Skip to content

Commit 3838c8d

Browse files
alex-tarsenm
andauthored
Update llvm/test/CodeGen/AMDGPU/triton_regression_no_waterfall.mir
Co-authored-by: Matt Arsenault <[email protected]>
1 parent 6c4287d commit 3838c8d

File tree

1 file changed

+0
-37
lines changed

1 file changed

+0
-37
lines changed

llvm/test/CodeGen/AMDGPU/triton_regression_no_waterfall.mir

Lines changed: 0 additions & 37 deletions
Original file line numberDiff line numberDiff line change
@@ -1,42 +1,5 @@
11
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 5
22
# RUN: llc -mtriple=amdgcn -mcpu=gfx942 -run-pass=si-fix-sgpr-copies -o - %s | FileCheck --check-prefix=GCN %s
3-
--- |
4-
source_filename = "test1.ll"
5-
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-p9:192:256:256:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7:8:9"
6-
target triple = "amdgcn"
7-
8-
define amdgpu_kernel void @test_should_convert_to_v_readfirstlane_b32(float %fval, i32 %arg1, i32 %arg2, ptr addrspace(4) %out) #0 {
9-
entry:
10-
%test_should_convert_to_v_readfirstlane_b32.kernarg.segment = call nonnull align 16 dereferenceable(280) ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
11-
%fval.kernarg.offset = getelementptr inbounds i8, ptr addrspace(4) %test_should_convert_to_v_readfirstlane_b32.kernarg.segment, i64 36, !amdgpu.uniform !0
12-
%0 = load <3 x i32>, ptr addrspace(4) %fval.kernarg.offset, align 4, !invariant.load !0
13-
%fval.load1 = extractelement <3 x i32> %0, i32 0
14-
%1 = bitcast i32 %fval.load1 to float
15-
%arg1.load2 = extractelement <3 x i32> %0, i32 1
16-
%arg2.load3 = extractelement <3 x i32> %0, i32 2
17-
%conv = fptoui float %1 to i32
18-
%shl = shl i32 %conv, 16
19-
%or = or i32 %shl, %arg1.load2
20-
%and = and i32 %or, %arg2.load3
21-
%shr = lshr i32 %and, 2
22-
%sgpr128_0 = insertelement <4 x i32> undef, i32 %shr, i32 0
23-
%sgpr128_1 = insertelement <4 x i32> %sgpr128_0, i32 %or, i32 1
24-
%sgpr128_2 = insertelement <4 x i32> %sgpr128_1, i32 %and, i32 2
25-
%sgpr128_3 = insertelement <4 x i32> %sgpr128_2, i32 %shr, i32 3
26-
call void @llvm.amdgcn.raw.buffer.store.i32(i32 %arg1.load2, <4 x i32> %sgpr128_3, i32 0, i32 0, i32 2)
27-
ret void
28-
}
29-
30-
declare void @llvm.amdgcn.raw.buffer.store.i32(i32, <4 x i32>, i32, i32, i32 immarg) #1
31-
32-
declare noundef align 4 ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr() #2
33-
34-
attributes #0 = { "target-cpu"="gfx942" }
35-
attributes #1 = { nocallback nofree nosync nounwind willreturn memory(write) "target-cpu"="gfx942" }
36-
attributes #2 = { nocallback nofree nosync nounwind speculatable willreturn memory(none) }
37-
38-
!0 = !{}
39-
...
403
---
414
name: test_should_convert_to_v_readfirstlane_b32
425
alignment: 1

0 commit comments

Comments
 (0)