[llvm] [TTI][X86] getShuffleCosts - for SK_PermuteTwoSrc, if the masks are known to be "inlane" no need to scale the costs by worst-case legalization (PR #117999)

Thu Nov 28 04:18:32 PST 2024

https://github.com/RKSimon created https://github.com/llvm/llvm-project/pull/117999

SK_PermuteTwoSrc legalization has to assume any of the legalised source registers could be referenced in split shuffles, but if we already know that each 128-bit lane only references elements from the same lane of the source operands, then this scaling won't occur.

Hopefully this can help with #113356 without us having to get full processShuffleMasks canonicalization finished first.

>From ba1f367d0166f797f7f2c9f9daee82996104a2d4 Mon Sep 17 00:00:00 2001
From: Simon Pilgrim <llvm-dev at redking.me.uk>
Date: Thu, 28 Nov 2024 12:16:57 +0000
Subject: [PATCH] [TTI][X86] getShuffleCosts - for SK_PermuteTwoSrc, if the
 masks are known to be "inlane" no need to scale the costs by worst-case
 legalization

SK_PermuteTwoSrc legalization has to assume any of the legalised source registers could be referenced in split shuffles, but if we already know that each 128-bit lane only references elements from the same lane of the source operands, then this scaling won't occur.

Hopefully this can help with #113356 without us having to get full processShuffleMasks canonicalization finished first.
---
 .../lib/Target/X86/X86TargetTransformInfo.cpp |  2 +-
 .../X86/shuffle-insert_subvector-codesize.ll  | 56 +++++++++----------
 .../X86/shuffle-insert_subvector-latency.ll   | 56 +++++++++----------
 .../shuffle-insert_subvector-sizelatency.ll   | 56 +++++++++----------
 .../CostModel/X86/shuffle-insert_subvector.ll | 56 +++++++++----------
 .../X86/shuffle-transpose-codesize.ll         | 56 +++++++++----------
 .../X86/shuffle-transpose-latency.ll          | 56 +++++++++----------
 .../X86/shuffle-transpose-sizelatency.ll      | 56 +++++++++----------
 .../CostModel/X86/shuffle-transpose.ll        | 56 +++++++++----------
 .../Transforms/PhaseOrdering/X86/pr94546.ll   | 20 +++----
 10 files changed, 232 insertions(+), 238 deletions(-)

diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
index 179e29e40614e7..764ed8bac0d528 100644
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
@@ -1775,7 +1775,7 @@ InstructionCost X86TTIImpl::getShuffleCost(
   }
 
