Updating CreateScalar to be intrinsic for Vector64/128/256 (#77798)

* Updating CreateScalar to be intrinsic for Vector64/128/256

* Applying formatting patch

* Fixing CreateScalar VecCon nodes created on import and find use before insert

* Applying formatting patch

* Ensure we zero extend TYP_BYTE and TYP_SHORT

* Ensure TYP_UBYTE and TYP_USHORT are also explicitly zero extended

* Fix the cast and add a comment explaining "why"

* Applying formatting patch

11 files changed, 407 insertions, 294 deletions

diff --git a/src/coreclr/jit/gentree.cpp b/src/coreclr/jit/gentree.cpp
index 8535f0e90c5..186fcabe5fb 100644
--- a/src/coreclr/jit/gentree.cpp
+++ b/src/coreclr/jit/gentree.cpp
@@ -17362,32 +17362,48 @@ bool GenTreeIntConCommon::AddrNeedsReloc(Compiler* comp)
 //     true if node represents a constant; otherwise, false
 bool GenTreeVecCon::IsHWIntrinsicCreateConstant(GenTreeHWIntrinsic* node, simd32_t& simd32Val)
 {
-    var_types simdType     = node->TypeGet();
-    var_types simdBaseType = node->GetSimdBaseType();
-    unsigned  simdSize     = node->GetSimdSize();
+    NamedIntrinsic intrinsic    = node->GetHWIntrinsicId();
+    var_types      simdType     = node->TypeGet();
+    var_types      simdBaseType = node->GetSimdBaseType();
+    unsigned       simdSize     = node->GetSimdSize();
 
     size_t argCnt    = node->GetOperandCount();
     size_t cnsArgCnt = 0;
 
-    switch (node->GetHWIntrinsicId())
+    switch (intrinsic)
     {
         case NI_Vector128_Create:
+        case NI_Vector128_CreateScalar:
         case NI_Vector128_CreateScalarUnsafe:
 #if defined(TARGET_XARCH)
         case NI_Vector256_Create:
+        case NI_Vector256_CreateScalar:
         case NI_Vector256_CreateScalarUnsafe:
 #elif defined(TARGET_ARM64)
         case NI_Vector64_Create:
+        case NI_Vector64_CreateScalar:
         case NI_Vector64_CreateScalarUnsafe:
 #endif
         {
+            // Zero out the simd32Val
+            simd32Val = {};
+
             // These intrinsics are meant to set the same value to every element.
             if ((argCnt == 1) && HandleArgForHWIntrinsicCreate(node->Op(1), 0, simd32Val, simdBaseType))
             {
-                // Now assign the rest of the arguments.
-                for (unsigned i = 1; i < simdSize / genTypeSize(simdBaseType); i++)
+// CreateScalar leaves the upper bits as zero
+
+#if defined(TARGET_XARCH)
+                if ((intrinsic != NI_Vector128_CreateScalar) && (intrinsic != NI_Vector256_CreateScalar))
+#elif defined(TARGET_ARM64)
+                if ((intrinsic != NI_Vector64_CreateScalar) && (intrinsic != NI_Vector128_CreateScalar))
+#endif
                 {
-                    HandleArgForHWIntrinsicCreate(node->Op(1), i, simd32Val, simdBaseType);
+                    // Now assign the rest of the arguments.
+                    for (unsigned i = 1; i < simdSize / genTypeSize(simdBaseType); i++)
+                    {
+                        HandleArgForHWIntrinsicCreate(node->Op(1), i, simd32Val, simdBaseType);
+                    }
                 }
 
                 cnsArgCnt = 1;
@@ -18974,6 +18990,13 @@ bool GenTree::isContainableHWIntrinsic() const
             return true;
         }
 
+        case NI_Vector128_get_Zero:
+        case NI_Vector256_get_Zero:
+        {
+            // These HWIntrinsic operations are contained as part of Sse41.Insert
+            return true;
+        }
+
         default:
         {
             return false;
diff --git a/src/coreclr/jit/hwintrinsicarm64.cpp b/src/coreclr/jit/hwintrinsicarm64.cpp
index 7b966f0a56d..e4653df3e6d 100644
--- a/src/coreclr/jit/hwintrinsicarm64.cpp
+++ b/src/coreclr/jit/hwintrinsicarm64.cpp
@@ -545,6 +545,7 @@ GenTree* Compiler::impSpecialIntrinsic(NamedIntrinsic        intrinsic,
             break;
         }
 
+        case NI_Vector64_CreateScalar:
         case NI_Vector64_CreateScalarUnsafe:
         {
             if (genTypeSize(simdBaseType) == 8)
@@ -556,12 +557,14 @@ GenTree* Compiler::impSpecialIntrinsic(NamedIntrinsic        intrinsic,
 
         case NI_Vector64_Create:
         case NI_Vector128_Create:
+        case NI_Vector128_CreateScalar:
         case NI_Vector128_CreateScalarUnsafe:
         {
             uint32_t simdLength = getSIMDVectorLength(simdSize, simdBaseType);
             assert((sig->numArgs == 1) || (sig->numArgs == simdLength));
 
