[llvm] [RISCV][TTI] Cost a subvector extract at a register boundary with exact vlen (PR #82405)

Philip Reames via llvm-commits llvm-commits at lists.llvm.org
Wed Feb 21 07:52:07 PST 2024


https://github.com/preames updated https://github.com/llvm/llvm-project/pull/82405

>From 5b3e54328ea2827cafbfe5d69e118e4d387b582a Mon Sep 17 00:00:00 2001
From: Philip Reames <preames at rivosinc.com>
Date: Tue, 20 Feb 2024 08:25:10 -0800
Subject: [PATCH 1/3] [RISCV][TTI] Cost a subvector extract at a register
 boundary with exact vlen

If we have exact vlen knowledge, we can figure out which indices correspond
to register boundaries.  Our lowering uses this knowledge to replace the
vslidedown.vi with a sub-register extract.  Our costs can reflect that as
well.
---
 .../Target/RISCV/RISCVTargetTransformInfo.cpp |  12 +
 .../RISCV/shuffle-extract_subvector.ll        | 229 ++++++++++++++++++
 2 files changed, 241 insertions(+)

diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index cf234f25bf9d8c..97b929f15f018a 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -436,6 +436,18 @@ InstructionCost RISCVTTIImpl::getShuffleCost(TTI::ShuffleKind Kind,
     if (Index == 0)
       return TTI::TCC_Free;
 
+    // If we're extracting a subvector at a sub-register boundary - which
+    // unfortunately we need exact vlen to identity - this is a subregister
+    // extract at worst and thus won't require a vslidedown.
+    if (std::pair<InstructionCost, MVT> SubLT = getTypeLegalizationCost(SubTp);
+        SubLT.second.isValid() && SubLT.second.isFixedLengthVector()) {
+      const unsigned MinVLen = ST->getRealMinVLen();
+      const unsigned MaxVLen = ST->getRealMaxVLen();
+      if (MinVLen == MaxVLen &&
+          SubLT.second.getScalarSizeInBits() * Index % MinVLen == 0)
+        return TTI::TCC_Free;
+    }
+
     // Example sequence:
     // vsetivli     zero, 4, e8, mf2, tu, ma (ignored)
     // vslidedown.vi  v8, v9, 2
diff --git a/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll b/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll
index 793786318a0a16..29a9ade4c5e508 100644
--- a/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll
@@ -1,5 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
 ; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv32 -mattr=+v,+f,+d,+zfh,+zvfh -riscv-v-vector-bits-min=-1 | FileCheck %s
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv32 -mattr=+v,+f,+d,+zfh,+zvfh -riscv-v-vector-bits-min=128 -riscv-v-vector-bits-max=128 | FileCheck --check-prefix=VLEN128 %s
 ; Check that we don't crash querying costs when vectors are not enabled.
 ; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv32
 
