Allow the user to control the MaxVectorTBitWidth

src/coreclr/inc/clrconfigvalues.h

@@ -357,12 +357,6 @@ RETAIL_CONFIG_DWORD_INFO(EXTERNAL_JitRegisterFP, W("JitRegisterFP"), 3, "Control
 RETAIL_CONFIG_DWORD_INFO(INTERNAL_JitELTHookEnabled, W("JitELTHookEnabled"), 0, "On ARM, setting this will emit Enter/Leave/TailCall callbacks")
 RETAIL_CONFIG_DWORD_INFO(INTERNAL_JitMemStats, W("JitMemStats"), 0, "Display JIT memory usage statistics")
 RETAIL_CONFIG_DWORD_INFO(INTERNAL_JitVNMapSelBudget, W("JitVNMapSelBudget"), 100, "Max # of MapSelect's considered for a particular top-level invocation.")
-#if defined(TARGET_AMD64) || defined(TARGET_X86) || defined(TARGET_ARM64)
-#define EXTERNAL_FeatureSIMD_Default 1
-#else // !(defined(TARGET_AMD64) || defined(TARGET_X86) || defined(TARGET_ARM64))
-#define EXTERNAL_FeatureSIMD_Default 0
-#endif // !(defined(TARGET_AMD64) || defined(TARGET_X86) || defined(TARGET_ARM64))
-RETAIL_CONFIG_DWORD_INFO(INTERNAL_SIMD16ByteOnly, W("SIMD16ByteOnly"), 0, "Limit maximum SIMD vector length to 16 bytes (used by x64_arm64_altjit)")
 RETAIL_CONFIG_DWORD_INFO(UNSUPPORTED_TrackDynamicMethodDebugInfo, W("TrackDynamicMethodDebugInfo"), 0, "Specifies whether debug info should be generated and tracked for dynamic methods")
 
 #ifdef FEATURE_MULTICOREJIT
@@ -745,15 +739,17 @@ RETAIL_CONFIG_DWORD_INFO(INTERNAL_GDBJitEmitDebugFrame, W("GDBJitEmitDebugFrame"
 #endif
 #endif
 
+RETAIL_CONFIG_DWORD_INFO(EXTERNAL_MaxVectorTBitWidth,           W("MaxVectorTBitWidth"),        0, "The maximum width, in bits, that Vector<T> is allowed to be. A value less than 128 is treated as the system default.")
+
 //
 // Hardware Intrinsic ISAs; keep in sync with jitconfigvalues.h
 //
 #if defined(TARGET_LOONGARCH64) || defined(TARGET_RISCV64)
 //TODO: should implement LoongArch64's features.
 //TODO-RISCV64-CQ: should implement RISCV64's features.
-RETAIL_CONFIG_DWORD_INFO(EXTERNAL_EnableHWIntrinsic,  W("EnableHWIntrinsic"),  0, "Allows Base+ hardware intrinsics to be disabled")
+RETAIL_CONFIG_DWORD_INFO(EXTERNAL_EnableHWIntrinsic,            W("EnableHWIntrinsic"),         0, "Allows Base+ hardware intrinsics to be disabled")
 #else
-RETAIL_CONFIG_DWORD_INFO(EXTERNAL_EnableHWIntrinsic,  W("EnableHWIntrinsic"),  1, "Allows Base+ hardware intrinsics to be disabled")
+RETAIL_CONFIG_DWORD_INFO(EXTERNAL_EnableHWIntrinsic,            W("EnableHWIntrinsic"),         1, "Allows Base+ hardware intrinsics to be disabled")
 #endif // defined(TARGET_LOONGARCH64) || defined(TARGET_RISCV64)
 
