Implement AVX SetHighLow

Author: FeiPengIntel

src/jit/hwintrinsiclistxarch.h

@@ -377,6 +377,7 @@ HARDWARE_INTRINSIC(AVX_RoundToNegativeInfinity,                      "RoundToNeg
 HARDWARE_INTRINSIC(AVX_RoundToPositiveInfinity,                      "RoundToPositiveInfinity",                          AVX,        10,           32,           1,           {INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_roundps,   INS_roundpd},           HW_Category_SimpleSIMD,                        HW_Flag_NoRMWSemantics)
 HARDWARE_INTRINSIC(AVX_RoundToZero,                                  "RoundToZero",                                      AVX,        11,           32,           1,           {INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_roundps,   INS_roundpd},           HW_Category_SimpleSIMD,                        HW_Flag_NoRMWSemantics)
 HARDWARE_INTRINSIC(AVX_SetVector256,                                 "SetVector256",                                     AVX,        -1,           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_NoCodeGen|HW_Flag_SecondArgMaybe64Bit)
+HARDWARE_INTRINSIC(AVX_SetHighLow,                                   "SetHighLow",                                       AVX,        -1,           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_NoCodeGen|HW_Flag_OneTypeGeneric)
 HARDWARE_INTRINSIC(AVX_SetAllVector256,                              "SetAllVector256",                                  AVX,        -1,           32,           1,           {INS_vpbroadcastb,INS_vpbroadcastb,INS_vpbroadcastw,INS_vpbroadcastw,INS_vpbroadcastd,INS_vpbroadcastd,INS_vpbroadcastq,INS_vpbroadcastq,INS_vbroadcastss,INS_vbroadcastsd},HW_Category_Helper,                HW_Flag_MultiIns|HW_Flag_SpecialImport|HW_Flag_OneTypeGeneric)
 HARDWARE_INTRINSIC(AVX_SetZeroVector256,                             "SetZeroVector256",                                 AVX,        -1,           32,           0,           {INS_pxor,      INS_pxor,      INS_pxor,      INS_pxor,      INS_pxor,      INS_pxor,      INS_pxor,      INS_pxor,      INS_xorps,     INS_xorpd},             HW_Category_Helper,                            HW_Flag_OneTypeGeneric|HW_Flag_NoRMWSemantics)
 HARDWARE_INTRINSIC(AVX_Shuffle,                                      "Shuffle",                                          AVX,        -1,           32,           3,           {INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_invalid,   INS_shufps,    INS_shufpd},            HW_Category_IMM,                               HW_Flag_NoRMWSemantics|HW_Flag_FullRangeIMM)

src/jit/hwintrinsicxarch.cpp

@@ -1236,6 +1236,16 @@ GenTree* Compiler::impAvxOrAvx2Intrinsic(NamedIntrinsic        intrinsic,
             break;
         }
 
+        case NI_AVX_SetHighLow:
+        {
+            baseType              = getBaseTypeOfSIMDType(sig->retTypeSigClass);
+            GenTree* lowerVector  = impSIMDPopStack(TYP_SIMD16);
+            GenTree* higherVector = impSIMDPopStack(TYP_SIMD16);
+            retNode               = gtNewSimdHWIntrinsicNode(TYP_SIMD32, lowerVector, higherVector, gtNewIconNode(1),
+                                               NI_AVX_InsertVector128, baseType, 32);
+            break;
+        }
+
         case NI_AVX_ExtractVector128:
         case NI_AVX2_ExtractVector128:
         {