@@ -20,6 +21,19 @@ define void @test_vXf64(<4 x double> %src256, <8 x double> %src512) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_4567 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %V512_567u = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 5, i32 6, i32 7, i32 poison>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
+; VLEN128-LABEL: 'test_vXf64'
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_01 = shufflevector <4 x double> %src256, <4 x double> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_23 = shufflevector <4 x double> %src256, <4 x double> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_01 = shufflevector <8 x double> %src512, <8 x double> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_23 = shufflevector <8 x double> %src512, <8 x double> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_45 = shufflevector <8 x double> %src512, <8 x double> undef, <2 x i32> <i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_67 = shufflevector <8 x double> %src512, <8 x double> undef, <2 x i32> <i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_0123 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_2345 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_4567 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %V512_567u = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 5, i32 6, i32 7, i32 poison>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %V256_01 = shufflevector <4 x double> %src256, <4 x double> undef, <2 x i32> <i32 0, i32 1>
   %V256_23 = shufflevector <4 x double> %src256, <4 x double> undef, <2 x i32> <i32 2, i32 3>
@@ -46,6 +60,18 @@ define void @test_vXi64(<4 x i64> %src256, <8 x i64> %src512) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_2345 = shufflevector <8 x i64> %src512, <8 x i64> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_4567 = shufflevector <8 x i64> %src512, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
+; VLEN128-LABEL: 'test_vXi64'
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_01 = shufflevector <4 x i64> %src256, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_23 = shufflevector <4 x i64> %src256, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_01 = shufflevector <8 x i64> %src512, <8 x i64> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_23 = shufflevector <8 x i64> %src512, <8 x i64> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_45 = shufflevector <8 x i64> %src512, <8 x i64> undef, <2 x i32> <i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_67 = shufflevector <8 x i64> %src512, <8 x i64> undef, <2 x i32> <i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_0123 = shufflevector <8 x i64> %src512, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_2345 = shufflevector <8 x i64> %src512, <8 x i64> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_4567 = shufflevector <8 x i64> %src512, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %V256_01 = shufflevector <4 x i64> %src256, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
   %V256_23 = shufflevector <4 x i64> %src256, <4 x i64> undef, <2 x i32> <i32 2, i32 3>
@@ -84,6 +110,31 @@ define void @test_vXi32(<4 x i32> %src128, <8 x i32> %src256, <16 x i32> %src512
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_01234567 = shufflevector <16 x i32> %src512, <16 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_89ABCDEF = shufflevector <16 x i32> %src512, <16 x i32> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
+; VLEN128-LABEL: 'test_vXi32'
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_01 = shufflevector <8 x i32> %src256, <8 x i32> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_23 = shufflevector <8 x i32> %src256, <8 x i32> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_45 = shufflevector <8 x i32> %src256, <8 x i32> undef, <2 x i32> <i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_67 = shufflevector <8 x i32> %src256, <8 x i32> undef, <2 x i32> <i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_0123 = shufflevector <8 x i32> %src256, <8 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_4567 = shufflevector <8 x i32> %src256, <8 x i32> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_01 = shufflevector <16 x i32> %src512, <16 x i32> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_23 = shufflevector <16 x i32> %src512, <16 x i32> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_45 = shufflevector <16 x i32> %src512, <16 x i32> undef, <2 x i32> <i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_67 = shufflevector <16 x i32> %src512, <16 x i32> undef, <2 x i32> <i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_89 = shufflevector <16 x i32> %src512, <16 x i32> undef, <2 x i32> <i32 8, i32 9>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_AB = shufflevector <16 x i32> %src512, <16 x i32> undef, <2 x i32> <i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_CD = shufflevector <16 x i32> %src512, <16 x i32> undef, <2 x i32> <i32 12, i32 13>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_EF = shufflevector <16 x i32> %src512, <16 x i32> undef, <2 x i32> <i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_0123 = shufflevector <16 x i32> %src512, <16 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_4567 = shufflevector <16 x i32> %src512, <16 x i32> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_89AB = shufflevector <16 x i32> %src512, <16 x i32> undef, <4 x i32> <i32 8, i32 9, i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_CDEF = shufflevector <16 x i32> %src512, <16 x i32> undef, <4 x i32> <i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_01234567 = shufflevector <16 x i32> %src512, <16 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_89ABCDEF = shufflevector <16 x i32> %src512, <16 x i32> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> undef, <2 x i32> <i32 0, i32 1>