 #if defined(TARGET_AMD64) || defined(TARGET_X86)

src/coreclr/inc/corinfoinstructionset.h

@@ -14,7 +14,7 @@
 enum CORINFO_InstructionSet
 {
     InstructionSet_ILLEGAL = 0,
-    InstructionSet_NONE = 63,
+    InstructionSet_NONE = 127,
 #ifdef TARGET_ARM64
     InstructionSet_ArmBase=1,
     InstructionSet_AdvSimd=2,
@@ -29,14 +29,15 @@ enum CORINFO_InstructionSet
     InstructionSet_Vector128=11,
     InstructionSet_Dczva=12,
     InstructionSet_Rcpc=13,
-    InstructionSet_ArmBase_Arm64=14,
-    InstructionSet_AdvSimd_Arm64=15,
-    InstructionSet_Aes_Arm64=16,
-    InstructionSet_Crc32_Arm64=17,
-    InstructionSet_Dp_Arm64=18,
-    InstructionSet_Rdm_Arm64=19,
-    InstructionSet_Sha1_Arm64=20,
-    InstructionSet_Sha256_Arm64=21,
+    InstructionSet_VectorT128=14,
+    InstructionSet_ArmBase_Arm64=15,
+    InstructionSet_AdvSimd_Arm64=16,
+    InstructionSet_Aes_Arm64=17,
+    InstructionSet_Crc32_Arm64=18,
+    InstructionSet_Dp_Arm64=19,
+    InstructionSet_Rdm_Arm64=20,
+    InstructionSet_Sha1_Arm64=21,
+    InstructionSet_Sha256_Arm64=22,
 #endif // TARGET_ARM64
 #ifdef TARGET_AMD64
     InstructionSet_X86Base=1,
@@ -71,35 +72,38 @@ enum CORINFO_InstructionSet
     InstructionSet_AVX512DQ_VL=30,
     InstructionSet_AVX512VBMI=31,
     InstructionSet_AVX512VBMI_VL=32,
-    InstructionSet_X86Base_X64=33,
-    InstructionSet_SSE_X64=34,
-    InstructionSet_SSE2_X64=35,
-    InstructionSet_SSE3_X64=36,
-    InstructionSet_SSSE3_X64=37,
-    InstructionSet_SSE41_X64=38,
-    InstructionSet_SSE42_X64=39,
-    InstructionSet_AVX_X64=40,
-    InstructionSet_AVX2_X64=41,
-    InstructionSet_AES_X64=42,
-    InstructionSet_BMI1_X64=43,
-    InstructionSet_BMI2_X64=44,
-    InstructionSet_FMA_X64=45,
-    InstructionSet_LZCNT_X64=46,
-    InstructionSet_PCLMULQDQ_X64=47,
-    InstructionSet_POPCNT_X64=48,
-    InstructionSet_AVXVNNI_X64=49,
-    InstructionSet_MOVBE_X64=50,
-    InstructionSet_X86Serialize_X64=51,
-    InstructionSet_AVX512F_X64=52,
-    InstructionSet_AVX512F_VL_X64=53,
-    InstructionSet_AVX512BW_X64=54,
-    InstructionSet_AVX512BW_VL_X64=55,
-    InstructionSet_AVX512CD_X64=56,
-    InstructionSet_AVX512CD_VL_X64=57,
-    InstructionSet_AVX512DQ_X64=58,
-    InstructionSet_AVX512DQ_VL_X64=59,
-    InstructionSet_AVX512VBMI_X64=60,
-    InstructionSet_AVX512VBMI_VL_X64=61,
+    InstructionSet_VectorT128=33,
+    InstructionSet_VectorT256=34,
+    InstructionSet_VectorT512=35,
+    InstructionSet_X86Base_X64=36,
+    InstructionSet_SSE_X64=37,
+    InstructionSet_SSE2_X64=38,
+    InstructionSet_SSE3_X64=39,
+    InstructionSet_SSSE3_X64=40,
+    InstructionSet_SSE41_X64=41,
+    InstructionSet_SSE42_X64=42,
+    InstructionSet_AVX_X64=43,
+    InstructionSet_AVX2_X64=44,
+    InstructionSet_AES_X64=45,
+    InstructionSet_BMI1_X64=46,
+    InstructionSet_BMI2_X64=47,
+    InstructionSet_FMA_X64=48,
+    InstructionSet_LZCNT_X64=49,
+    InstructionSet_PCLMULQDQ_X64=50,
