From 94c163c34e4d1d9f2fd7887f198a6fac3c71e271 Mon Sep 17 00:00:00 2001
From: Tim Renouf <tpr.llvm@botech.co.uk>
Date: Fri, 22 Mar 2019 15:53:50 +0000
Subject: [PATCH] InstCombineSimplifyDemanded: Allow v3 results for AMDGCN
 buffer and image intrinsics

This helps to avoid the situation where RA spots that only 3 of the
v4f32 result of a load are used, and immediately reallocates the 4th
register for something else, requiring a stall waiting for the load.

Differential Revision: https://reviews.llvm.org/D58906

Change-Id: I947661edfd5715f62361a02b100f14aeeada29aa
llvm-svn: 356768
---
 .../InstCombine/InstCombineSimplifyDemanded.cpp    |  3 +-
 .../AMDGPU/amdgcn-demanded-vector-elts.ll          | 84 ++++++++++------------
 2 files changed, 37 insertions(+), 50 deletions(-)

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp b/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
index b79a4d7..2dea7ee 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
@@ -1003,8 +1003,7 @@ Value *InstCombiner::simplifyAMDGCNMemoryIntrinsicDemanded(IntrinsicInst *II,
       NewDMask = ConstantInt::get(DMask->getType(), NewDMaskVal);
   }
 
-  // TODO: Handle 3 vectors when supported in code gen.
-  unsigned NewNumElts = PowerOf2Ceil(DemandedElts.countPopulation());
+  unsigned NewNumElts = DemandedElts.countPopulation();
   if (!NewNumElts)
     return UndefValue::get(II->getType());
 
diff --git a/llvm/test/Transforms/InstCombine/AMDGPU/amdgcn-demanded-vector-elts.ll b/llvm/test/Transforms/InstCombine/AMDGPU/amdgcn-demanded-vector-elts.ll
index 6c03ed2..9c45cf5 100644
--- a/llvm/test/Transforms/InstCombine/AMDGPU/amdgcn-demanded-vector-elts.ll
+++ b/llvm/test/Transforms/InstCombine/AMDGPU/amdgcn-demanded-vector-elts.ll
@@ -75,8 +75,8 @@ define amdgpu_ps float @extract_elt1_buffer_load_v4f32(<4 x i32> inreg %rsrc, i3
 }
 
 ; CHECK-LABEL: @extract_elt2_buffer_load_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i1 false, i1 false)
-; CHECK-NEXT: %elt1 = extractelement <4 x float> %data, i32 2
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.buffer.load.v3f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i1 false, i1 false)
+; CHECK-NEXT: %elt1 = extractelement <3 x float> %data, i32 2
 ; CHECK-NEXT: ret float %elt1
 define amdgpu_ps float @extract_elt2_buffer_load_v4f32(<4 x i32> inreg %rsrc, i32 %idx, i32 %ofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i1 false, i1 false)
@@ -104,8 +104,8 @@ define amdgpu_ps <2 x float> @extract_elt0_elt1_buffer_load_v4f32(<4 x i32> inre
 }
 
 ; CHECK-LABEL: @extract_elt1_elt2_buffer_load_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i1 false, i1 false)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <2 x i32> <i32 1, i32 2>
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.buffer.load.v3f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i1 false, i1 false)
+; CHECK-NEXT: %shuf = shufflevector <3 x float> %data, <3 x float> undef, <2 x i32> <i32 1, i32 2>
 ; CHECK-NEXT: ret <2 x float> %shuf
 define amdgpu_ps <2 x float> @extract_elt1_elt2_buffer_load_v4f32(<4 x i32> inreg %rsrc, i32 %idx, i32 %ofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i1 false, i1 false)
@@ -124,9 +124,8 @@ define amdgpu_ps <2 x float> @extract_elt2_elt3_buffer_load_v4f32(<4 x i32> inre
 }
 