   %V128_23 = shufflevector <4 x i32> %src128, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
@@ -169,6 +220,65 @@ define void @test_vXi16(<4 x i16> %src64, <8 x i16> %src128, <16 x i16> %src256,
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01_02_03_04_05_06_07_08_09_0A_0B_0C_0D_0E_0F = shufflevector <32 x i16> %src512, <32 x i16> undef, <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 14, i32 15>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_10_11_12_13_14_15_16_17_18_19_1A_1B_1C_1D_1E_1F = shufflevector <32 x i16> %src512, <32 x i16> undef, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
+; VLEN128-LABEL: 'test_vXi16'
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V64_01 = shufflevector <4 x i16> %src64, <4 x i16> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_23 = shufflevector <4 x i16> %src64, <4 x i16> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V128_01 = shufflevector <8 x i16> %src128, <8 x i16> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_23 = shufflevector <8 x i16> %src128, <8 x i16> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_45 = shufflevector <8 x i16> %src128, <8 x i16> undef, <2 x i32> <i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_67 = shufflevector <8 x i16> %src128, <8 x i16> undef, <2 x i32> <i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V128_0123 = shufflevector <8 x i16> %src128, <8 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_4567 = shufflevector <8 x i16> %src128, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_01 = shufflevector <16 x i16> %src256, <16 x i16> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_23 = shufflevector <16 x i16> %src256, <16 x i16> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_45 = shufflevector <16 x i16> %src256, <16 x i16> undef, <2 x i32> <i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_67 = shufflevector <16 x i16> %src256, <16 x i16> undef, <2 x i32> <i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_89 = shufflevector <16 x i16> %src256, <16 x i16> undef, <2 x i32> <i32 8, i32 9>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_AB = shufflevector <16 x i16> %src256, <16 x i16> undef, <2 x i32> <i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_CD = shufflevector <16 x i16> %src256, <16 x i16> undef, <2 x i32> <i32 12, i32 13>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_EF = shufflevector <16 x i16> %src256, <16 x i16> undef, <2 x i32> <i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_0123 = shufflevector <16 x i16> %src256, <16 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_2345 = shufflevector <16 x i16> %src256, <16 x i16> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_4567 = shufflevector <16 x i16> %src256, <16 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_6789 = shufflevector <16 x i16> %src256, <16 x i16> undef, <4 x i32> <i32 6, i32 7, i32 8, i32 9>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_89AB = shufflevector <16 x i16> %src256, <16 x i16> undef, <4 x i32> <i32 8, i32 9, i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_CDEF = shufflevector <16 x i16> %src256, <16 x i16> undef, <4 x i32> <i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_01234567 = shufflevector <16 x i16> %src256, <16 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_89ABCDEF = shufflevector <16 x i16> %src256, <16 x i16> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01 = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_02_03 = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_04_05 = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_06_07 = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_08_09 = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 8, i32 9>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_0A_0B = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_0C_0D = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 12, i32 13>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_0E_0F = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_10_11 = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 16, i32 17>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_12_13 = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 18, i32 19>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_14_15 = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 20, i32 21>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_16_17 = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 22, i32 23>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_18_19 = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 24, i32 25>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_1A_1B = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 26, i32 27>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_1C_1D = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 28, i32 29>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_1E_1F = shufflevector <32 x i16> %src512, <32 x i16> undef, <2 x i32> <i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01_02_03 = shufflevector <32 x i16> %src512, <32 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_02_03_04_05 = shufflevector <32 x i16> %src512, <32 x i16> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_04_05_06_07 = shufflevector <32 x i16> %src512, <32 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_06_07_08_09 = shufflevector <32 x i16> %src512, <32 x i16> undef, <4 x i32> <i32 6, i32 7, i32 8, i32 