+    InstructionSet_POPCNT_X64=51,
+    InstructionSet_AVXVNNI_X64=52,
+    InstructionSet_MOVBE_X64=53,
+    InstructionSet_X86Serialize_X64=54,
+    InstructionSet_AVX512F_X64=55,
+    InstructionSet_AVX512F_VL_X64=56,
+    InstructionSet_AVX512BW_X64=57,
+    InstructionSet_AVX512BW_VL_X64=58,
+    InstructionSet_AVX512CD_X64=59,
+    InstructionSet_AVX512CD_VL_X64=60,
+    InstructionSet_AVX512DQ_X64=61,
+    InstructionSet_AVX512DQ_VL_X64=62,
+    InstructionSet_AVX512VBMI_X64=63,
+    InstructionSet_AVX512VBMI_VL_X64=64,
 #endif // TARGET_AMD64
 #ifdef TARGET_X86
     InstructionSet_X86Base=1,
@@ -134,43 +138,46 @@ enum CORINFO_InstructionSet
     InstructionSet_AVX512DQ_VL=30,
     InstructionSet_AVX512VBMI=31,
     InstructionSet_AVX512VBMI_VL=32,
-    InstructionSet_X86Base_X64=33,
-    InstructionSet_SSE_X64=34,
-    InstructionSet_SSE2_X64=35,
-    InstructionSet_SSE3_X64=36,
-    InstructionSet_SSSE3_X64=37,
-    InstructionSet_SSE41_X64=38,
-    InstructionSet_SSE42_X64=39,
-    InstructionSet_AVX_X64=40,
-    InstructionSet_AVX2_X64=41,
-    InstructionSet_AES_X64=42,
-    InstructionSet_BMI1_X64=43,
-    InstructionSet_BMI2_X64=44,
-    InstructionSet_FMA_X64=45,
-    InstructionSet_LZCNT_X64=46,
-    InstructionSet_PCLMULQDQ_X64=47,
-    InstructionSet_POPCNT_X64=48,
-    InstructionSet_AVXVNNI_X64=49,
-    InstructionSet_MOVBE_X64=50,
-    InstructionSet_X86Serialize_X64=51,
-    InstructionSet_AVX512F_X64=52,
-    InstructionSet_AVX512F_VL_X64=53,
-    InstructionSet_AVX512BW_X64=54,
-    InstructionSet_AVX512BW_VL_X64=55,
-    InstructionSet_AVX512CD_X64=56,
-    InstructionSet_AVX512CD_VL_X64=57,
-    InstructionSet_AVX512DQ_X64=58,
-    InstructionSet_AVX512DQ_VL_X64=59,
-    InstructionSet_AVX512VBMI_X64=60,
-    InstructionSet_AVX512VBMI_VL_X64=61,
+    InstructionSet_VectorT128=33,
+    InstructionSet_VectorT256=34,
+    InstructionSet_VectorT512=35,
+    InstructionSet_X86Base_X64=36,
+    InstructionSet_SSE_X64=37,
+    InstructionSet_SSE2_X64=38,
+    InstructionSet_SSE3_X64=39,
+    InstructionSet_SSSE3_X64=40,
+    InstructionSet_SSE41_X64=41,
+    InstructionSet_SSE42_X64=42,
+    InstructionSet_AVX_X64=43,
+    InstructionSet_AVX2_X64=44,
+    InstructionSet_AES_X64=45,
+    InstructionSet_BMI1_X64=46,
+    InstructionSet_BMI2_X64=47,
+    InstructionSet_FMA_X64=48,
+    InstructionSet_LZCNT_X64=49,
+    InstructionSet_PCLMULQDQ_X64=50,
+    InstructionSet_POPCNT_X64=51,
+    InstructionSet_AVXVNNI_X64=52,
+    InstructionSet_MOVBE_X64=53,
+    InstructionSet_X86Serialize_X64=54,
+    InstructionSet_AVX512F_X64=55,
+    InstructionSet_AVX512F_VL_X64=56,
+    InstructionSet_AVX512BW_X64=57,
+    InstructionSet_AVX512BW_VL_X64=58,
+    InstructionSet_AVX512CD_X64=59,
+    InstructionSet_AVX512CD_VL_X64=60,
+    InstructionSet_AVX512DQ_X64=61,
+    InstructionSet_AVX512DQ_VL_X64=62,
+    InstructionSet_AVX512VBMI_X64=63,
+    InstructionSet_AVX512VBMI_VL_X64=64,
 #endif // TARGET_X86
 