 ; CHECK-LABEL: @extract_elt0_elt1_elt2_buffer_load_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i1 false, i1 false)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
-; CHECK-NEXT: ret <3 x float> %shuf
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.buffer.load.v3f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i1 false, i1 false)
+; CHECK-NEXT: ret <3 x float> %data
 define amdgpu_ps <3 x float> @extract_elt0_elt1_elt2_buffer_load_v4f32(<4 x i32> inreg %rsrc, i32 %idx, i32 %ofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i1 false, i1 false)
   %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
@@ -390,8 +389,8 @@ define amdgpu_ps float @extract_elt1_raw_buffer_load_v4f32(<4 x i32> inreg %rsrc
 }
 
 ; CHECK-LABEL: @extract_elt2_raw_buffer_load_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %elt1 = extractelement <4 x float> %data, i32 2
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.raw.buffer.load.v3f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: %elt1 = extractelement <3 x float> %data, i32 2
 ; CHECK-NEXT: ret float %elt1
 define amdgpu_ps float @extract_elt2_raw_buffer_load_v4f32(<4 x i32> inreg %rsrc, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
@@ -419,8 +418,8 @@ define amdgpu_ps <2 x float> @extract_elt0_elt1_raw_buffer_load_v4f32(<4 x i32>
 }
 
 ; CHECK-LABEL: @extract_elt1_elt2_raw_buffer_load_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <2 x i32> <i32 1, i32 2>
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.raw.buffer.load.v3f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: %shuf = shufflevector <3 x float> %data, <3 x float> undef, <2 x i32> <i32 1, i32 2>
 ; CHECK-NEXT: ret <2 x float> %shuf
 define amdgpu_ps <2 x float> @extract_elt1_elt2_raw_buffer_load_v4f32(<4 x i32> inreg %rsrc, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
@@ -439,9 +438,8 @@ define amdgpu_ps <2 x float> @extract_elt2_elt3_raw_buffer_load_v4f32(<4 x i32>
 }
 
 ; CHECK-LABEL: @extract_elt0_elt1_elt2_raw_buffer_load_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
-; CHECK-NEXT: ret <3 x float> %shuf
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.raw.buffer.load.v3f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: ret <3 x float> %data
 define amdgpu_ps <3 x float> @extract_elt0_elt1_elt2_raw_buffer_load_v4f32(<4 x i32> inreg %rsrc, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
   %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
@@ -630,8 +628,8 @@ define amdgpu_ps float @extract_elt1_raw_buffer_load_format_v4f32(<4 x i32> inre
 }
 
 ; CHECK-LABEL: @extract_elt2_raw_buffer_load_format_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %elt1 = extractelement <4 x float> %data, i32 2
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.raw.buffer.load.format.v3f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: %elt1 = extractelement <3 x float> %data, i32 2
 ; CHECK-NEXT: ret float %elt1
 define amdgpu_ps float @extract_elt2_raw_buffer_load_format_v4f32(<4 x i32> inreg %rsrc, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
@@ -659,8 +657,8 @@ define amdgpu_ps <2 x float> @extract_elt0_elt1_raw_buffer_load_format_v4f32(<4
 }
 
 ; CHECK-LABEL: @extract_elt1_elt2_raw_buffer_load_format_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <2 x i32> <i32 1, i32 2>
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.raw.buffer.load.format.v3f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: %shuf = shufflevector <3 x float> %data, <3 x float> undef, <2 x i32> <i32 1, i32 2>
 ; CHECK-NEXT: ret <2 x float> %shuf
 define amdgpu_ps <2 x float> @extract_elt1_elt2_raw_buffer_load_format_v4f32(<4 x i32> inreg %rsrc, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
@@ -679,9 +677,8 @@ define amdgpu_ps <2 x float> @extract_elt2_elt3_raw_buffer_load_format_v4f32(<4
 }
 
 ; CHECK-LABEL: @extract_elt0_elt1_elt2_raw_buffer_load_format_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
-; CHECK-NEXT: ret <3 x float> %shuf
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.raw.buffer.load.format.v3f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: ret <3 x float> %data
 define amdgpu_ps <3 x float> @extract_elt0_elt1_elt2_raw_buffer_load_format_v4f32(<4 x i32> inreg %rsrc, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.raw.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %ofs, i32 %sofs, i32 0)
   %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
@@ -870,8 +867,8 @@ define amdgpu_ps float @extract_elt1_struct_buffer_load_v4f32(<4 x i32> inreg %r
 }
 