9>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_08_09_0A_0B = shufflevector <32 x i16> %src512, <32 x i16> undef, <4 x i32> <i32 8, i32 9, i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_0C_0D_0E_0F = shufflevector <32 x i16> %src512, <32 x i16> undef, <4 x i32> <i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_10_11_12_13 = shufflevector <32 x i16> %src512, <32 x i16> undef, <4 x i32> <i32 16, i32 17, i32 18, i32 19>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_14_15_16_17 = shufflevector <32 x i16> %src512, <32 x i16> undef, <4 x i32> <i32 20, i32 21, i32 22, i32 23>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_18_19_1A_1B = shufflevector <32 x i16> %src512, <32 x i16> undef, <4 x i32> <i32 24, i32 25, i32 26, i32 27>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_1C_1D_1E_1F = shufflevector <32 x i16> %src512, <32 x i16> undef, <4 x i32> <i32 28, i32 29, i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01_02_03_04_05_06_07 = shufflevector <32 x i16> %src512, <32 x i16> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_08_09_0A_0B_0C_0D_0E_0F = shufflevector <32 x i16> %src512, <32 x i16> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_10_11_12_13_14_15_16_17 = shufflevector <32 x i16> %src512, <32 x i16> undef, <8 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_18_19_1A_1B_1C_1D_1E_1F = shufflevector <32 x i16> %src512, <32 x i16> undef, <8 x i32> <i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01_02_03_04_05_06_07_08_09_0A_0B_0C_0D_0E_0F = shufflevector <32 x i16> %src512, <32 x i16> undef, <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 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_10_11_12_13_14_15_16_17_18_19_1A_1B_1C_1D_1E_1F = shufflevector <32 x i16> %src512, <32 x i16> undef, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %V64_01 = shufflevector <4 x i16> %src64, <4 x i16> undef, <2 x i32> <i32 0, i32 1>
   %V64_23 = shufflevector <4 x i16> %src64, <4 x i16> undef, <2 x i32> <i32 2, i32 3>
@@ -348,6 +458,125 @@ define void @test_vXi8(<8 x i8> %src64, <16 x i8> %src128, <32 x i8> %src256, <6
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01_02_03_04_05_06_07_08_09_0A_0B_0C_0D_0E_0F_10_11_12_13_14_15_16_17_18_19_1A_1B_1C_1D_1E_1F = shufflevector <64 x i8> %src512, <64 x i8> undef, <32 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 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_20_21_22_23_24_25_26_27_28_29_2A_2B_2C_2D_2E_2F_30_31_32_33_34_35_36_37_38_39_3A_3B_3C_3D_3E_3F = shufflevector <64 x i8> %src512, <64 x i8> undef, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+;
+; VLEN128-LABEL: 'test_vXi8'
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V64_01 = shufflevector <8 x i8> %src64, <8 x i8> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_23 = shufflevector <8 x i8> %src64, <8 x i8> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_45 = shufflevector <8 x i8> %src64, <8 x i8> undef, <2 x i32> <i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_67 = shufflevector <8 x i8> %src64, <8 x i8> undef, <2 x i32> <i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V64_0123 = shufflevector <8 x i8> %src64, <8 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64_4567 = shufflevector <8 x i8> %src64, <8 x i8> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V128_01 = shufflevector <16 x i8> %src128, <16 x i8> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_23 = shufflevector <16 x i8> %src128, <16 x i8> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_45 = shufflevector <16 x i8> %src128, <16 x i8> undef, <2 x i32> <i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_67 = shufflevector <16 x i8> %src128, <16 x i8> undef, <2 x i32> <i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_89 = shufflevector <16 x i8> %src128, <16 x i8> undef, <2 x i32> <i32 8, i32 9>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_AB = shufflevector <16 x i8> %src128, <16 x i8> undef, <2 x i32> <i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_CD = shufflevector <16 x i8> %src128, <16 x i8> undef, <2 x i32> <i32 12, i32 13>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_EF = shufflevector <16 x i8> %src128, <16 x i8> undef, <2 x i32> <i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V128_0123 = shufflevector <16 x i8> %src128, <16 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_2345 = shufflevector <16 x i8> %src128, <16 x i8> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_4567 = shufflevector <16 x i8> %src128, <16 x i8> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_6789 = shufflevector <16 x i8> %src128, <16 x i8> undef, <4 x i32> <i32 6, i32 7, i32 8, i32 9>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_89AB = shufflevector <16 x i8> %src128, <16 x i8> undef, <4 x i32> <i32 8, i32 9, i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_CDEF = shufflevector <16 x i8> %src128, <16 x i8> undef, <4 x i32> <i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V128_01234567 = shufflevector <16 x i8> %src128, <16 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V128_89ABCDEF = shufflevector <16 x i8> %src128, <16 x i8> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_00_01 = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_02_03 = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_04_05 = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_06_07 = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_08_09 = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 8, i32 9>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0A_0B = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0C_0D = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 12, i32 