 };
 
 struct CORINFO_InstructionSetFlags
 {
 private:
-    static const int32_t FlagsFieldCount = 1;
+    static const int32_t FlagsFieldCount = 2;
     static const int32_t BitsPerFlagsField = sizeof(uint64_t) * 8;
     uint64_t _flags[FlagsFieldCount] = { };
 
@@ -404,6 +411,8 @@ inline CORINFO_InstructionSetFlags EnsureInstructionSetFlagsAreValid(CORINFO_Ins
             resultflags.RemoveInstructionSet(InstructionSet_Vector64);
         if (resultflags.HasInstructionSet(InstructionSet_Vector128) && !resultflags.HasInstructionSet(InstructionSet_AdvSimd))
             resultflags.RemoveInstructionSet(InstructionSet_Vector128);
+        if (resultflags.HasInstructionSet(InstructionSet_VectorT128) && !resultflags.HasInstructionSet(InstructionSet_AdvSimd))
+            resultflags.RemoveInstructionSet(InstructionSet_VectorT128);
 #endif // TARGET_ARM64
 #ifdef TARGET_AMD64
         if (resultflags.HasInstructionSet(InstructionSet_X86Base) && !resultflags.HasInstructionSet(InstructionSet_X86Base_X64))
@@ -594,6 +603,12 @@ inline CORINFO_InstructionSetFlags EnsureInstructionSetFlagsAreValid(CORINFO_Ins
             resultflags.RemoveInstructionSet(InstructionSet_AVX512VBMI_VL);
         if (resultflags.HasInstructionSet(InstructionSet_AVX512VBMI_VL) && !resultflags.HasInstructionSet(InstructionSet_AVX512BW_VL))
             resultflags.RemoveInstructionSet(InstructionSet_AVX512VBMI_VL);
+        if (resultflags.HasInstructionSet(InstructionSet_VectorT128) && !resultflags.HasInstructionSet(InstructionSet_SSE2))
+            resultflags.RemoveInstructionSet(InstructionSet_VectorT128);
+        if (resultflags.HasInstructionSet(InstructionSet_VectorT256) && !resultflags.HasInstructionSet(InstructionSet_AVX2))
+            resultflags.RemoveInstructionSet(InstructionSet_VectorT256);
+        if (resultflags.HasInstructionSet(InstructionSet_VectorT512) && !resultflags.HasInstructionSet(InstructionSet_AVX512F))
+            resultflags.RemoveInstructionSet(InstructionSet_VectorT512);
         if (resultflags.HasInstructionSet(InstructionSet_AVX512F) && !resultflags.HasInstructionSet(InstructionSet_AVX512BW_VL))
             resultflags.RemoveInstructionSet(InstructionSet_AVX512F);
         if (resultflags.HasInstructionSet(InstructionSet_AVX512F) && !resultflags.HasInstructionSet(InstructionSet_AVX512CD_VL))
@@ -674,6 +689,12 @@ inline CORINFO_InstructionSetFlags EnsureInstructionSetFlagsAreValid(CORINFO_Ins
             resultflags.RemoveInstructionSet(InstructionSet_AVX512VBMI_VL);
         if (resultflags.HasInstructionSet(InstructionSet_AVX512VBMI_VL) && !resultflags.HasInstructionSet(InstructionSet_AVX512BW_VL))
             resultflags.RemoveInstructionSet(InstructionSet_AVX512VBMI_VL);
+        if (resultflags.HasInstructionSet(InstructionSet_VectorT128) && !resultflags.HasInstructionSet(InstructionSet_SSE2))
+            resultflags.RemoveInstructionSet(InstructionSet_VectorT128);
+        if (resultflags.HasInstructionSet(InstructionSet_VectorT256) && !resultflags.HasInstructionSet(InstructionSet_AVX2))
+            resultflags.RemoveInstructionSet(InstructionSet_VectorT256);
+        if (resultflags.HasInstructionSet(InstructionSet_VectorT512) && !resultflags.HasInstructionSet(InstructionSet_AVX512F))
+            resultflags.RemoveInstructionSet(InstructionSet_VectorT512);
         if (resultflags.HasInstructionSet(InstructionSet_AVX512F) && !resultflags.HasInstructionSet(InstructionSet_AVX512BW_VL))
             resultflags.RemoveInstructionSet(InstructionSet_AVX512F);
         if (resultflags.HasInstructionSet(InstructionSet_AVX512F) && !resultflags.HasInstructionSet(InstructionSet_AVX512CD_VL))
@@ -738,6 +759,8 @@ inline const char *InstructionSetToString(CORINFO_InstructionSet instructionSet)
             return "Dczva";
         case InstructionSet_Rcpc :
             return "Rcpc";
+        case InstructionSet_VectorT128 :
+            return "VectorT128";
 #endif // TARGET_ARM64
 #ifdef TARGET_AMD64
         case InstructionSet_X86Base :
@@ -862,6 +885,12 @@ inline const char *InstructionSetToString(CORINFO_InstructionSet instructionSet)
             return "AVX512VBMI_VL";
         case InstructionSet_AVX512VBMI_VL_X64 :
             return "AVX512VBMI_VL_X64";
+        case InstructionSet_VectorT128 :
+            return "VectorT128";
+        case InstructionSet_VectorT256 :
+            return "VectorT256";
+        case InstructionSet_VectorT512 :
+            return "VectorT512";
 #endif // TARGET_AMD64
 #ifdef TARGET_X86
         case InstructionSet_X86Base :
@@ -928,6 +957,12 @@ inline const char *InstructionSetToString(CORINFO_InstructionSet instructionSet)
             return "AVX512VBMI";
         case InstructionSet_AVX512VBMI_VL :
             return "AVX512VBMI_VL";
+        case InstructionSet_VectorT128 :
+            return "VectorT128";
+        case InstructionSet_VectorT256 :
+            return "VectorT256";
+        case InstructionSet_VectorT512 :
+            return "VectorT512";
 #endif // TARGET_X86
 