-            bool isConstant = true;
+            bool isConstant     = true;
+            bool isCreateScalar = (intrinsic == NI_Vector64_CreateScalar) || (intrinsic == NI_Vector128_CreateScalar);
 
             if (varTypeIsFloating(simdBaseType))
             {
@@ -620,7 +623,12 @@ GenTree* Compiler::impSpecialIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd16Val.u8[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val       = {};
+                            vecCon->gtSimd32Val.u8[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < simdLength - 1; index++)
                             {
@@ -641,7 +649,12 @@ GenTree* Compiler::impSpecialIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd16Val.u16[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val        = {};
+                            vecCon->gtSimd32Val.u16[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < (simdLength - 1); index++)
                             {
@@ -662,7 +675,12 @@ GenTree* Compiler::impSpecialIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd16Val.u32[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val        = {};
+                            vecCon->gtSimd32Val.u32[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < (simdLength - 1); index++)
                             {
@@ -683,7 +701,12 @@ GenTree* Compiler::impSpecialIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd16Val.u64[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val        = {};
+                            vecCon->gtSimd32Val.u64[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < (simdLength - 1); index++)
                             {
@@ -703,7 +726,12 @@ GenTree* Compiler::impSpecialIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd16Val.f32[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val        = {};
+                            vecCon->gtSimd32Val.f32[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < (simdLength - 1); index++)
                             {
@@ -723,7 +751,12 @@ GenTree* Compiler::impSpecialIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd16Val.f64[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val        = {};
+                            vecCon->gtSimd32Val.f64[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < (simdLength - 1); index++)
                             {
diff --git a/src/coreclr/jit/hwintrinsiccodegenxarch.cpp b/src/coreclr/jit/hwintrinsiccodegenxarch.cpp
index 8bd12f54d8d..b2b08c8d828 100644
--- a/src/coreclr/jit/hwintrinsiccodegenxarch.cpp
+++ b/src/coreclr/jit/hwintrinsiccodegenxarch.cpp
@@ -596,6 +596,13 @@ void CodeGen::genHWIntrinsic_R_R_RM_I(GenTreeHWIntrinsic* node, instruction ins,
 
     regNumber op1Reg = op1->GetRegNum();
 
+    if ((ins == INS_insertps) && (op1Reg == REG_NA))
+    {
+        // insertps is special and can contain op1 when it is zero
+        assert(op1->isContained() && op1->IsVectorZero());
+        op1Reg = targetReg;
+    }
+
     assert(targetReg != REG_NA);
     assert(op1Reg != REG_NA);
 
diff --git a/src/coreclr/jit/hwintrinsiclistarm64.h b/src/coreclr/jit/hwintrinsiclistarm64.h
index 92c1c0848b6..cbe956299d1 100644
--- a/src/coreclr/jit/hwintrinsiclistarm64.h
+++ b/src/coreclr/jit/hwintrinsiclistarm64.h
@@ -39,7 +39,8 @@ HARDWARE_INTRINSIC(Vector64,      ConvertToInt64,
 HARDWARE_INTRINSIC(Vector64,      ConvertToSingle,                                                   8,      1,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
 HARDWARE_INTRINSIC(Vector64,      ConvertToUInt32,                                                   8,      1,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
 HARDWARE_INTRINSIC(Vector64,      ConvertToUInt64,                                                   8,      1,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
-HARDWARE_INTRINSIC(Vector64,      Create,                                                            8,     -1,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_mov,            INS_mov,            INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
+HARDWARE_INTRINSIC(Vector64,      Create,                                                            8,     -1,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
+HARDWARE_INTRINSIC(Vector64,      CreateScalar,                                                      8,     -1,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
 HARDWARE_INTRINSIC(Vector64,      CreateScalarUnsafe,                                                8,      1,     {INS_ins,            INS_ins,            INS_ins,            INS_ins,            INS_ins,            INS_ins,            INS_invalid,        INS_invalid,        INS_fmov,           INS_invalid},     HW_Category_SIMD,                  HW_Flag_SpecialImport|HW_Flag_SpecialCodeGen|HW_Flag_SupportsContainment)
 HARDWARE_INTRINSIC(Vector64,      Divide,                                                            8,      2,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
 HARDWARE_INTRINSIC(Vector64,      Dot,                                                               8,      2,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_BaseTypeFromFirstArg|HW_Flag_NoCodeGen)
@@ -143,6 +144,7 @@ HARDWARE_INTRINSIC(Vector128,     ConvertToSingle,
 HARDWARE_INTRINSIC(Vector128,     ConvertToUInt32,                                                  16,      1,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
 HARDWARE_INTRINSIC(Vector128,     ConvertToUInt64,                                                  16,      1,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
 HARDWARE_INTRINSIC(Vector128,     Create,                                                           16,     -1,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
+HARDWARE_INTRINSIC(Vector128,     CreateScalar,                                                     16,     -1,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
 HARDWARE_INTRINSIC(Vector128,     CreateScalarUnsafe,                                               16,      1,     {INS_ins,            INS_ins,            INS_ins,            INS_ins,            INS_ins,            INS_ins,            INS_ins,            INS_ins,            INS_fmov,           INS_fmov},        HW_Category_SIMD,                  HW_Flag_SpecialImport|HW_Flag_SpecialCodeGen|HW_Flag_SupportsContainment)
 HARDWARE_INTRINSIC(Vector128,     Divide,                                                           16,      2,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
 HARDWARE_INTRINSIC(Vector128,     Dot,                                                              16,      2,     {INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid,        INS_invalid},     HW_Category_Helper,                HW_Flag_SpecialImport|HW_Flag_BaseTypeFromFirstArg|HW_Flag_NoCodeGen)
diff --git a/src/coreclr/jit/hwintrinsiclistxarch.h b/src/coreclr/jit/hwintrinsiclistxarch.h
index 37738201789..cb2610315ec 100644
--- a/src/coreclr/jit/hwintrinsiclistxarch.h
+++ b/src/coreclr/jit/hwintrinsiclistxarch.h
@@ -57,6 +57,7 @@ HARDWARE_INTRINSIC(Vector128,       ConvertToSingle,
 HARDWARE_INTRINSIC(Vector128,       ConvertToUInt32,                            16,             1,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
 HARDWARE_INTRINSIC(Vector128,       ConvertToUInt64,                            16,             1,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
 HARDWARE_INTRINSIC(Vector128,       Create,                                     16,            -1,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
+HARDWARE_INTRINSIC(Vector128,       CreateScalar,                               16,             1,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
 HARDWARE_INTRINSIC(Vector128,       CreateScalarUnsafe,                         16,             1,      {INS_movd,              INS_movd,               INS_movd,               INS_movd,               INS_movd,               INS_movd,               INS_movd,               INS_movd,               INS_movss,              INS_movsdsse2},         HW_Category_SIMDScalar,             HW_Flag_SpecialImport|HW_Flag_SpecialCodeGen|HW_Flag_NoRMWSemantics)
 HARDWARE_INTRINSIC(Vector128,       Divide,                                     16,             2,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
 HARDWARE_INTRINSIC(Vector128,       Dot,                                        16,             2,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_BaseTypeFromFirstArg|HW_Flag_NoCodeGen)
@@ -158,6 +159,7 @@ HARDWARE_INTRINSIC(Vector256,       ConvertToSingle,
 HARDWARE_INTRINSIC(Vector256,       ConvertToUInt32,                            32,             1,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
 HARDWARE_INTRINSIC(Vector256,       ConvertToUInt64,                            32,             1,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_BaseTypeFromFirstArg)
 HARDWARE_INTRINSIC(Vector256,       Create,                                     32,            -1,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_AvxOnlyCompatible)
+HARDWARE_INTRINSIC(Vector256,       CreateScalar,                               32,            -1,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_NoCodeGen|HW_Flag_AvxOnlyCompatible)
 HARDWARE_INTRINSIC(Vector256,       CreateScalarUnsafe,                         32,             1,      {INS_movd,              INS_movd,               INS_movd,               INS_movd,               INS_movd,               INS_movd,               INS_movd,               INS_movd,               INS_movss,              INS_movsdsse2},         HW_Category_SIMDScalar,             HW_Flag_SpecialImport|HW_Flag_SpecialCodeGen|HW_Flag_NoRMWSemantics|HW_Flag_AvxOnlyCompatible)
 HARDWARE_INTRINSIC(Vector256,       Divide,                                     32,             2,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_NoCodeGen)
 HARDWARE_INTRINSIC(Vector256,       Dot,                                        32,             2,      {INS_invalid,           INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid,            INS_invalid},           HW_Category_Helper,                 HW_Flag_SpecialImport|HW_Flag_BaseTypeFromFirstArg|HW_Flag_NoCodeGen)
diff --git a/src/coreclr/jit/hwintrinsicxarch.cpp b/src/coreclr/jit/hwintrinsicxarch.cpp
index 04bab108a2e..8eea6cb6c1a 100644
--- a/src/coreclr/jit/hwintrinsicxarch.cpp
+++ b/src/coreclr/jit/hwintrinsicxarch.cpp
@@ -899,13 +899,16 @@ GenTree* Compiler::impBaseIntrinsic(NamedIntrinsic        intrinsic,
 