13>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0E_0F = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_10_11 = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 16, i32 17>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_12_13 = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 18, i32 19>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_14_15 = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 20, i32 21>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_16_17 = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 22, i32 23>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_18_19 = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 24, i32 25>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_1A_1B = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 26, i32 27>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_1C_1D = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 28, i32 29>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_1E_1F = shufflevector <32 x i8> %src256, <32 x i8> undef, <2 x i32> <i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_00_01_02_03 = shufflevector <32 x i8> %src256, <32 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_02_03_04_05 = shufflevector <32 x i8> %src256, <32 x i8> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_04_05_06_07 = shufflevector <32 x i8> %src256, <32 x i8> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_06_07_08_09 = shufflevector <32 x i8> %src256, <32 x i8> undef, <4 x i32> <i32 6, i32 7, i32 8, i32 9>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_08_09_0A_0B = shufflevector <32 x i8> %src256, <32 x i8> undef, <4 x i32> <i32 8, i32 9, i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_0C_0D_0E_0F = shufflevector <32 x i8> %src256, <32 x i8> undef, <4 x i32> <i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_10_11_12_13 = shufflevector <32 x i8> %src256, <32 x i8> undef, <4 x i32> <i32 16, i32 17, i32 18, i32 19>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_14_15_16_17 = shufflevector <32 x i8> %src256, <32 x i8> undef, <4 x i32> <i32 20, i32 21, i32 22, i32 23>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_18_19_1A_1B = shufflevector <32 x i8> %src256, <32 x i8> undef, <4 x i32> <i32 24, i32 25, i32 26, i32 27>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_1C_1D_1E_1F = shufflevector <32 x i8> %src256, <32 x i8> undef, <4 x i32> <i32 28, i32 29, i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_00_01_02_03_04_05_06_07 = shufflevector <32 x i8> %src256, <32 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_08_09_0A_0B_0C_0D_0E_0F = shufflevector <32 x i8> %src256, <32 x i8> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_10_11_12_13_14_15_16_17 = shufflevector <32 x i8> %src256, <32 x i8> undef, <8 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V256_18_19_1A_1B_1C_1D_1E_1F = shufflevector <32 x i8> %src256, <32 x i8> undef, <8 x i32> <i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_00_01_02_03_04_05_06_07_08_09_0A_0B_0C_0D_0E_0F = shufflevector <32 x i8> %src256, <32 x i8> undef, <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 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V256_10_11_12_13_14_15_16_17_18_19_1A_1B_1C_1D_1E_1F = shufflevector <32 x i8> %src256, <32 x i8> undef, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 0, i32 1>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_02_03 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_04_05 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_06_07 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_08_09 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 8, i32 9>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_0A_0B = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_0C_0D = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 12, i32 13>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_0E_0F = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_10_11 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 16, i32 17>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_12_13 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 18, i32 19>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_14_15 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 20, i32 21>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_16_17 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 22, i32 23>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_18_19 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 24, i32 25>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_1A_1B = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 26, i32 27>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_1C_1D = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 28, i32 29>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_1E_1F = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_20_21 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 32, i32 33>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_22_23 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 34, i32 35>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_24_25 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 36, i32 37>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_26_27 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 38, i32 39>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_28_29 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 40, i32 41>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_2A_2B = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 42, i32 43>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_2C_2D = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 