[CostModel][X86] SK_InsertSubvector inserted into the lowest subvector should be treated as SK_Select blend (#145892)

X86 uses implicit widening and BLEND/MOV shuffles in these cases - otherwise we still treat it as a SK_PermuteTwoSrc

Author: RKSimon

llvm/lib/Target/X86/X86TargetTransformInfo.cpp

@@ -1676,8 +1676,10 @@ InstructionCost X86TTIImpl::getShuffleCost(TTI::ShuffleKind Kind,
         SubLT.second == MVT::f32 && (Index == 0 || ST->hasSSE41()))
       return 1;
 
-    // If the insertion isn't aligned, treat it like a 2-op shuffle.
-    Kind = TTI::SK_PermuteTwoSrc;
+    // If the insertion is the lowest subvector then it will be blended
+    // otherwise treat it like a 2-op shuffle.
+    Kind =
+        (Index == 0 && LT.first == 1) ? TTI::SK_Select : TTI::SK_PermuteTwoSrc;
   }
 
   // Handle some common (illegal) sub-vector types as they are often very cheap

llvm/test/Transforms/PhaseOrdering/X86/hadd.ll

@@ -290,13 +290,10 @@ define <16 x i16> @add_v16i16_0123u56789uBCDEF(<16 x i16> %a, <16 x i16> %b) {
 ; SSE2-NEXT:    ret <16 x i16> [[RESULT]]
 ;
 ; SSE4-LABEL: @add_v16i16_0123u56789uBCDEF(
-; SSE4-NEXT:    [[TMP1:%.*]] = shufflevector <16 x i16> [[A:%.*]], <16 x i16> [[B:%.*]], <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 18, i32 20, i32 22, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP2:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> [[B]], <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 19, i32 21, i32 23, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP3:%.*]] = shufflevector <16 x i16> [[TMP1]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 24, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
+; SSE4-NEXT:    [[TMP2:%.*]] = shufflevector <16 x i16> [[A:%.*]], <16 x i16> [[B:%.*]], <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 19, i32 21, i32 23, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
 ; SSE4-NEXT:    [[TMP4:%.*]] = shufflevector <16 x i16> [[TMP2]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 25, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP5:%.*]] = shufflevector <16 x i16> [[TMP3]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 8, i32 26, i32 poison, i32 30, i32 poison, i32 poison, i32 poison, i32 poison>
 ; SSE4-NEXT:    [[TMP6:%.*]] = shufflevector <16 x i16> [[TMP4]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 8, i32 27, i32 poison, i32 31, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP7:%.*]] = shufflevector <16 x i16> [[TMP5]], <16 x i16> [[B]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 8, i32 9, i32 poison, i32 11, i32 24, i32 26, i32 28, i32 30>
+; SSE4-NEXT:    [[TMP7:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> [[B]], <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 18, i32 20, i32 22, i32 8, i32 10, i32 poison, i32 14, i32 24, i32 26, i32 28, i32 30>
 ; SSE4-NEXT:    [[TMP8:%.*]] = shufflevector <16 x i16> [[TMP6]], <16 x i16> [[B]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 8, i32 9, i32 poison, i32 11, i32 25, i32 27, i32 29, i32 31>
 ; SSE4-NEXT:    [[TMP9:%.*]] = add <16 x i16> [[TMP7]], [[TMP8]]
 ; SSE4-NEXT:    ret <16 x i16> [[TMP9]]
@@ -404,11 +401,9 @@ define <16 x i16> @add_v16i16_FEuCBA98765432u0(<16 x i16> %a, <16 x i16> %b) {
 ; SSE2-NEXT:    ret <16 x i16> [[RESULT]]
 ;
 ; SSE4-LABEL: @add_v16i16_FEuCBA98765432u0(
-; SSE4-NEXT:    [[TMP1:%.*]] = shufflevector <16 x i16> [[A:%.*]], <16 x i16> [[B:%.*]], <16 x i32> <i32 0, i32 poison, i32 4, i32 6, i32 16, i32 18, i32 20, i32 22, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP2:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> [[B]], <16 x i32> <i32 1, i32 poison, i32 5, i32 7, i32 17, i32 19, i32 21, i32 23, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP3:%.*]] = shufflevector <16 x i16> [[TMP1]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 poison, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 24, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
+; SSE4-NEXT:    [[TMP2:%.*]] = shufflevector <16 x i16> [[A:%.*]], <16 x i16> [[B:%.*]], <16 x i32> <i32 1, i32 poison, i32 5, i32 7, i32 17, i32 19, i32 21, i32 23, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
 ; SSE4-NEXT:    [[TMP10:%.*]] = shufflevector <16 x i16> [[TMP2]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 poison, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 25, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP4:%.*]] = shufflevector <16 x i16> [[TMP3]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 poison, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 poison, i32 27, i32 28, i32 poison, i32 poison, i32 poison, i32 poison>
+; SSE4-NEXT:    [[TMP4:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> [[B]], <16 x i32> <i32 0, i32 poison, i32 4, i32 6, i32 16, i32 18, i32 20, i32 22, i32 8, i32 poison, i32 11, i32 12, i32 poison, i32 poison, i32 poison, i32 poison>
 ; SSE4-NEXT:    [[TMP5:%.*]] = shufflevector <16 x i16> [[TMP10]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 poison, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 poison, i32 26, i32 29, i32 poison, i32 poison, i32 poison, i32 poison>
 ; SSE4-NEXT:    [[TMP6:%.*]] = add <16 x i16> [[TMP4]], [[TMP5]]
 ; SSE4-NEXT:    [[TMP7:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> [[B]], <16 x i32> <i32 14, i32 24, i32 28, i32 30, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>