         case NI_Vector128_Create:
         case NI_Vector256_Create:
+        case NI_Vector128_CreateScalar:
+        case NI_Vector256_CreateScalar:
         case NI_Vector128_CreateScalarUnsafe:
         case NI_Vector256_CreateScalarUnsafe:
         {
             uint32_t simdLength = getSIMDVectorLength(simdSize, simdBaseType);
             assert((sig->numArgs == 1) || (sig->numArgs == simdLength));
 
-            bool isConstant = true;
+            bool isConstant     = true;
+            bool isCreateScalar = (intrinsic == NI_Vector128_CreateScalar) || (intrinsic == NI_Vector256_CreateScalar);
 
             if (varTypeIsFloating(simdBaseType))
             {
@@ -964,7 +967,12 @@ GenTree* Compiler::impBaseIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd32Val.u8[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val       = {};
+                            vecCon->gtSimd32Val.u8[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < simdLength - 1; index++)
                             {
@@ -985,7 +993,12 @@ GenTree* Compiler::impBaseIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd32Val.u16[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val        = {};
+                            vecCon->gtSimd32Val.u16[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < (simdLength - 1); index++)
                             {
@@ -1006,7 +1019,12 @@ GenTree* Compiler::impBaseIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd32Val.u32[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val        = {};
+                            vecCon->gtSimd32Val.u32[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < (simdLength - 1); index++)
                             {
@@ -1027,7 +1045,12 @@ GenTree* Compiler::impBaseIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd32Val.u64[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val        = {};
+                            vecCon->gtSimd32Val.u64[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < (simdLength - 1); index++)
                             {
@@ -1047,7 +1070,12 @@ GenTree* Compiler::impBaseIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd32Val.f32[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val        = {};
+                            vecCon->gtSimd32Val.f32[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < (simdLength - 1); index++)
                             {
@@ -1067,7 +1095,12 @@ GenTree* Compiler::impBaseIntrinsic(NamedIntrinsic        intrinsic,
                             vecCon->gtSimd32Val.f64[simdLength - 1 - index] = cnsVal;
                         }
 
-                        if (sig->numArgs == 1)
+                        if (isCreateScalar)
+                        {
+                            vecCon->gtSimd32Val        = {};
+                            vecCon->gtSimd32Val.f64[0] = cnsVal;
+                        }
+                        else if (sig->numArgs == 1)
                         {
                             for (uint32_t index = 0; index < (simdLength - 1); index++)
                             {
diff --git a/src/coreclr/jit/lowerarmarch.cpp b/src/coreclr/jit/lowerarmarch.cpp
index 1401de10b47..7f988dabad6 100644
--- a/src/coreclr/jit/lowerarmarch.cpp
+++ b/src/coreclr/jit/lowerarmarch.cpp
@@ -1100,6 +1100,8 @@ GenTree* Lowering::LowerHWIntrinsic(GenTreeHWIntrinsic* node)
     {
         case NI_Vector64_Create:
         case NI_Vector128_Create:
+        case NI_Vector64_CreateScalar:
+        case NI_Vector128_CreateScalar:
         {
             // We don't directly support the Vector64.Create or Vector128.Create methods in codegen
             // and instead lower them to other intrinsic nodes in LowerHWIntrinsicCreate so we expect
@@ -1439,8 +1441,9 @@ GenTree* Lowering::LowerHWIntrinsicCreate(GenTreeHWIntrinsic* node)
     assert(varTypeIsArithmetic(simdBaseType));
     assert(simdSize != 0);
 