44, i32 45>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_2E_2F = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 46, i32 47>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_30_31 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 48, i32 49>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_32_33 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 50, i32 51>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_34_35 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 52, i32 53>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_36_37 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 54, i32 55>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_38_39 = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 56, i32 57>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_3A_3B = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 58, i32 59>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_3C_3D = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 60, i32 61>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_3E_3F = shufflevector <64 x i8> %src512, <64 x i8> undef, <2 x i32> <i32 62, i32 63>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01_02_03 = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_04_05_06_07 = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_08_09_0A_0B = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 8, i32 9, i32 10, i32 11>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_0C_0D_0E_0F = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_10_11_12_13 = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 16, i32 17, i32 18, i32 19>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_14_15_16_17 = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 20, i32 21, i32 22, i32 23>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_18_19_1A_1B = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 24, i32 25, i32 26, i32 27>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_1C_1D_1E_1F = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 28, i32 29, i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_20_21_22_23 = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 32, i32 33, i32 34, i32 35>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_24_25_26_27 = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 36, i32 37, i32 38, i32 39>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_28_29_2A_2B = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 40, i32 41, i32 42, i32 43>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_2C_2D_2E_2F = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 44, i32 45, i32 46, i32 47>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_30_31_32_33 = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 48, i32 49, i32 50, i32 51>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_34_35_36_37 = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 52, i32 53, i32 54, i32 55>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_38_39_3A_3B = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 56, i32 57, i32 58, i32 59>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_3C_3D_3E_3F = shufflevector <64 x i8> %src512, <64 x i8> undef, <4 x i32> <i32 60, i32 61, i32 62, i32 63>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01_02_03_04_05_06_07 = shufflevector <64 x i8> %src512, <64 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_08_09_0A_0B_0C_0D_0E_0F = shufflevector <64 x i8> %src512, <64 x i8> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_10_11_12_13_14_15_16_17 = shufflevector <64 x i8> %src512, <64 x i8> undef, <8 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_18_19_1A_1B_1C_1D_1E_1F = shufflevector <64 x i8> %src512, <64 x i8> undef, <8 x i32> <i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_20_21_22_23_24_25_26_27 = shufflevector <64 x i8> %src512, <64 x i8> undef, <8 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_28_29_2A_2B_2C_2D_2E_2F = shufflevector <64 x i8> %src512, <64 x i8> undef, <8 x i32> <i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_30_31_32_33_34_35_36_37 = shufflevector <64 x i8> %src512, <64 x i8> undef, <8 x i32> <i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_38_39_3A_3B_3C_3D_3E_3F = shufflevector <64 x i8> %src512, <64 x i8> undef, <8 x i32> <i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01_02_03_04_05_06_07_08_09_0A_0B_0C_0D_0E_0F = shufflevector <64 x i8> %src512, <64 x i8> undef, <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 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_10_11_12_13_14_15_16_17_18_19_1A_1B_1C_1D_1E_1F = shufflevector <64 x i8> %src512, <64 x i8> undef, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_20_21_22_23_24_25_26_27_28_29_2A_2B_2C_2D_2E_2F = shufflevector <64 x i8> %src512, <64 x i8> undef, <16 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_30_31_32_33_34_35_36_37_38_39_3A_3B_3C_3D_3E_3F = shufflevector <64 x i8> %src512, <64 x i8> undef, <16 x i32> <i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01_02_03_04_05_06_07_08_09_0A_0B_0C_0D_0E_0F_10_11_12_13_14_15_16_17_18_19_1A_1B_1C_1D_1E_1F = shufflevector <64 x i8> %src512, <64 x i8> undef, <32 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 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_20_21_22_23_24_25_26_27_28_29_2A_2B_2C_2D_2E_2F_30_31_32_33_34_35_36_37_38_39_3A_3B_3C_3D_3E_3F = shufflevector <64 x i8> %src512, <64 x i8> undef, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %V64_01 = shufflevector <8 x i8> %src64, <8 x i8> undef, <2 x i32> <i32 0, i32 1>
   %V64_23 = shufflevector <8 x i8> %src64, <8 x i8> undef, <2 x i32> <i32 2, i32 3>

