[llvm] r239368 - [NVPTX] run SROA after NVPTXFavorNonGenericAddrSpaces
Jingyue Wu
jingyue at google.com
Mon Jun 8 17:05:56 PDT 2015
Author: jingyue
Date: Mon Jun 8 19:05:56 2015
New Revision: 239368
URL: http://llvm.org/viewvc/llvm-project?rev=239368&view=rev
Log:
[NVPTX] run SROA after NVPTXFavorNonGenericAddrSpaces
Summary:
This cleans up most allocas NVPTXLowerKernelArgs emits for byval
parameters.
Test Plan: makes bug21465.ll more stronger to verify no redundant local load/store.
Reviewers: eliben, jholewinski
Reviewed By: eliben, jholewinski
Subscribers: jholewinski, llvm-commits
Differential Revision: http://reviews.llvm.org/D10322
Modified:
llvm/trunk/lib/Target/NVPTX/NVPTXTargetMachine.cpp
llvm/trunk/test/CodeGen/NVPTX/bug21465.ll
Modified: llvm/trunk/lib/Target/NVPTX/NVPTXTargetMachine.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/NVPTX/NVPTXTargetMachine.cpp?rev=239368&r1=239367&r2=239368&view=diff
==============================================================================
--- llvm/trunk/lib/Target/NVPTX/NVPTXTargetMachine.cpp (original)
+++ llvm/trunk/lib/Target/NVPTX/NVPTXTargetMachine.cpp Mon Jun 8 19:05:56 2015
@@ -165,6 +165,11 @@ void NVPTXPassConfig::addIRPasses() {
addPass(createGenericToNVVMPass());
addPass(createNVPTXLowerKernelArgsPass(&getNVPTXTargetMachine()));
addPass(createNVPTXFavorNonGenericAddrSpacesPass());
+ // NVPTXLowerKernelArgs emits alloca for byval parameters which can often
+ // be eliminated by SROA. We do not run SROA right after NVPTXLowerKernelArgs
+ // because we plan to merge NVPTXLowerKernelArgs and
+ // NVPTXFavorNonGenericAddrSpaces into one pass.
+ addPass(createSROAPass());
// FavorNonGenericAddrSpaces shortcuts unnecessary addrspacecasts, and leave
// them unused. We could remove dead code in an ad-hoc manner, but that
// requires manual work and might be error-prone.
Modified: llvm/trunk/test/CodeGen/NVPTX/bug21465.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/NVPTX/bug21465.ll?rev=239368&r1=239367&r2=239368&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/NVPTX/bug21465.ll (original)
+++ llvm/trunk/test/CodeGen/NVPTX/bug21465.ll Mon Jun 8 19:05:56 2015
@@ -14,9 +14,10 @@ entry:
; CHECK: addrspacecast %struct.S* %input to %struct.S addrspace(101)*
%b = getelementptr inbounds %struct.S, %struct.S* %input, i64 0, i32 1
%0 = load i32, i32* %b, align 4
-; PTX: ld.param.u32 %r{{[0-9]+}}, {{\[}}[[BASE:%rd[0-9]+]]{{\]}}
-; PTX-NEXT: ld.param.u32 %r{{[0-9]+}}, {{\[}}[[BASE]]+4{{\]}}
+; PTX-NOT: ld.param.u32 {{%r[0-9]+}}, [{{%rd[0-9]+}}]
+; PTX: ld.param.u32 [[value:%r[0-9]+]], [{{%rd[0-9]+}}+4]
store i32 %0, i32* %output, align 4
+; PTX-NEXT: st.global.u32 [{{%rd[0-9]+}}], [[value]]
ret void
}
More information about the llvm-commits
mailing list