-    bool   isConstant = GenTreeVecCon::IsHWIntrinsicCreateConstant(node, simd32Val);
-    size_t argCnt     = node->GetOperandCount();
+    bool   isConstant     = GenTreeVecCon::IsHWIntrinsicCreateConstant(node, simd32Val);
+    bool   isCreateScalar = (intrinsicId == NI_Vector64_CreateScalar) || (intrinsicId == NI_Vector128_CreateScalar);
+    size_t argCnt         = node->GetOperandCount();
 
     // Check if we have a cast that we can remove. Note that "IsValidConstForMovImm"
     // will reset Op(1) if it finds such a cast, so we do not need to handle it here.
@@ -1482,6 +1485,22 @@ GenTree* Lowering::LowerHWIntrinsicCreate(GenTreeHWIntrinsic* node)
     }
     else if (argCnt == 1)
     {
+        if (isCreateScalar)
+        {
+            GenTree* op1 = node->Op(1);
+
+            GenTree* tmp = comp->gtNewZeroConNode(simdType);
+            BlockRange().InsertBefore(op1, tmp);
+            LowerNode(tmp);
+
+            GenTree* idx = comp->gtNewIconNode(0);
+            BlockRange().InsertAfter(tmp, idx);
+            LowerNode(idx);
+
+            node->ResetHWIntrinsicId(NI_AdvSimd_Insert, comp, tmp, idx, op1);
+            return LowerNode(node);
+        }
+
         // We have the following (where simd is simd8 or simd16):
         //          /--*  op1  T
         //   node = *  HWINTRINSIC   simd   T Create
diff --git a/src/coreclr/jit/lowerxarch.cpp b/src/coreclr/jit/lowerxarch.cpp
index 44d6781d081..40ccaebec0f 100644
--- a/src/coreclr/jit/lowerxarch.cpp
+++ b/src/coreclr/jit/lowerxarch.cpp
@@ -1023,6 +1023,8 @@ GenTree* Lowering::LowerHWIntrinsic(GenTreeHWIntrinsic* node)
 
         case NI_Vector128_Create:
         case NI_Vector256_Create:
+        case NI_Vector128_CreateScalar:
+        case NI_Vector256_CreateScalar:
         {
             // We don't directly support the Vector128.Create or Vector256.Create methods in codegen
             // and instead lower them to other intrinsic nodes in LowerHWIntrinsicCreate so we expect
@@ -1705,8 +1707,9 @@ GenTree* Lowering::LowerHWIntrinsicCreate(GenTreeHWIntrinsic* node)
     GenTree* tmp2 = nullptr;
     GenTree* tmp3 = nullptr;
 
-    bool   isConstant = GenTreeVecCon::IsHWIntrinsicCreateConstant(node, simd32Val);
-    size_t argCnt     = node->GetOperandCount();
+    bool   isConstant     = GenTreeVecCon::IsHWIntrinsicCreateConstant(node, simd32Val);
+    bool   isCreateScalar = (intrinsicId == NI_Vector128_CreateScalar) || (intrinsicId == NI_Vector256_CreateScalar);
+    size_t argCnt         = node->GetOperandCount();
 