>From 2c525a8042bb24e6bc61a968fd28a81b24d31094 Mon Sep 17 00:00:00 2001
From: Philip Reames <preames at rivosinc.com>
Date: Tue, 20 Feb 2024 16:47:08 -0800
Subject: [PATCH 2/3] Fix a bug with insufficiently aligned extracts

---
 llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp | 12 ++++++++----
 .../CostModel/RISCV/shuffle-extract_subvector.ll   | 14 +++++++-------
 2 files changed, 15 insertions(+), 11 deletions(-)

diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index 97b929f15f018a..ca11440bbed31c 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -436,15 +436,19 @@ InstructionCost RISCVTTIImpl::getShuffleCost(TTI::ShuffleKind Kind,
     if (Index == 0)
       return TTI::TCC_Free;
 
-    // If we're extracting a subvector at a sub-register boundary - which
-    // unfortunately we need exact vlen to identity - this is a subregister
-    // extract at worst and thus won't require a vslidedown.
+    // If we're extracting a subvector of at most m1 size at a sub-register
+    // boundary - which unfortunately we need exact vlen to identity - this is
+    // a subregister extract at worst and thus won't require a vslidedown.
+    // TODO: Extend for aligned m2, m4 subvector extracts
+    // TODO: Extend for misalgined (but contained) extracts
+    // TODO: Extend for scalable subvector types
     if (std::pair<InstructionCost, MVT> SubLT = getTypeLegalizationCost(SubTp);
         SubLT.second.isValid() && SubLT.second.isFixedLengthVector()) {
       const unsigned MinVLen = ST->getRealMinVLen();
       const unsigned MaxVLen = ST->getRealMaxVLen();
       if (MinVLen == MaxVLen &&
-          SubLT.second.getScalarSizeInBits() * Index % MinVLen == 0)
+          SubLT.second.getScalarSizeInBits() * Index % MinVLen == 0 &&
+          SubLT.second.getSizeInBits() <= MinVLen)
         return TTI::TCC_Free;
     }
 