llvm/test/Transforms/PhaseOrdering/X86/hsub.ll

@@ -290,13 +290,10 @@ define <16 x i16> @sub_v16i16_0123u56789uBCDEF(<16 x i16> %a, <16 x i16> %b) {
 ; SSE2-NEXT:    ret <16 x i16> [[RESULT]]
 ;
 ; SSE4-LABEL: @sub_v16i16_0123u56789uBCDEF(
-; SSE4-NEXT:    [[TMP1:%.*]] = shufflevector <16 x i16> [[A:%.*]], <16 x i16> [[B:%.*]], <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 18, i32 20, i32 22, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP2:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> [[B]], <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 19, i32 21, i32 23, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP3:%.*]] = shufflevector <16 x i16> [[TMP1]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 24, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
+; SSE4-NEXT:    [[TMP2:%.*]] = shufflevector <16 x i16> [[A:%.*]], <16 x i16> [[B:%.*]], <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 19, i32 21, i32 23, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
 ; SSE4-NEXT:    [[TMP4:%.*]] = shufflevector <16 x i16> [[TMP2]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 25, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP5:%.*]] = shufflevector <16 x i16> [[TMP3]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 8, i32 26, i32 poison, i32 30, i32 poison, i32 poison, i32 poison, i32 poison>
 ; SSE4-NEXT:    [[TMP6:%.*]] = shufflevector <16 x i16> [[TMP4]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 8, i32 27, i32 poison, i32 31, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP7:%.*]] = shufflevector <16 x i16> [[TMP5]], <16 x i16> [[B]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 8, i32 9, i32 poison, i32 11, i32 24, i32 26, i32 28, i32 30>
+; SSE4-NEXT:    [[TMP7:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> [[B]], <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 18, i32 20, i32 22, i32 8, i32 10, i32 poison, i32 14, i32 24, i32 26, i32 28, i32 30>
 ; SSE4-NEXT:    [[TMP8:%.*]] = shufflevector <16 x i16> [[TMP6]], <16 x i16> [[B]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 5, i32 6, i32 7, i32 8, i32 9, i32 poison, i32 11, i32 25, i32 27, i32 29, i32 31>
 ; SSE4-NEXT:    [[TMP9:%.*]] = sub <16 x i16> [[TMP7]], [[TMP8]]
 ; SSE4-NEXT:    ret <16 x i16> [[TMP9]]
@@ -398,11 +395,9 @@ define <16 x i16> @sub_v16i16_FEuCBA98765432u0(<16 x i16> %a, <16 x i16> %b) {
 ; SSE2-NEXT:    ret <16 x i16> [[RESULT]]
 ;
 ; SSE4-LABEL: @sub_v16i16_FEuCBA98765432u0(
-; SSE4-NEXT:    [[TMP1:%.*]] = shufflevector <16 x i16> [[A:%.*]], <16 x i16> [[B:%.*]], <16 x i32> <i32 0, i32 poison, i32 4, i32 6, i32 16, i32 18, i32 20, i32 22, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP2:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> [[B]], <16 x i32> <i32 1, i32 poison, i32 5, i32 7, i32 17, i32 19, i32 21, i32 23, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP3:%.*]] = shufflevector <16 x i16> [[TMP1]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 poison, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 24, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
+; SSE4-NEXT:    [[TMP2:%.*]] = shufflevector <16 x i16> [[A:%.*]], <16 x i16> [[B:%.*]], <16 x i32> <i32 1, i32 poison, i32 5, i32 7, i32 17, i32 19, i32 21, i32 23, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
 ; SSE4-NEXT:    [[TMP10:%.*]] = shufflevector <16 x i16> [[TMP2]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 poison, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 25, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
-; SSE4-NEXT:    [[TMP4:%.*]] = shufflevector <16 x i16> [[TMP3]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 poison, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 poison, i32 26, i32 28, i32 poison, i32 poison, i32 poison, i32 poison>
+; SSE4-NEXT:    [[TMP4:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> [[B]], <16 x i32> <i32 0, i32 poison, i32 4, i32 6, i32 16, i32 18, i32 20, i32 22, i32 8, i32 poison, i32 10, i32 12, i32 poison, i32 poison, i32 poison, i32 poison>
 ; SSE4-NEXT:    [[TMP5:%.*]] = shufflevector <16 x i16> [[TMP10]], <16 x i16> [[A]], <16 x i32> <i32 0, i32 poison, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 poison, i32 27, i32 29, i32 poison, i32 poison, i32 poison, i32 poison>
 ; SSE4-NEXT:    [[TMP6:%.*]] = sub <16 x i16> [[TMP4]], [[TMP5]]
 ; SSE4-NEXT:    [[TMP7:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> [[B]], <16 x i32> <i32 14, i32 24, i32 28, i32 30, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>

