX512VBMI_VL;
case READYTORUN_INSTRUCTION_Avx10v1: return InstructionSet_AVX10v1;
case READYTORUN_INSTRUCTION_Avx10v1_V256: return InstructionSet_AVX10v1_V256;
case READYTORUN_INSTRUCTION_Avx10v1_V512: return InstructionSet_AVX10v1_V512;
case READYTORUN_INSTRUCTION_VectorT128: return InstructionSet_VectorT128;
case READYTORUN_INSTRUCTION_VectorT256: return InstructionSet_VectorT256;
case READYTORUN_INSTRUCTION_VectorT512: return InstructionSet_VectorT512;
case READYTORUN_INSTRUCTION_EVEX: return InstructionSet_EVEX;
#endif // TARGET_AMD64
#ifdef TARGET_X86
case READYTORUN_INSTRUCTION_X86Base: return InstructionSet_X86Base;
Expand Down Expand Up @@ -1186,11 +1190,11 @@ inline CORINFO_InstructionSet InstructionSetFromR2RInstructionSet(ReadyToRunInst
case READYTORUN_INSTRUCTION_Avx512Vbmi: return InstructionSet_AVX512VBMI;
case READYTORUN_INSTRUCTION_Avx512Vbmi_VL: return InstructionSet_AVX512VBMI_VL;
case READYTORUN_INSTRUCTION_Avx10v1: return InstructionSet_AVX10v1;
case READYTORUN_INSTRUCTION_Avx10v1_V256: return InstructionSet_AVX10v1_V256;
case READYTORUN_INSTRUCTION_Avx10v1_V512: return InstructionSet_AVX10v1_V512;
case READYTORUN_INSTRUCTION_VectorT128: return InstructionSet_VectorT128;
case READYTORUN_INSTRUCTION_VectorT256: return InstructionSet_VectorT256;
case READYTORUN_INSTRUCTION_VectorT512: return InstructionSet_VectorT512;
case READYTORUN_INSTRUCTION_EVEX: return InstructionSet_EVEX;
#endif // TARGET_X86

default:
Expand Down
10 changes: 5 additions & 5 deletions src/coreclr/inc/jiteeversionguid.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,11 +43,11 @@ typedef const GUID *LPCGUID;
#define GUID_DEFINED
#endif // !GUID_DEFINED

constexpr GUID JITEEVersionIdentifier = { /* 227e46fa-1be3-4770-b613-4a239e7c28aa */
0x227e46fa,
0x1be3,
0x4770,
{0xb6, 0x13, 0x4a, 0x23, 0x9e, 0x7c, 0x28, 0xaa}
constexpr GUID JITEEVersionIdentifier = { /* 58e03d68-4c56-43a3-ab80-d3ca21dc6526 */
0x58e03d68,
0x4c56,
0x43a3,
{0xab, 0x80, 0xd3, 0xca, 0x21, 0xdc, 0x65, 0x26}
};

//////////////////////////////////////////////////////////////////////////////////////////////////////////
Expand Down
2 changes: 1 addition & 1 deletion src/coreclr/inc/readytoruninstructionset.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -53,8 +53,8 @@ enum ReadyToRunInstructionSet
READYTORUN_INSTRUCTION_Rcpc2=42,
READYTORUN_INSTRUCTION_Sve=43,
READYTORUN_INSTRUCTION_Avx10v1=44,
READYTORUN_INSTRUCTION_Avx10v1_V256=45,
READYTORUN_INSTRUCTION_Avx10v1_V512=46,
READYTORUN_INSTRUCTION_EVEX=47,

};

Expand Down
4 changes: 2 additions & 2 deletions src/coreclr/jit/assertionprop.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3184,8 +3184,8 @@ bool Compiler::optIsProfitableToSubstitute(GenTree* dest, BasicBlock* destBlock,
return (simdBaseType == TYP_FLOAT) && vecCon->IsZero();
}

case NI_AVX512F_CompareEqualMask:
case NI_AVX512F_CompareNotEqualMask:
case NI_EVEX_CompareEqualMask:
case NI_EVEX_CompareNotEqualMask:
{
// We can optimize when the constant is zero, but only
// for non floating-point since +0.0 == -0.0
Expand Down
34 changes: 32 additions & 2 deletions src/coreclr/jit/codegencommon.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1827,7 +1827,22 @@ void CodeGen::genGenerateMachineCode()
#if defined(TARGET_X86)
if (compiler->canUseEvexEncoding())
{
printf("X86 with AVX512");
if (compiler->compOpportunisticallyDependsOn(InstructionSet_AVX10v1))
{
if (compiler->compOpportunisticallyDependsOn(InstructionSet_AVX10v1_V512))
{
printf("X86 with AVX10/512");
}
else
{
printf("X86 with AVX10/256");
}
}
else
{
assert(compiler->compIsaSupportedDebugOnly(InstructionSet_AVX512F));
printf("X86 with AVX512");
}
}
else if (compiler->canUseVexEncoding())
{
Expand All @@ -1840,7 +1855,22 @@ void CodeGen::genGenerateMachineCode()
#elif defined(TARGET_AMD64)
if (compiler->canUseEvexEncoding())
{
printf("X64 with AVX512");
if (compiler->compOpportunisticallyDependsOn(InstructionSet_AVX10v1))
{
if (compiler->compOpportunisticallyDependsOn(InstructionSet_AVX10v1_V512))
{
printf("X86 with AVX10/512");
}
else
{
printf("X86 with AVX10/256");
}
}
else
{
assert(compiler->compIsaSupportedDebugOnly(InstructionSet_AVX512F));
printf("X86 with AVX512");
}
}
else if (compiler->canUseVexEncoding())
{
Expand Down
Loading
Loading