replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

AMDGPU: Remove old intrinsic uses 

llvm-svn: 303305
This commit is contained in:
Matt Arsenault 2017-05-17 21:38:21 +00:00
parent cd41bc8dec
commit a53292779a
5 changed files with 13 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
llvm/test/CodeGen/AMDGPU/coalescer_distribute.ll
View File

@ -5,7 +5,7 @@ target triple = "amdgcn--"
define spir_kernel void @hoge() {
bb:
%tmp = tail call i32 @llvm.r600.read.tidig.x()
%tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
br i1 undef, label %bb2, label %bb23
bb2:
@ -50,4 +50,4 @@ bb34:
ret void
}
declare i32 @llvm.r600.read.tidig.x()
declare i32 @llvm.amdgcn.workitem.id.x()

8
llvm/test/CodeGen/AMDGPU/endcf-loop-header.ll
View File

@ -1,4 +1,4 @@
; RUN: llc < %s -march=amdgcn -mcpu=SI -verify-machineinstrs | FileCheck %s
; RUN: llc < %s -march=amdgcn -verify-machineinstrs | FileCheck %s
; This tests that the llvm.SI.end.cf intrinsic is not inserted into the
; loop block. This intrinsic will be lowered to s_or_b64 by the code
@ -14,7 +14,7 @@
; CHECK: s_cbranch_execnz [[LOOP_LABEL]]
define amdgpu_kernel void @test(i32 addrspace(1)* %out) {
entry:
%cond = call i32 @llvm.r600.read.tidig.x() #0
%cond = call i32 @llvm.amdgcn.workitem.id.x() #0
%tmp0 = icmp eq i32 %cond, 0
br i1 %tmp0, label %if, label %loop
@ -34,6 +34,6 @@ done:
ret void
}
declare i32 @llvm.r600.read.tidig.x() #0
declare i32 @llvm.amdgcn.workitem.id.x() #0
attributes #0 = { readnone }
attributes #0 = { nounwind readnone }

6
llvm/test/CodeGen/AMDGPU/i1-copy-phi.ll
View File

@ -12,7 +12,7 @@
; SI: s_endpgm
define amdgpu_kernel void @br_i1_phi(i32 %arg) {
bb:
%tidig = call i32 @llvm.r600.read.tidig.x() #0
%tidig = call i32 @llvm.amdgcn.workitem.id.x()
%cmp = trunc i32 %tidig to i1
br i1 %cmp, label %bb2, label %bb3
@ -32,6 +32,6 @@ bb6: ; preds = %bb4, %bb3
ret void
}
declare i32 @llvm.r600.read.tidig.x() #0
declare i32 @llvm.amdgcn.workitem.id.x() #0
attributes #0 = { readnone }
attributes #0 = { nounwind readnone }

6
llvm/test/CodeGen/AMDGPU/llvm.amdgcn.s.barrier.ll
View File

@ -7,14 +7,13 @@
; GFX9: flat_store_dword
; GFX9-NOT: s_waitcnt
; GCN: s_barrier
define amdgpu_kernel void @test_barrier(i32 addrspace(1)* %out) #0 {
define amdgpu_kernel void @test_barrier(i32 addrspace(1)* %out, i32 %size) #0 {
entry:
%tmp = call i32 @llvm.amdgcn.workitem.id.x()
%tmp1 = getelementptr i32, i32 addrspace(1)* %out, i32 %tmp
store i32 %tmp, i32 addrspace(1)* %tmp1
call void @llvm.amdgcn.s.barrier()
%tmp2 = call i32 @llvm.r600.read.local.size.x()
%tmp3 = sub i32 %tmp2, 1
%tmp3 = sub i32 %size, 1
%tmp4 = sub i32 %tmp3, %tmp
%tmp5 = getelementptr i32, i32 addrspace(1)* %out, i32 %tmp4
%tmp6 = load i32, i32 addrspace(1)* %tmp5
@ -24,7 +23,6 @@ entry:
declare void @llvm.amdgcn.s.barrier() #1
declare i32 @llvm.amdgcn.workitem.id.x() #2
declare i32 @llvm.r600.read.local.size.x() #2
attributes #0 = { nounwind }
attributes #1 = { convergent nounwind }

4
llvm/test/CodeGen/AMDGPU/vgpr-spill-emergency-stack-slot-compute.ll
View File

@ -186,7 +186,7 @@ bb12: ; preds = %bb145, %bb
%tmp140 = phi float [ 0.000000e+00, %bb ], [ %tmp405, %bb145 ]
%tmp141 = phi float [ 0.000000e+00, %bb ], [ %tmp406, %bb145 ]
%tmp142 = bitcast float %tmp95 to i32
%tid = call i32 @llvm.r600.read.tidig.x() #1
%tid = call i32 @llvm.amdgcn.workitem.id.x() #1
%tmp143 = icmp sgt i32 %tmp142, %tid
br i1 %tmp143, label %bb144, label %bb145
@ -593,7 +593,7 @@ bb145: ; preds = %bb12
br label %bb12
}
declare i32 @llvm.r600.read.tidig.x() #1
declare i32 @llvm.amdgcn.workitem.id.x() #1
attributes #0 = { nounwind }
attributes #1 = { nounwind readnone }