   // For 2-input shuffles, we must account for splitting the 2 inputs into many.
-  if (Kind == TTI::SK_PermuteTwoSrc && LT.first != 1) {
+  if (Kind == TTI::SK_PermuteTwoSrc && !IsInLaneShuffle && LT.first != 1) {
     // We assume that source and destination have the same vector type.
     InstructionCost NumOfDests = LT.first;
     InstructionCost NumOfShufflesPerDest = LT.first * 2 - 1;
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-codesize.ll b/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-codesize.ll
index 7f9a84f0cb0f39..22da05a7225060 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-codesize.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-codesize.ll
@@ -143,13 +143,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -174,13 +174,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -205,13 +205,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -242,7 +242,7 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -273,7 +273,7 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -364,13 +364,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -395,13 +395,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -426,13 +426,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -463,7 +463,7 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -494,7 +494,7 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -595,15 +595,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -630,15 +630,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -665,15 +665,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-latency.ll b/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-latency.ll
index 00248958d18620..264fc95ef6b568 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-latency.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-latency.ll
@@ -143,13 +143,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -174,13 +174,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -205,13 +205,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -242,7 +242,7 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -273,7 +273,7 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -364,13 +364,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -395,13 +395,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -426,13 +426,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -463,7 +463,7 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -494,7 +494,7 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -595,15 +595,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -630,15 +630,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -665,15 +665,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-sizelatency.ll b/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-sizelatency.ll
index 3845d582f8b111..b072232943e594 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-sizelatency.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector-sizelatency.ll
@@ -143,13 +143,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -174,13 +174,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -205,13 +205,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -242,7 +242,7 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -273,7 +273,7 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -364,13 +364,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -395,13 +395,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -426,13 +426,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -463,7 +463,7 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -494,7 +494,7 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -595,15 +595,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -630,15 +630,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -665,15 +665,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector.ll b/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector.ll
index 5082eed6fd8794..a1386117126c7c 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-insert_subvector.ll
@@ -143,13 +143,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -174,13 +174,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -205,13 +205,13 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x float> %src128, <4 x float> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x float> %src256, <8 x float> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -242,7 +242,7 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -273,7 +273,7 @@ define void @test_vXf32(<2 x float> %src64, <4 x float> %src128, <8 x float> %sr
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x float> %src256, <8 x float> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512_67 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x float> %src512, <16 x float> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -364,13 +364,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -395,13 +395,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -426,13 +426,13 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> %src64_128, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> %src64_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -463,7 +463,7 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -494,7 +494,7 @@ define void @test_vXi32(<2 x i32> %src64, <4 x i32> %src128, <8 x i32> %src256,
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> %src128_256, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> %src64_512, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -595,15 +595,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -630,15 +630,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
@@ -665,15 +665,15 @@ define void @test_vXi16(<2 x i16> %src32, <4 x i16> %src64, <8 x i16> %src128, <
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 4, i32 5, i32 6, i32 7>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> %src32_128, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 16, i32 17, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 16, i32 17, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 16, i32 17, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 16, i32 17, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> %src64_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 16, i32 17>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 16, i32 17, i32 18, i32 19, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 12, i32 13, i32 14, i32 15>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> %src128_256, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-transpose-codesize.ll b/llvm/test/Analysis/CostModel/X86/shuffle-transpose-codesize.ll
index 7783161375916f..c8662a41b8e3a8 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-transpose-codesize.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-transpose-codesize.ll
@@ -19,14 +19,14 @@
 define void @test_vXf64(<2 x double> %a128, <2 x double> %b128, <4 x double> %a256, <4 x double> %b256, <8 x double> %a512, <8 x double> %b512) {
 ; SSE-LABEL: 'test_vXf64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %a128, <2 x double> %b128, <2 x i32> <i32 0, i32 2>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXf64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %a128, <2 x double> %b128, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXf64'
@@ -44,14 +44,14 @@ define void @test_vXf64(<2 x double> %a128, <2 x double> %b128, <4 x double> %a2
 define void @test_vXi64(<2 x i64> %a128, <2 x i64> %b128, <4 x i64> %a256, <4 x i64> %b256, <8 x i64> %a512, <8 x i64> %b512) {