     if (isConstant)
     {
@@ -1745,6 +1748,156 @@ GenTree* Lowering::LowerHWIntrinsicCreate(GenTreeHWIntrinsic* node)
     }
     else if (argCnt == 1)
     {
+        if (isCreateScalar)
+        {
+            switch (simdBaseType)
+            {
+                case TYP_BYTE:
+                case TYP_UBYTE:
+                {
+                    // Types need to be explicitly zero-extended to ensure upper-bits are zero
+                    //
+                    // We need to explicitly use TYP_UBYTE since unsigned is ignored for small types
+                    // Explicitly handle both BYTE and UBYTE to account for reinterpret casts and the like
+                    //
+                    // The from type is INT since that is the input type tracked by IR, where-as the target
+                    // type needs to be UBYTE so it implicitly zero-extends back to TYP_INT
+
+                    tmp1 = comp->gtNewCastNode(TYP_INT, op1, /* unsigned */ true, TYP_UBYTE);
+                    BlockRange().InsertAfter(op1, tmp1);
+                    LowerNode(tmp1);
+
+                    node->ChangeHWIntrinsicId(NI_SSE2_ConvertScalarToVector128Int32, tmp1);
+                    node->SetSimdBaseJitType(CORINFO_TYPE_INT);
+                    break;
+                }
+
+                case TYP_SHORT:
+                case TYP_USHORT:
+                {
+                    // Types need to be explicitly zero-extended to ensure upper-bits are zero
+                    //
+                    // We need to explicitly use TYP_USHORT since unsigned is ignored for small types
+                    // Explicitly handle both SHORT and USHORT to account for reinterpret casts and the like
+                    //
+                    // The from type is INT since that is the input type tracked by IR, where-as the target
+                    // type needs to be USHORT so it implicitly zero-extends back to TYP_INT
+
+                    tmp1 = comp->gtNewCastNode(TYP_INT, op1, /* unsigned */ true, TYP_USHORT);
+                    BlockRange().InsertAfter(op1, tmp1);
+                    LowerNode(tmp1);
+
+                    node->ChangeHWIntrinsicId(NI_SSE2_ConvertScalarToVector128Int32, tmp1);
+                    node->SetSimdBaseJitType(CORINFO_TYPE_INT);
+                    break;
+                }
+
+                case TYP_INT:
+                {
+                    node->ChangeHWIntrinsicId(NI_SSE2_ConvertScalarToVector128Int32);
+                    break;
+                }
+
+                case TYP_UINT:
+                {
+                    node->ChangeHWIntrinsicId(NI_SSE2_ConvertScalarToVector128UInt32);
+                    break;
+                }
+
+#if defined(TARGET_AMD64)
+                case TYP_LONG:
+                {
+                    node->ChangeHWIntrinsicId(NI_SSE2_X64_ConvertScalarToVector128Int64);
+                    break;
+                }
+
+                case TYP_ULONG:
+                {
+                    node->ChangeHWIntrinsicId(NI_SSE2_X64_ConvertScalarToVector128UInt64);
+                    break;
+                }
+#endif // TARGET_AMD64
+
+                case TYP_FLOAT:
+                {
+                    tmp1 = comp->gtNewZeroConNode(simdType);
+                    BlockRange().InsertBefore(op1, tmp1);
+                    LowerNode(tmp1);
+
+                    if (comp->compOpportunisticallyDependsOn(InstructionSet_SSE41))
+                    {
+                        // Sse41.Insert has:
+                        //  * Bits 0-3: zmask
+                        //  * Bits 4-5: count_d
+                        //  * Bits 6-7: count_s (register form only)
+                        //
+                        // We want zmask 0b1110 (0xE) to zero elements 1/2/3
+                        // We want count_d 0b00 (0x0) to insert the value to element 0
+                        // We want count_s 0b00 (0x0) as we're just taking element 0 of the source
+
+                        idx = comp->gtNewIconNode(0x0E);
+                        BlockRange().InsertAfter(op1, idx);
+                        LowerNode(idx);
+
+                        node->ResetHWIntrinsicId(NI_SSE41_Insert, comp, tmp1, op1, idx);
+                    }
+                    else
+                    {
+                        node->ResetHWIntrinsicId(NI_SSE_MoveScalar, comp, tmp1, op1);
+                    }
+                    break;
+                }
+
+                case TYP_DOUBLE:
+                {
+                    tmp1 = comp->gtNewZeroConNode(simdType);
+                    BlockRange().InsertBefore(op1, tmp1);
+                    LowerNode(tmp1);
+
+                    node->ResetHWIntrinsicId(NI_SSE2_MoveScalar, comp, tmp1, op1);
+                    break;
+                }
+
+                default:
+                {
+                    unreached();
+                }
+            }
+
+            if (simdSize == 32)
+            {
+                // We're creating a Vector256 scalar so we need to treat the original op as Vector128,
+                // we need to unsafely extend up to Vector256 (which is actually safe since the 128-bit
+                // op will zero extend up to 256-bits), and then we need to replace the original use
+                // with the new TYP_SIMD32 node.
+
+                node->ChangeType(TYP_SIMD16);
+                node->SetSimdSize(16);
+                LowerNode(node);
+
+                tmp2 = comp->gtNewSimdHWIntrinsicNode(TYP_SIMD32, node, NI_Vector128_ToVector256Unsafe, simdBaseJitType,
+                                                      16);
+
+                LIR::Use use;
+                bool     foundUse = BlockRange().TryGetUse(node, &use);
+                BlockRange().InsertAfter(node, tmp2);
+
+                if (foundUse)
+                {
+                    use.ReplaceWith(tmp2);
+                }
+                else
+                {
+                    node->ClearUnusedValue();
+                    tmp2->SetUnusedValue();
+                }
+
+                node = tmp2->AsHWIntrinsic();
+            }
+
+            return LowerNode(node);
+        }
+
         // We have the following (where simd is simd16 or simd32):
         //          /--*  op1  T
         //   node = *  HWINTRINSIC   simd   T Create
@@ -1822,6 +1975,8 @@ GenTree* Lowering::LowerHWIntrinsicCreate(GenTreeHWIntrinsic* node)
             return LowerNode(node);
         }
 
+        assert(intrinsicId == NI_Vector128_Create);
+
         // We will be constructing the following parts:
         //          /--*  op1  T
         //   tmp1 = *  HWINTRINSIC   simd16 T CreateScalarUnsafe
@@ -6940,7 +7095,6 @@ void Lowering::ContainCheckHWIntrinsic(GenTreeHWIntrinsic* node)
                         case NI_SSSE3_AlignRight:
                         case NI_SSE41_Blend:
                         case NI_SSE41_DotProduct:
-                        case NI_SSE41_Insert:
                         case NI_SSE41_X64_Insert:
                         case NI_SSE41_MultipleSumAbsoluteDifferences:
                         case NI_AVX_Blend:
@@ -6968,6 +7122,53 @@ void Lowering::ContainCheckHWIntrinsic(GenTreeHWIntrinsic* node)
                             break;
                         }
 