         default:
@@ -958,6 +993,7 @@ inline CORINFO_InstructionSet InstructionSetFromR2RInstructionSet(ReadyToRunInst
         case READYTORUN_INSTRUCTION_Sha256: return InstructionSet_Sha256;
         case READYTORUN_INSTRUCTION_Atomics: return InstructionSet_Atomics;
         case READYTORUN_INSTRUCTION_Rcpc: return InstructionSet_Rcpc;
+        case READYTORUN_INSTRUCTION_VectorT128: return InstructionSet_VectorT128;
 #endif // TARGET_ARM64
 #ifdef TARGET_AMD64
         case READYTORUN_INSTRUCTION_X86Base: return InstructionSet_X86Base;
@@ -989,6 +1025,9 @@ inline CORINFO_InstructionSet InstructionSetFromR2RInstructionSet(ReadyToRunInst
         case READYTORUN_INSTRUCTION_Avx512DQ_VL: return InstructionSet_AVX512DQ_VL;
         case READYTORUN_INSTRUCTION_Avx512Vbmi: return InstructionSet_AVX512VBMI;
         case READYTORUN_INSTRUCTION_Avx512Vbmi_VL: return InstructionSet_AVX512VBMI_VL;
+        case READYTORUN_INSTRUCTION_VectorT128: return InstructionSet_VectorT128;
+        case READYTORUN_INSTRUCTION_VectorT256: return InstructionSet_VectorT256;
+        case READYTORUN_INSTRUCTION_VectorT512: return InstructionSet_VectorT512;
 #endif // TARGET_AMD64
 #ifdef TARGET_X86
         case READYTORUN_INSTRUCTION_X86Base: return InstructionSet_X86Base;
@@ -1020,6 +1059,9 @@ inline CORINFO_InstructionSet InstructionSetFromR2RInstructionSet(ReadyToRunInst
         case READYTORUN_INSTRUCTION_Avx512DQ_VL: return InstructionSet_AVX512DQ_VL;
         case READYTORUN_INSTRUCTION_Avx512Vbmi: return InstructionSet_AVX512VBMI;
         case READYTORUN_INSTRUCTION_Avx512Vbmi_VL: return InstructionSet_AVX512VBMI_VL;
+        case READYTORUN_INSTRUCTION_VectorT128: return InstructionSet_VectorT128;
+        case READYTORUN_INSTRUCTION_VectorT256: return InstructionSet_VectorT256;
+        case READYTORUN_INSTRUCTION_VectorT512: return InstructionSet_VectorT512;
 #endif // TARGET_X86
 