 ; CHECK-LABEL: @extract_elt2_struct_buffer_load_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %elt1 = extractelement <4 x float> %data, i32 2
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.struct.buffer.load.v3f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: %elt1 = extractelement <3 x float> %data, i32 2
 ; CHECK-NEXT: ret float %elt1
 define amdgpu_ps float @extract_elt2_struct_buffer_load_v4f32(<4 x i32> inreg %rsrc, i32 %idx, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
@@ -899,8 +896,8 @@ define amdgpu_ps <2 x float> @extract_elt0_elt1_struct_buffer_load_v4f32(<4 x i3
 }
 
 ; CHECK-LABEL: @extract_elt1_elt2_struct_buffer_load_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <2 x i32> <i32 1, i32 2>
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.struct.buffer.load.v3f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: %shuf = shufflevector <3 x float> %data, <3 x float> undef, <2 x i32> <i32 1, i32 2>
 ; CHECK-NEXT: ret <2 x float> %shuf
 define amdgpu_ps <2 x float> @extract_elt1_elt2_struct_buffer_load_v4f32(<4 x i32> inreg %rsrc, i32 %idx, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
@@ -919,9 +916,8 @@ define amdgpu_ps <2 x float> @extract_elt2_elt3_struct_buffer_load_v4f32(<4 x i3
 }
 
 ; CHECK-LABEL: @extract_elt0_elt1_elt2_struct_buffer_load_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
-; CHECK-NEXT: ret <3 x float> %shuf
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.struct.buffer.load.v3f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: ret <3 x float> %data
 define amdgpu_ps <3 x float> @extract_elt0_elt1_elt2_struct_buffer_load_v4f32(<4 x i32> inreg %rsrc, i32 %idx, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
   %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
@@ -1110,8 +1106,8 @@ define amdgpu_ps float @extract_elt1_struct_buffer_load_format_v4f32(<4 x i32> i
 }
 
 ; CHECK-LABEL: @extract_elt2_struct_buffer_load_format_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %elt1 = extractelement <4 x float> %data, i32 2
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.struct.buffer.load.format.v3f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: %elt1 = extractelement <3 x float> %data, i32 2
 ; CHECK-NEXT: ret float %elt1
 define amdgpu_ps float @extract_elt2_struct_buffer_load_format_v4f32(<4 x i32> inreg %rsrc, i32 %idx, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
@@ -1139,8 +1135,8 @@ define amdgpu_ps <2 x float> @extract_elt0_elt1_struct_buffer_load_format_v4f32(
 }
 
 ; CHECK-LABEL: @extract_elt1_elt2_struct_buffer_load_format_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <2 x i32> <i32 1, i32 2>
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.struct.buffer.load.format.v3f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: %shuf = shufflevector <3 x float> %data, <3 x float> undef, <2 x i32> <i32 1, i32 2>
 ; CHECK-NEXT: ret <2 x float> %shuf
 define amdgpu_ps <2 x float> @extract_elt1_elt2_struct_buffer_load_format_v4f32(<4 x i32> inreg %rsrc, i32 %idx, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
@@ -1159,9 +1155,8 @@ define amdgpu_ps <2 x float> @extract_elt2_elt3_struct_buffer_load_format_v4f32(
 }
 
 ; CHECK-LABEL: @extract_elt0_elt1_elt2_struct_buffer_load_format_v4f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
-; CHECK-NEXT: ret <3 x float> %shuf
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.struct.buffer.load.format.v3f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
+; CHECK-NEXT: ret <3 x float> %data
 define amdgpu_ps <3 x float> @extract_elt0_elt1_elt2_struct_buffer_load_format_v4f32(<4 x i32> inreg %rsrc, i32 %idx, i32 %ofs, i32 %sofs) #0 {
   %data = call <4 x float> @llvm.amdgcn.struct.buffer.load.format.v4f32(<4 x i32> %rsrc, i32 %idx, i32 %ofs, i32 %sofs, i32 0)
   %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
@@ -1458,9 +1453,8 @@ define amdgpu_ps <3 x float> @extract_elt0_elt1_elt2_dmask_0101_image_sample_1d_
 }
 