 ; SSE-LABEL: 'test_vXi64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %a128, <2 x i64> %b128, <2 x i32> <i32 0, i32 2>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXi64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %a128, <2 x i64> %b128, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXi64'
@@ -70,15 +70,15 @@ define void @test_vXf32(<2 x float> %a64, <2 x float> %b64, <4 x float> %a128, <
 ; SSE-LABEL: 'test_vXf32'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x float> %a64, <2 x float> %b64, <2 x i32> <i32 0, i32 2>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x float> %a128, <4 x float> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXf32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x float> %a64, <2 x float> %b64, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x float> %a128, <4 x float> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXf32'
@@ -99,15 +99,15 @@ define void @test_vXi32(<2 x i32> %a64, <2 x i32> %b64, <4 x i32> %a128, <4 x i3
 ; SSE-LABEL: 'test_vXi32'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x i32> %a64, <2 x i32> %b64, <2 x i32> <i32 0, i32 2>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXi32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x i32> %a64, <2 x i32> %b64, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXi32'
@@ -129,24 +129,24 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXi16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXi16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX1-LABEL: 'test_vXi16'
@@ -154,7 +154,7 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX2-LABEL: 'test_vXi16'
@@ -162,7 +162,7 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'test_vXi16'
@@ -203,8 +203,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 364 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXi8'
@@ -212,8 +212,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXi8'
@@ -221,8 +221,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX1-LABEL: 'test_vXi8'
@@ -231,7 +231,7 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX2-LABEL: 'test_vXi8'
@@ -240,7 +240,7 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'test_vXi8'
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-transpose-latency.ll b/llvm/test/Analysis/CostModel/X86/shuffle-transpose-latency.ll
index f96bb0d8e6ed09..a207bc8e20f5fb 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-transpose-latency.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-transpose-latency.ll
@@ -19,14 +19,14 @@
 define void @test_vXf64(<2 x double> %a128, <2 x double> %b128, <4 x double> %a256, <4 x double> %b256, <8 x double> %a512, <8 x double> %b512) {
 ; SSE-LABEL: 'test_vXf64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %a128, <2 x double> %b128, <2 x i32> <i32 0, i32 2>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXf64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %a128, <2 x double> %b128, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXf64'
@@ -44,14 +44,14 @@ define void @test_vXf64(<2 x double> %a128, <2 x double> %b128, <4 x double> %a2
 define void @test_vXi64(<2 x i64> %a128, <2 x i64> %b128, <4 x i64> %a256, <4 x i64> %b256, <8 x i64> %a512, <8 x i64> %b512) {
 ; SSE-LABEL: 'test_vXi64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %a128, <2 x i64> %b128, <2 x i32> <i32 0, i32 2>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXi64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %a128, <2 x i64> %b128, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXi64'
@@ -70,15 +70,15 @@ define void @test_vXf32(<2 x float> %a64, <2 x float> %b64, <4 x float> %a128, <
 ; SSE-LABEL: 'test_vXf32'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x float> %a64, <2 x float> %b64, <2 x i32> <i32 0, i32 2>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x float> %a128, <4 x float> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXf32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x float> %a64, <2 x float> %b64, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x float> %a128, <4 x float> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXf32'
@@ -99,15 +99,15 @@ define void @test_vXi32(<2 x i32> %a64, <2 x i32> %b64, <4 x i32> %a128, <4 x i3
 ; SSE-LABEL: 'test_vXi32'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x i32> %a64, <2 x i32> %b64, <2 x i32> <i32 0, i32 2>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXi32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x i32> %a64, <2 x i32> %b64, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXi32'
@@ -129,24 +129,24 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXi16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXi16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX1-LABEL: 'test_vXi16'
@@ -154,7 +154,7 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX2-LABEL: 'test_vXi16'
@@ -162,7 +162,7 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'test_vXi16'
@@ -203,8 +203,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 364 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXi8'
@@ -212,8 +212,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXi8'
@@ -221,8 +221,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX1-LABEL: 'test_vXi8'
@@ -231,7 +231,7 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX2-LABEL: 'test_vXi8'
@@ -240,7 +240,7 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'test_vXi8'
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-transpose-sizelatency.ll b/llvm/test/Analysis/CostModel/X86/shuffle-transpose-sizelatency.ll
index e9927c61e77417..75130deda1df82 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-transpose-sizelatency.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-transpose-sizelatency.ll
@@ -19,14 +19,14 @@
 define void @test_vXf64(<2 x double> %a128, <2 x double> %b128, <4 x double> %a256, <4 x double> %b256, <8 x double> %a512, <8 x double> %b512) {
 ; SSE-LABEL: 'test_vXf64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %a128, <2 x double> %b128, <2 x i32> <i32 0, i32 2>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXf64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %a128, <2 x double> %b128, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXf64'
@@ -44,14 +44,14 @@ define void @test_vXf64(<2 x double> %a128, <2 x double> %b128, <4 x double> %a2
 define void @test_vXi64(<2 x i64> %a128, <2 x i64> %b128, <4 x i64> %a256, <4 x i64> %b256, <8 x i64> %a512, <8 x i64> %b512) {
 ; SSE-LABEL: 'test_vXi64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %a128, <2 x i64> %b128, <2 x i32> <i32 0, i32 2>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXi64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %a128, <2 x i64> %b128, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXi64'
@@ -70,15 +70,15 @@ define void @test_vXf32(<2 x float> %a64, <2 x float> %b64, <4 x float> %a128, <
 ; SSE-LABEL: 'test_vXf32'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x float> %a64, <2 x float> %b64, <2 x i32> <i32 0, i32 2>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x float> %a128, <4 x float> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXf32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x float> %a64, <2 x float> %b64, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x float> %a128, <4 x float> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXf32'
@@ -99,15 +99,15 @@ define void @test_vXi32(<2 x i32> %a64, <2 x i32> %b64, <4 x i32> %a128, <4 x i3
