mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-25 20:23:11 +01:00
551dde9f3b
See bug 34765: https://bugs.llvm.org//show_bug.cgi?id=34765 Reviewers: tamazov, SamWot, arsenm, vpykhtin Differential Revision: https://reviews.llvm.org/D40088 llvm-svn: 318675
20 lines
925 B
LLVM
20 lines
925 B
LLVM
; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -verify-machineinstrs< %s | FileCheck -check-prefix=SI %s
|
|
; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tonga -verify-machineinstrs< %s | FileCheck -check-prefix=SI %s
|
|
|
|
; Copy VGPR -> SGPR used twice as an instruction operand, which is then
|
|
; used in an REG_SEQUENCE that also needs to be handled.
|
|
|
|
; SI-LABEL: {{^}}test_dup_operands:
|
|
; SI: v_add_{{[iu]}}32_e32
|
|
define amdgpu_kernel void @test_dup_operands(<2 x i32> addrspace(1)* noalias %out, <2 x i32> addrspace(1)* noalias %in) {
|
|
%a = load <2 x i32>, <2 x i32> addrspace(1)* %in
|
|
%lo = extractelement <2 x i32> %a, i32 0
|
|
%hi = extractelement <2 x i32> %a, i32 1
|
|
%add = add i32 %lo, %lo
|
|
%vec0 = insertelement <2 x i32> undef, i32 %add, i32 0
|
|
%vec1 = insertelement <2 x i32> %vec0, i32 %hi, i32 1
|
|
store <2 x i32> %vec1, <2 x i32> addrspace(1)* %out, align 8
|
|
ret void
|
|
}
|
|
|