         default:

src/coreclr/inc/jiteeversionguid.h

@@ -43,11 +43,11 @@ typedef const GUID *LPCGUID;
 #define GUID_DEFINED
 #endif // !GUID_DEFINED
 
-constexpr GUID JITEEVersionIdentifier = { /* d4414be1-70e4-46ac-8866-ca3a6c2f8422 */
-    0xd4414be1,
-    0x70e4,
-    0x46ac,
-    {0x88, 0x66, 0xca, 0x3a, 0x6c, 0x2f, 0x84, 0x22}
+constexpr GUID JITEEVersionIdentifier = { /* fda2f9dd-6b3e-4ecd-a7b8-79e5edf1f072 */
+    0xfda2f9dd,
+    0x6b3e,
+    0x4ecd,
+    {0xa7, 0xb8, 0x79, 0xe5, 0xed, 0xf1, 0xf0, 0x72}
   };
 
 //////////////////////////////////////////////////////////////////////////////////////////////////////////

src/coreclr/inc/readytoruninstructionset.h

@@ -47,6 +47,9 @@ enum ReadyToRunInstructionSet
     READYTORUN_INSTRUCTION_Avx512DQ_VL=36,
     READYTORUN_INSTRUCTION_Avx512Vbmi=37,
     READYTORUN_INSTRUCTION_Avx512Vbmi_VL=38,
+    READYTORUN_INSTRUCTION_VectorT128=39,
+    READYTORUN_INSTRUCTION_VectorT256=40,
+    READYTORUN_INSTRUCTION_VectorT512=41,
 
 };
 

src/coreclr/jit/codegenxarch.cpp

@@ -3401,15 +3401,15 @@ void CodeGen::genCodeForCpBlkUnroll(GenTreeBlk* node)
     assert(srcOffset < (INT32_MAX - static_cast<int>(size)));
     assert(dstOffset < (INT32_MAX - static_cast<int>(size)));
 
-    if (size >= XMM_REGSIZE_BYTES)
+    // Get the largest SIMD register available if the size is large enough
+    unsigned regSize = compiler->roundDownSIMDSize(size);
+
+    if ((size >= regSize) && (regSize > 0))
     {
         regNumber tempReg = node->GetSingleTempReg(RBM_ALLFLOAT);
 
         instruction simdMov = simdUnalignedMovIns();
 
-        // Get the largest SIMD register available if the size is large enough
-        unsigned regSize = compiler->roundDownSIMDSize(size);
-
         auto emitSimdMovs = [&]() {
             if (srcLclNum != BAD_VAR_NUM)
             {