+                        case NI_SSE41_Insert:
+                        {
+                            GenTree* lastOp = node->Op(numArgs);
+
+                            if ((simdBaseType == TYP_FLOAT) && lastOp->IsCnsIntOrI())
+                            {
+                                // Sse41.Insert has:
+                                //  * Bits 0-3: zmask
+                                //  * Bits 4-5: count_d
+                                //  * Bits 6-7: count_s (register form only)
+                                //
+                                // Where zmask specifies which elements to zero
+                                // Where count_d specifies the destination index the value is being inserted to
+                                // Where count_s specifies the source index of the value being inserted
+
+                                ssize_t ival = lastOp->AsIntConCommon()->IconValue();
+
+                                ssize_t zmask   = (ival & 0x0F);
+                                ssize_t count_d = (ival & 0x30) >> 4;
+                                ssize_t count_s = (ival & 0xC0) >> 6;
+
+                                if (op1->IsVectorZero())
+                                {
+                                    // When op1 is zero, we can contain op1 and modify the mask
+                                    // to zero everything except for the element we're inserting to
+
+                                    MakeSrcContained(node, op1);
+
+                                    zmask |= ~(1 << count_d);
+                                    zmask &= 0x0F;
+
+                                    ival = (count_s << 6) | (count_d << 4) | (zmask);
+                                    lastOp->AsIntConCommon()->SetIconValue(ival);
+                                }
+                            }
+
+                            if (TryGetContainableHWIntrinsicOp(node, &op2, &supportsRegOptional))
+                            {
+                                MakeSrcContained(node, op2);
+                            }
+                            else if (supportsRegOptional)
+                            {
+                                op2->SetRegOptional();
+                            }
+                            break;
+                        }
+
                         default:
                         {
                             assert(!"Unhandled containment for ternary hardware intrinsic with immediate operand");
diff --git a/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector128.cs b/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector128.cs
index 98d99645709..a1d433731df 100644
--- a/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector128.cs
+++ b/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector128.cs
@@ -1156,6 +1156,7 @@ namespace System.Runtime.Intrinsics
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{T}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
         /// <exception cref="NotSupportedException">The type of <paramref name="value" /> (<typeparamref name="T" />) is not supported.</exception>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static unsafe Vector128<T> CreateScalar<T>(T value)
             where T : struct => Vector64.CreateScalar(value).ToVector128();
@@ -1163,241 +1164,91 @@ namespace System.Runtime.Intrinsics
         /// <summary>Creates a new <see cref="Vector128{Byte}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{Byte}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<byte> CreateScalar(byte value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector128<byte>.Zero, 0, value);
-            }
-            else if (Sse2.IsSupported)
-            {
-                // ConvertScalarToVector128 only deals with 32/64-bit inputs and we need to ensure all upper-bits are zeroed, so we call
-                // the UInt32 overload to ensure zero extension. We can then just treat the result as byte and return.
-                return Sse2.ConvertScalarToVector128UInt32(value).AsByte();
-            }
-            else
-            {
-                return CreateScalar<byte>(value);
-            }
-        }
+        public static unsafe Vector128<byte> CreateScalar(byte value) => CreateScalar<byte>(value);
 
         /// <summary>Creates a new <see cref="Vector128{Double}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{Double}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<double> CreateScalar(double value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector128<double>.Zero, 0, value);
-            }
-            else if (Sse2.IsSupported)
-            {
-                return Sse2.MoveScalar(Vector128<double>.Zero, CreateScalarUnsafe(value));
-            }
-            else
-            {
-                return CreateScalar<double>(value);
-            }
-        }
+        public static unsafe Vector128<double> CreateScalar(double value) => CreateScalar<double>(value);
 
         /// <summary>Creates a new <see cref="Vector128{Int16}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{Int16}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<short> CreateScalar(short value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector128<short>.Zero, 0, value);
-            }
-            else if (Sse2.IsSupported)
-            {
-                // ConvertScalarToVector128 only deals with 32/64-bit inputs and we need to ensure all upper-bits are zeroed, so we cast
-                // to ushort and call the UInt32 overload to ensure zero extension. We can then just treat the result as short and return.
-                return Sse2.ConvertScalarToVector128UInt32((ushort)(value)).AsInt16();
-            }
-            else
-            {
-                return CreateScalar<short>(value);
-            }
-        }
+        public static unsafe Vector128<short> CreateScalar(short value) => CreateScalar<short>(value);
 
         /// <summary>Creates a new <see cref="Vector128{Int32}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{Int32}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<int> CreateScalar(int value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector128<int>.Zero, 0, value);
-            }
-            else if (Sse2.IsSupported)
-            {
-                return Sse2.ConvertScalarToVector128Int32(value);
-            }
-            else
-            {
-                return CreateScalar<int>(value);
-            }
-        }
+        public static unsafe Vector128<int> CreateScalar(int value) => CreateScalar<int>(value);
 
         /// <summary>Creates a new <see cref="Vector128{Int64}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{Int64}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<long> CreateScalar(long value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector128<long>.Zero, 0, value);
-            }
-            else if (Sse2.X64.IsSupported)
-            {
-                return Sse2.X64.ConvertScalarToVector128Int64(value);
-            }
-            else
-            {
-                return CreateScalar<long>(value);
-            }
-        }
+        public static unsafe Vector128<long> CreateScalar(long value) => CreateScalar<long>(value);
 
         /// <summary>Creates a new <see cref="Vector128{IntPtr}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{IntPtr}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<nint> CreateScalar(nint value)
-        {
-#if TARGET_64BIT
-            return CreateScalar((long)(value)).AsNInt();
-#else
-            return CreateScalar((int)(value)).AsNInt();
-#endif
-        }
+        public static unsafe Vector128<nint> CreateScalar(nint value) => CreateScalar<nint>(value);
 
         /// <summary>Creates a new <see cref="Vector128{UIntPtr}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{UIntPtr}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<nuint> CreateScalar(nuint value)
-        {
-#if TARGET_64BIT
-            return CreateScalar((ulong)(value)).AsNUInt();
-#else
-            return CreateScalar((uint)(value)).AsNUInt();
-#endif
-        }
+        public static unsafe Vector128<nuint> CreateScalar(nuint value) => CreateScalar<nuint>(value);
 