 ; SSE-LABEL: 'test_vXi32'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x i32> %a64, <2 x i32> %b64, <2 x i32> <i32 0, i32 2>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXi32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x i32> %a64, <2 x i32> %b64, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXi32'
@@ -129,24 +129,24 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXi16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXi16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX1-LABEL: 'test_vXi16'
@@ -154,7 +154,7 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX2-LABEL: 'test_vXi16'
@@ -162,7 +162,7 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'test_vXi16'
@@ -203,8 +203,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 364 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXi8'
@@ -212,8 +212,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXi8'
@@ -221,8 +221,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX1-LABEL: 'test_vXi8'
@@ -231,7 +231,7 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX2-LABEL: 'test_vXi8'
@@ -240,7 +240,7 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'test_vXi8'
diff --git a/llvm/test/Analysis/CostModel/X86/shuffle-transpose.ll b/llvm/test/Analysis/CostModel/X86/shuffle-transpose.ll
index 4e32b45f4e4140..5c50ec1feb0582 100644
--- a/llvm/test/Analysis/CostModel/X86/shuffle-transpose.ll
+++ b/llvm/test/Analysis/CostModel/X86/shuffle-transpose.ll
@@ -19,14 +19,14 @@
 define void @test_vXf64(<2 x double> %a128, <2 x double> %b128, <4 x double> %a256, <4 x double> %b256, <8 x double> %a512, <8 x double> %b512) {
 ; SSE-LABEL: 'test_vXf64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %a128, <2 x double> %b128, <2 x i32> <i32 0, i32 2>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXf64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x double> %a128, <2 x double> %b128, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x double> %a256, <4 x double> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x double> %a512, <8 x double> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXf64'
@@ -44,14 +44,14 @@ define void @test_vXf64(<2 x double> %a128, <2 x double> %b128, <4 x double> %a2
 define void @test_vXi64(<2 x i64> %a128, <2 x i64> %b128, <4 x i64> %a256, <4 x i64> %b256, <8 x i64> %a512, <8 x i64> %b512) {
 ; SSE-LABEL: 'test_vXi64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %a128, <2 x i64> %b128, <2 x i32> <i32 0, i32 2>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXi64'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128 = shufflevector <2 x i64> %a128, <2 x i64> %b128, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <4 x i64> %a256, <4 x i64> %b256, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <8 x i64> %a512, <8 x i64> %b512, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXi64'
@@ -70,15 +70,15 @@ define void @test_vXf32(<2 x float> %a64, <2 x float> %b64, <4 x float> %a128, <
 ; SSE-LABEL: 'test_vXf32'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x float> %a64, <2 x float> %b64, <2 x i32> <i32 0, i32 2>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x float> %a128, <4 x float> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXf32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x float> %a64, <2 x float> %b64, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x float> %a128, <4 x float> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x float> %a256, <8 x float> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <16 x float> %a512, <16 x float> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXf32'
@@ -99,15 +99,15 @@ define void @test_vXi32(<2 x i32> %a64, <2 x i32> %b64, <4 x i32> %a128, <4 x i3
 ; SSE-LABEL: 'test_vXi32'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x i32> %a64, <2 x i32> %b64, <2 x i32> <i32 0, i32 2>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX-LABEL: 'test_vXi32'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <2 x i32> %a64, <2 x i32> %b64, <2 x i32> <i32 0, i32 2>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V128 = shufflevector <4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test_vXi32'
@@ -129,24 +129,24 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 224 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXi16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXi16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <2 x i16> %a32, <2 x i16> %b32, <2 x i32> <i32 0, i32 2>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'test_vXi16'
@@ -154,7 +154,7 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'test_vXi16'
@@ -162,7 +162,7 @@ define void @test_vXi16(<2 x i16> %a32, <2 x i16> %b32, <4 x i16> %a64, <4 x i16
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <4 x i16> %a64, <4 x i16> %b64, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <8 x i16> %a128, <8 x i16> %b128, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <16 x i16> %a256, <16 x i16> %b256, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <32 x i16> %a512, <32 x i16> %b512, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'test_vXi16'
@@ -203,8 +203,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 364 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSSE3-LABEL: 'test_vXi8'
@@ -212,8 +212,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'test_vXi8'
@@ -221,8 +221,8 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32 = shufflevector <4 x i8> %a32, <4 x i8> %b32, <4 x i32> <i32 0, i32 4, i32 2, i32 6>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'test_vXi8'
@@ -231,7 +231,7 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'test_vXi8'
@@ -240,7 +240,7 @@ define void @test_vXi8(<2 x i8> %a16, <2 x i8> %b16, <4 x i8> %a32, <4 x i8> %b3
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V64 = shufflevector <8 x i8> %a64, <8 x i8> %b64, <8 x i32> <i32 0, i32 8, i32 2, i32 10, i32 4, i32 12, i32 6, i32 14>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V128 = shufflevector <16 x i8> %a128, <16 x i8> %b128, <16 x i32> <i32 0, i32 16, i32 2, i32 18, i32 4, i32 20, i32 6, i32 22, i32 8, i32 24, i32 10, i32 26, i32 12, i32 28, i32 14, i32 30>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256 = shufflevector <32 x i8> %a256, <32 x i8> %b256, <32 x i32> <i32 0, i32 32, i32 2, i32 34, i32 4, i32 36, i32 6, i32 38, i32 8, i32 40, i32 10, i32 42, i32 12, i32 44, i32 14, i32 46, i32 16, i32 48, i32 18, i32 50, i32 20, i32 52, i32 22, i32 54, i32 24, i32 56, i32 26, i32 58, i32 28, i32 60, i32 30, i32 62>
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512 = shufflevector <64 x i8> %a512, <64 x i8> %b512, <64 x i32> <i32 0, i32 64, i32 2, i32 66, i32 4, i32 68, i32 6, i32 70, i32 8, i32 72, i32 10, i32 74, i32 12, i32 76, i32 14, i32 78, i32 16, i32 80, i32 18, i32 82, i32 20, i32 84, i32 22, i32 86, i32 24, i32 88, i32 26, i32 90, i32 28, i32 92, i32 30, i32 94, i32 32, i32 96, i32 34, i32 98, i32 36, i32 100, i32 38, i32 102, i32 40, i32 104, i32 42, i32 106, i32 44, i32 108, i32 46, i32 110, i32 48, i32 112, i32 50, i32 114, i32 52, i32 116, i32 54, i32 118, i32 56, i32 120, i32 58, i32 122, i32 60, i32 124, i32 62, i32 126>
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'test_vXi8'
diff --git a/llvm/test/Transforms/PhaseOrdering/X86/pr94546.ll b/llvm/test/Transforms/PhaseOrdering/X86/pr94546.ll
index ec9bfc542e32f5..610de3142df310 100644
--- a/llvm/test/Transforms/PhaseOrdering/X86/pr94546.ll
+++ b/llvm/test/Transforms/PhaseOrdering/X86/pr94546.ll
@@ -9,18 +9,11 @@
 ; RUN: opt -mtriple=x86_64-- -mcpu=x86-64-v3 -passes="default<O3>" -S < %s  | FileCheck %s --check-prefixes=AVX,AVX2
 