 ; CHECK-LABEL: @extract_elt0_elt1_elt2_dmask_0111_image_sample_1d_v4f32_f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 7, float %s, <8 x i32> %sampler, <4 x i32> %rsrc, i1 false, i32 0, i32 0)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
-; CHECK-NEXT: ret <3 x float> %shuf
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.image.sample.1d.v3f32.f32(i32 7, float %s, <8 x i32> %sampler, <4 x i32> %rsrc, i1 false, i32 0, i32 0)
+; CHECK-NEXT: ret <3 x float> %data
 define amdgpu_ps <3 x float> @extract_elt0_elt1_elt2_dmask_0111_image_sample_1d_v4f32_f32(float %s, <8 x i32> inreg %sampler, <4 x i32> inreg %rsrc) #0 {
   %data = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 7, float %s, <8 x i32> %sampler, <4 x i32> %rsrc, i1 false, i32 0, i32 0)
   %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
@@ -1468,9 +1462,8 @@ define amdgpu_ps <3 x float> @extract_elt0_elt1_elt2_dmask_0111_image_sample_1d_
 }
 
 ; CHECK-LABEL: @extract_elt0_elt1_elt2_dmask_1111_image_sample_1d_v4f32_f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 7, float %s, <8 x i32> %sampler, <4 x i32> %rsrc, i1 false, i32 0, i32 0)
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
-; CHECK-NEXT: ret <3 x float> %shuf
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.image.sample.1d.v3f32.f32(i32 7, float %s, <8 x i32> %sampler, <4 x i32> %rsrc, i1 false, i32 0, i32 0)
+; CHECK-NEXT: ret <3 x float> %data
 define amdgpu_ps <3 x float> @extract_elt0_elt1_elt2_dmask_1111_image_sample_1d_v4f32_f32(float %s, <8 x i32> inreg %sampler, <4 x i32> inreg %rsrc) #0 {
   %data = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %s, <8 x i32> %sampler, <4 x i32> %rsrc, i1 false, i32 0, i32 0)
   %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
@@ -1593,13 +1586,8 @@ declare <4 x float> @llvm.amdgcn.image.sample.lz.1d.v4f32.f32(i32, float, <8 x i
 ; --------------------------------------------------------------------
 
 ; CHECK-LABEL: @extract_elt1_elt2_elt3_image_sample_cd_1d_v4f32_f32_f32(
-; CHECK-NEXT: %data = call <4 x float> @llvm.amdgcn.image.sample.cd.1d.v4f32.f32.f32(i32 14, float %dsdh, float %dsdv, float %s, <8 x i32> %sampler, <4 x i32> %rsrc, i1 false, i32 0, i32 0)
-
-; TODO: This should be combined with the following shufflevector
-; CHECK-NEXT: %1 = shufflevector <4 x float> %data, <4 x float> undef, <4 x i32> <i32 undef, i32 0, i32 1, i32 2>
-
-; CHECK-NEXT: %shuf = shufflevector <4 x float> %1, <4 x float> undef, <3 x i32> <i32 1, i32 2, i32 3>
-; CHECK-NEXT: ret <3 x float> %shuf
+; CHECK-NEXT: %data = call <3 x float> @llvm.amdgcn.image.sample.cd.1d.v3f32.f32.f32(i32 14, float %dsdh, float %dsdv, float %s, <8 x i32> %sampler, <4 x i32> %rsrc, i1 false, i32 0, i32 0)
+; CHECK-NEXT: ret <3 x float> %data
 define amdgpu_ps <3 x float> @extract_elt1_elt2_elt3_image_sample_cd_1d_v4f32_f32_f32(float %dsdh, float %dsdv, float %s, <8 x i32> inreg %sampler, <4 x i32> inreg %rsrc) #0 {
   %data = call <4 x float> @llvm.amdgcn.image.sample.cd.1d.v4f32.f32.f32(i32 15, float %dsdh, float %dsdv, float %s, <8 x i32> %sampler, <4 x i32> %rsrc, i1 false, i32 0, i32 0)
   %shuf = shufflevector <4 x float> %data, <4 x float> undef, <3 x i32> <i32 1, i32 2, i32 3>
-- 
2.7.4