         /// <summary>Creates a new <see cref="Vector128{SByte}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{SByte}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<sbyte> CreateScalar(sbyte value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector128<sbyte>.Zero, 0, value);
-            }
-            else if (Sse2.IsSupported)
-            {
-                // ConvertScalarToVector128 only deals with 32/64-bit inputs and we need to ensure all upper-bits are zeroed, so we cast
-                // to byte and call the UInt32 overload to ensure zero extension. We can then just treat the result as sbyte and return.
-                return Sse2.ConvertScalarToVector128UInt32((byte)(value)).AsSByte();
-            }
-            else
-            {
-                return CreateScalar<sbyte>(value);
-            }
-        }
+        public static unsafe Vector128<sbyte> CreateScalar(sbyte value) => CreateScalar<sbyte>(value);
 
         /// <summary>Creates a new <see cref="Vector128{Single}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{Single}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<float> CreateScalar(float value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector128<float>.Zero, 0, value);
-            }
-            else if (Sse.IsSupported)
-            {
-                return Sse.MoveScalar(Vector128<float>.Zero, CreateScalarUnsafe(value));
-            }
-            else
-            {
-                return CreateScalar<float>(value);
-            }
-        }
+        public static unsafe Vector128<float> CreateScalar(float value) => CreateScalar<float>(value);
 
         /// <summary>Creates a new <see cref="Vector128{UInt16}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{UInt16}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<ushort> CreateScalar(ushort value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector128<ushort>.Zero, 0, value);
-            }
-            else if (Sse2.IsSupported)
-            {
-                // ConvertScalarToVector128 only deals with 32/64-bit inputs and we need to ensure all upper-bits are zeroed, so we call
-                // the UInt32 overload to ensure zero extension. We can then just treat the result as ushort and return.
-                return Sse2.ConvertScalarToVector128UInt32(value).AsUInt16();
-            }
-            else
-            {
-                return CreateScalar<ushort>(value);
-            }
-        }
+        public static unsafe Vector128<ushort> CreateScalar(ushort value) => CreateScalar<ushort>(value);
 
         /// <summary>Creates a new <see cref="Vector128{UInt32}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{UInt32}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<uint> CreateScalar(uint value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector128<uint>.Zero, 0, value);
-            }
-            else if (Sse2.IsSupported)
-            {
-                return Sse2.ConvertScalarToVector128UInt32(value);
-            }
-            else
-            {
-                return CreateScalar<uint>(value);
-            }
-        }
+        public static unsafe Vector128<uint> CreateScalar(uint value) => CreateScalar<uint>(value);
 
         /// <summary>Creates a new <see cref="Vector128{UInt64}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector128{UInt64}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector128<ulong> CreateScalar(ulong value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector128<ulong>.Zero, 0, value);
-            }
-            else if (Sse2.X64.IsSupported)
-            {
-                return Sse2.X64.ConvertScalarToVector128UInt64(value);
-            }
-            else
-            {
-                return CreateScalar<ulong>(value);
-            }
-        }
+        public static unsafe Vector128<ulong> CreateScalar(ulong value) => CreateScalar<ulong>(value);
 
         /// <summary>Creates a new <see cref="Vector128{T}" /> instance with the first element initialized to the specified value and the remaining elements left uninitialized.</summary>
         /// <typeparam name="T">The type of the elements in the vector.</typeparam>
diff --git a/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector256.cs b/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector256.cs
index 8a6f11b1015..552236ddc4e 100644
--- a/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector256.cs
+++ b/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector256.cs
@@ -1151,6 +1151,7 @@ namespace System.Runtime.Intrinsics
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{T}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
         /// <exception cref="NotSupportedException">The type of <paramref name="value" /> (<typeparamref name="T" />) is not supported.</exception>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<T> CreateScalar<T>(T value)
             where T : struct => Vector128.CreateScalar(value).ToVector256();
@@ -1158,42 +1159,49 @@ namespace System.Runtime.Intrinsics
         /// <summary>Creates a new <see cref="Vector256{Byte}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{Byte}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<byte> CreateScalar(byte value) => CreateScalar<byte>(value);
 
         /// <summary>Creates a new <see cref="Vector256{Double}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{Double}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<double> CreateScalar(double value) => CreateScalar<double>(value);
 
         /// <summary>Creates a new <see cref="Vector256{Int16}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{Int16}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<short> CreateScalar(short value) => CreateScalar<short>(value);
 
         /// <summary>Creates a new <see cref="Vector256{Int32}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{Int32}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<int> CreateScalar(int value) => CreateScalar<int>(value);
 
         /// <summary>Creates a new <see cref="Vector256{Int64}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{Int64}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<long> CreateScalar(long value) => CreateScalar<long>(value);
 
         /// <summary>Creates a new <see cref="Vector256{IntPtr}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{IntPtr}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<nint> CreateScalar(nint value) => CreateScalar<nint>(value);
 
         /// <summary>Creates a new <see cref="Vector256{UIntPtr}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{UIntPtr}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<nuint> CreateScalar(nuint value) => CreateScalar<nuint>(value);
@@ -1201,6 +1209,7 @@ namespace System.Runtime.Intrinsics
         /// <summary>Creates a new <see cref="Vector256{SByte}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{SByte}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<sbyte> CreateScalar(sbyte value) => CreateScalar<sbyte>(value);
@@ -1208,12 +1217,14 @@ namespace System.Runtime.Intrinsics
         /// <summary>Creates a new <see cref="Vector256{Single}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{Single}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<float> CreateScalar(float value) => CreateScalar<float>(value);
 