llvm/test/Transforms/SLPVectorizer/X86/extracts-non-extendable.ll

@@ -7,26 +7,16 @@ define void @test(i64 %v) {
 ; CHECK-NEXT:  [[BB:.*:]]
 ; CHECK-NEXT:    [[TMP0:%.*]] = insertelement <2 x i64> <i64 0, i64 poison>, i64 [[V]], i32 1
 ; CHECK-NEXT:    [[TMP1:%.*]] = or <2 x i64> zeroinitializer, [[TMP0]]
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
-; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq i64 0, [[TMP2]]
-; CHECK-NEXT:    [[TMP4:%.*]] = icmp eq i64 0, 0
-; CHECK-NEXT:    [[TMP5:%.*]] = and i1 [[TMP3]], [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i64 0, 0
-; CHECK-NEXT:    [[TMP7:%.*]] = and i1 [[TMP5]], [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = icmp eq i64 0, 0
-; CHECK-NEXT:    [[TMP9:%.*]] = and i1 [[TMP7]], [[TMP8]]
-; CHECK-NEXT:    [[TMP10:%.*]] = and i1 [[TMP9]], false
-; CHECK-NEXT:    [[TMP11:%.*]] = icmp eq i64 0, [[TMP2]]
-; CHECK-NEXT:    [[TMP12:%.*]] = and i1 [[TMP10]], [[TMP11]]
-; CHECK-NEXT:    [[TMP13:%.*]] = icmp eq i64 0, 0
-; CHECK-NEXT:    [[TMP14:%.*]] = and i1 [[TMP12]], [[TMP13]]
-; CHECK-NEXT:    [[TMP15:%.*]] = icmp eq i64 0, 0
-; CHECK-NEXT:    [[TMP16:%.*]] = and i1 [[TMP14]], [[TMP15]]
-; CHECK-NEXT:    [[TMP17:%.*]] = icmp eq i64 0, 0
-; CHECK-NEXT:    [[TMP18:%.*]] = and i1 [[TMP16]], [[TMP17]]
+; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <2 x i64> [[TMP1]], <2 x i64> poison, <8 x i32> <i32 1, i32 poison, i32 poison, i32 poison, i32 1, i32 poison, i32 poison, i32 poison>
+; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <8 x i64> [[TMP2]], <8 x i64> <i64 poison, i64 0, i64 0, i64 0, i64 poison, i64 0, i64 0, i64 0>, <8 x i32> <i32 0, i32 9, i32 10, i32 11, i32 4, i32 13, i32 14, i32 15>
+; CHECK-NEXT:    [[TMP4:%.*]] = icmp eq <8 x i64> zeroinitializer, [[TMP3]]
 ; CHECK-NEXT:    [[TMP19:%.*]] = icmp ult i64 0, 0
+; CHECK-NEXT:    [[TMP6:%.*]] = freeze <8 x i1> [[TMP4]]
+; CHECK-NEXT:    [[TMP18:%.*]] = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> [[TMP6]])
 ; CHECK-NEXT:    [[TMP20:%.*]] = select i1 [[TMP19]], i1 [[TMP18]], i1 false
-; CHECK-NEXT:    br i1 [[TMP20]], label %[[BB_I107_PREHEADER:.*]], label %[[BB_I27_I_PREHEADER:.*]]
+; CHECK-NEXT:    [[TMP8:%.*]] = freeze i1 [[TMP20]]
+; CHECK-NEXT:    [[OP_RDX1:%.*]] = select i1 [[TMP8]], i1 false, i1 false
+; CHECK-NEXT:    br i1 [[OP_RDX1]], label %[[BB_I107_PREHEADER:.*]], label %[[BB_I27_I_PREHEADER:.*]]
 ; CHECK:       [[BB_I107_PREHEADER]]:
 ; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <2 x i64> [[TMP1]], i32 0
 ; CHECK-NEXT:    [[DOTSROA_1278_10_EXTRACT_SHIFT83_I1622_1:%.*]] = xor i64 0, [[TMP21]]