 define <4 x double> @PR94546(<4 x double> %a, <4 x double> %b) {
-; SSE2-LABEL: @PR94546(
-; SSE2-NEXT:    [[TMP1:%.*]] = shufflevector <4 x double> [[A:%.*]], <4 x double> [[B:%.*]], <2 x i32> <i32 0, i32 6>
-; SSE2-NEXT:    [[TMP2:%.*]] = shufflevector <4 x double> [[A]], <4 x double> [[B]], <2 x i32> <i32 1, i32 7>
-; SSE2-NEXT:    [[TMP3:%.*]] = fadd <2 x double> [[TMP1]], [[TMP2]]
-; SSE2-NEXT:    [[TMP4:%.*]] = shufflevector <2 x double> [[TMP3]], <2 x double> poison, <4 x i32> <i32 0, i32 poison, i32 poison, i32 1>
-; SSE2-NEXT:    ret <4 x double> [[TMP4]]
-;
-; SSE4-LABEL: @PR94546(
-; SSE4-NEXT:    [[TMP1:%.*]] = shufflevector <4 x double> [[A:%.*]], <4 x double> [[B:%.*]], <4 x i32> <i32 0, i32 poison, i32 poison, i32 6>
-; SSE4-NEXT:    [[TMP2:%.*]] = shufflevector <4 x double> [[A]], <4 x double> [[B]], <4 x i32> <i32 1, i32 poison, i32 poison, i32 7>
-; SSE4-NEXT:    [[TMP3:%.*]] = fadd <4 x double> [[TMP1]], [[TMP2]]
-; SSE4-NEXT:    ret <4 x double> [[TMP3]]
+; SSE-LABEL: @PR94546(
+; SSE-NEXT:    [[TMP1:%.*]] = shufflevector <4 x double> [[A:%.*]], <4 x double> [[B:%.*]], <4 x i32> <i32 0, i32 poison, i32 poison, i32 6>
+; SSE-NEXT:    [[TMP2:%.*]] = shufflevector <4 x double> [[A]], <4 x double> [[B]], <4 x i32> <i32 1, i32 poison, i32 poison, i32 7>
+; SSE-NEXT:    [[TMP3:%.*]] = fadd <4 x double> [[TMP1]], [[TMP2]]
+; SSE-NEXT:    ret <4 x double> [[TMP3]]
 ;
 ; AVX-LABEL: @PR94546(
 ; AVX-NEXT:    [[TMP1:%.*]] = shufflevector <4 x double> [[A:%.*]], <4 x double> [[B:%.*]], <4 x i32> <i32 0, i32 poison, i32 poison, i32 6>
@@ -50,4 +43,5 @@ define <4 x double> @PR94546(<4 x double> %a, <4 x double> %b) {
 ;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
 ; AVX1: {{.*}}
 ; AVX2: {{.*}}
-; SSE: {{.*}}
+; SSE2: {{.*}}
+; SSE4: {{.*}}