         /// <summary>Creates a new <see cref="Vector256{UInt16}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{UInt16}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<ushort> CreateScalar(ushort value) => CreateScalar<ushort>(value);
@@ -1221,6 +1232,7 @@ namespace System.Runtime.Intrinsics
         /// <summary>Creates a new <see cref="Vector256{UInt32}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{UInt32}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<uint> CreateScalar(uint value) => CreateScalar<uint>(value);
@@ -1228,6 +1240,7 @@ namespace System.Runtime.Intrinsics
         /// <summary>Creates a new <see cref="Vector256{UInt64}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector256{UInt64}" /> instance with the first element initialized to <paramref name="value" /> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Vector256<ulong> CreateScalar(ulong value) => CreateScalar<ulong>(value);
diff --git a/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector64.cs b/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector64.cs
index dad208aeb29..6a897cd9036 100644
--- a/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector64.cs
+++ b/src/libraries/System.Private.CoreLib/src/System/Runtime/Intrinsics/Vector64.cs
@@ -859,6 +859,7 @@ namespace System.Runtime.Intrinsics
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{Byte}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
         /// <exception cref="NotSupportedException">The type of <paramref name="value" /> (<typeparamref name="T" />) is not supported.</exception>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static unsafe Vector64<T> CreateScalar<T>(T value)
             where T : struct
@@ -871,160 +872,88 @@ namespace System.Runtime.Intrinsics
         /// <summary>Creates a new <see cref="Vector64{Byte}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{Byte}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector64<byte> CreateScalar(byte value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector64<byte>.Zero, 0, value);
-            }
-            else
-            {
-                return CreateScalar<byte>(value);
-            }
-        }
+        public static unsafe Vector64<byte> CreateScalar(byte value) => CreateScalar<byte>(value);
 
         /// <summary>Creates a new <see cref="Vector64{Double}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{Double}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static unsafe Vector64<double> CreateScalar(double value) => CreateScalar<double>(value);
 
         /// <summary>Creates a new <see cref="Vector64{Int16}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{Int16}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector64<short> CreateScalar(short value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector64<short>.Zero, 0, value);
-            }
-            else
-            {
-                return CreateScalar<short>(value);
-            }
-        }
+        public static unsafe Vector64<short> CreateScalar(short value) => CreateScalar<short>(value);
 
         /// <summary>Creates a new <see cref="Vector64{Int32}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{Int32}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector64<int> CreateScalar(int value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector64<int>.Zero, 0, value);
-            }
-            else
-            {
-                return CreateScalar<int>(value);
-            }
-        }
+        public static unsafe Vector64<int> CreateScalar(int value) => CreateScalar<int>(value);
 
         /// <summary>Creates a new <see cref="Vector64{Int64}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{Int64}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static unsafe Vector64<long> CreateScalar(long value) => CreateScalar<long>(value);
 
         /// <summary>Creates a new <see cref="Vector64{IntPtr}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{IntPtr}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector64<nint> CreateScalar(nint value)
-        {
-#if TARGET_64BIT
-            return CreateScalar((long)(value)).AsNInt();
-#else
-            return CreateScalar((int)(value)).AsNInt();
-#endif
-        }
+        public static unsafe Vector64<nint> CreateScalar(nint value) => CreateScalar<nint>(value);
 
         /// <summary>Creates a new <see cref="Vector64{UIntPtr}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{UIntPtr}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector64<nuint> CreateScalar(nuint value)
-        {
-#if TARGET_64BIT
-            return CreateScalar((ulong)(value)).AsNUInt();
-#else
-            return CreateScalar((uint)(value)).AsNUInt();
-#endif
-        }
+        public static unsafe Vector64<nuint> CreateScalar(nuint value) => CreateScalar<nuint>(value);
 
         /// <summary>Creates a new <see cref="Vector64{SByte}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{SByte}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector64<sbyte> CreateScalar(sbyte value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector64<sbyte>.Zero, 0, value);
-            }
-            else
-            {
-                return CreateScalar<sbyte>(value);
-            }
-        }
+        public static unsafe Vector64<sbyte> CreateScalar(sbyte value) => CreateScalar<sbyte>(value);
 
         /// <summary>Creates a new <see cref="Vector64{Single}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{Single}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector64<float> CreateScalar(float value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector64<float>.Zero, 0, value);
-            }
-            else
-            {
-                return CreateScalar<float>(value);
-            }
-        }
+        public static unsafe Vector64<float> CreateScalar(float value) => CreateScalar<float>(value);
 
         /// <summary>Creates a new <see cref="Vector64{UInt16}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{UInt16}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector64<ushort> CreateScalar(ushort value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector64<ushort>.Zero, 0, value);
-            }
-            else
-            {
-                return CreateScalar<ushort>(value);
-            }
-        }
+        public static unsafe Vector64<ushort> CreateScalar(ushort value) => CreateScalar<ushort>(value);
 
         /// <summary>Creates a new <see cref="Vector64{UInt32}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{UInt32}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
-        public static unsafe Vector64<uint> CreateScalar(uint value)
-        {
-            if (AdvSimd.IsSupported)
-            {
-                return AdvSimd.Insert(Vector64<uint>.Zero, 0, value);
-            }
-            else
-            {
-                return CreateScalar<uint>(value);
-            }
-        }
+        public static unsafe Vector64<uint> CreateScalar(uint value) => CreateScalar<uint>(value);
 
         /// <summary>Creates a new <see cref="Vector64{UInt64}" /> instance with the first element initialized to the specified value and the remaining elements initialized to zero.</summary>
         /// <param name="value">The value that element 0 will be initialized to.</param>
         /// <returns>A new <see cref="Vector64{UInt64}" /> instance with the first element initialized to <paramref name="value"/> and the remaining elements initialized to zero.</returns>
+        [Intrinsic]
         [CLSCompliant(false)]
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static unsafe Vector64<ulong> CreateScalar(ulong value) => CreateScalar<ulong>(value);