diff --git a/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll b/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll
index 29a9ade4c5e508..3ac2b7e26650ab 100644
--- a/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/shuffle-extract_subvector.ll
@@ -30,8 +30,8 @@ define void @test_vXf64(<4 x double> %src256, <8 x double> %src512) {
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_45 = shufflevector <8 x double> %src512, <8 x double> undef, <2 x i32> <i32 4, i32 5>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_67 = shufflevector <8 x double> %src512, <8 x double> undef, <2 x i32> <i32 6, i32 7>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_0123 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_2345 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
-; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_4567 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_2345 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_4567 = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of -1 for instruction: %V512_567u = shufflevector <8 x double> %src512, <8 x double> undef, <4 x i32> <i32 5, i32 6, i32 7, i32 poison>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
@@ -69,8 +69,8 @@ define void @test_vXi64(<4 x i64> %src256, <8 x i64> %src512) {
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_45 = shufflevector <8 x i64> %src512, <8 x i64> undef, <2 x i32> <i32 4, i32 5>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_67 = shufflevector <8 x i64> %src512, <8 x i64> undef, <2 x i32> <i32 6, i32 7>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_0123 = shufflevector <8 x i64> %src512, <8 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_2345 = shufflevector <8 x i64> %src512, <8 x i64> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
-; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_4567 = shufflevector <8 x i64> %src512, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_2345 = shufflevector <8 x i64> %src512, <8 x i64> undef, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_4567 = shufflevector <8 x i64> %src512, <8 x i64> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %V256_01 = shufflevector <4 x i64> %src256, <4 x i64> undef, <2 x i32> <i32 0, i32 1>
@@ -133,7 +133,7 @@ define void @test_vXi32(<4 x i32> %src128, <8 x i32> %src256, <16 x i32> %src512
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_89AB = shufflevector <16 x i32> %src512, <16 x i32> undef, <4 x i32> <i32 8, i32 9, i32 10, i32 11>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_CDEF = shufflevector <16 x i32> %src512, <16 x i32> undef, <4 x i32> <i32 12, i32 13, i32 14, i32 15>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_01234567 = shufflevector <16 x i32> %src512, <16 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_89ABCDEF = shufflevector <16 x i32> %src512, <16 x i32> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_89ABCDEF = shufflevector <16 x i32> %src512, <16 x i32> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %V128_01 = shufflevector <4 x i32> %src128, <4 x i32> undef, <2 x i32> <i32 0, i32 1>
@@ -277,7 +277,7 @@ define void @test_vXi16(<4 x i16> %src64, <8 x i16> %src128, <16 x i16> %src256,
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_10_11_12_13_14_15_16_17 = shufflevector <32 x i16> %src512, <32 x i16> undef, <8 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_18_19_1A_1B_1C_1D_1E_1F = shufflevector <32 x i16> %src512, <32 x i16> undef, <8 x i32> <i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01_02_03_04_05_06_07_08_09_0A_0B_0C_0D_0E_0F = shufflevector <32 x i16> %src512, <32 x i16> undef, <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 14, i32 15>
-; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_10_11_12_13_14_15_16_17_18_19_1A_1B_1C_1D_1E_1F = shufflevector <32 x i16> %src512, <32 x i16> undef, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_10_11_12_13_14_15_16_17_18_19_1A_1B_1C_1D_1E_1F = shufflevector <32 x i16> %src512, <32 x i16> undef, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %V64_01 = shufflevector <4 x i16> %src64, <4 x i16> undef, <2 x i32> <i32 0, i32 1>
@@ -575,7 +575,7 @@ define void @test_vXi8(<8 x i8> %src64, <16 x i8> %src128, <32 x i8> %src256, <6
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_20_21_22_23_24_25_26_27_28_29_2A_2B_2C_2D_2E_2F = shufflevector <64 x i8> %src512, <64 x i8> undef, <16 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_30_31_32_33_34_35_36_37_38_39_3A_3B_3C_3D_3E_3F = shufflevector <64 x i8> %src512, <64 x i8> undef, <16 x i32> <i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_00_01_02_03_04_05_06_07_08_09_0A_0B_0C_0D_0E_0F_10_11_12_13_14_15_16_17_18_19_1A_1B_1C_1D_1E_1F = shufflevector <64 x i8> %src512, <64 x i8> undef, <32 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 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
-; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V512_20_21_22_23_24_25_26_27_28_29_2A_2B_2C_2D_2E_2F_30_31_32_33_34_35_36_37_38_39_3A_3B_3C_3D_3E_3F = shufflevector <64 x i8> %src512, <64 x i8> undef, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
+; VLEN128-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V512_20_21_22_23_24_25_26_27_28_29_2A_2B_2C_2D_2E_2F_30_31_32_33_34_35_36_37_38_39_3A_3B_3C_3D_3E_3F = shufflevector <64 x i8> %src512, <64 x i8> undef, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
 ; VLEN128-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %V64_01 = shufflevector <8 x i8> %src64, <8 x i8> undef, <2 x i32> <i32 0, i32 1>

>From 962163f8a0717b93b39c138a924f7d9e045ea2ce Mon Sep 17 00:00:00 2001
From: Philip Reames <preames at rivosinc.com>
Date: Wed, 21 Feb 2024 07:52:00 -0800
Subject: [PATCH 3/3] Update llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp

Co-authored-by: Luke Lau <luke_lau at icloud.com>
---
 llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index ca11440bbed31c..f04968d82e86e2 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -437,7 +437,7 @@ InstructionCost RISCVTTIImpl::getShuffleCost(TTI::ShuffleKind Kind,
       return TTI::TCC_Free;
 
     // If we're extracting a subvector of at most m1 size at a sub-register
-    // boundary - which unfortunately we need exact vlen to identity - this is
+    // boundary - which unfortunately we need exact vlen to identify - this is
     // a subregister extract at worst and thus won't require a vslidedown.
     // TODO: Extend for aligned m2, m4 subvector extracts
     // TODO: Extend for misalgined (but contained) extracts



More information about the llvm-commits mailing list