[PATCH] D98746: [clang][amdgpu] Use implicit code object default

JonChesterfield abandoned this revision. JonChesterfield added a comment. Abandoning in favour of the more conservative D101095 Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D98746/new/ https://reviews.llvm.org/D9

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

JonChesterfield added a comment. Revision at D101095 implements the suggestion above. The argument is passed to llc whenever an argument was passed to clang. Front end tests / device runtime libraries that are happy with the default can then build without the

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

JonChesterfield added a comment. Posted D101077 which is a refactor. Expect to shortly post a replacement for this which implements @yaxunl's suggestion above, i.e. pass the argument to llc whenever the user specified one to clang Repository: rG LLVM Github

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

JonChesterfield added a comment. Great context, thanks guys. I had missed that part of the compiler. We presently have a dependency edge from clang to the amdgcn target in llvm. The drawback I noticed here is it stops clang unconditionally building a runtime library for amdgcn. This patch, or a

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

yaxunl added inline comments. Comment at: clang/lib/Driver/ToolChains/Clang.cpp:1115-1124 + // Currently defaults to 3 in AMDGPUBaseInfo.cpp + // Using that default lets clang emit IR for amdgcn when llvm has been built + // without that target, provided the user wants this co

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

t-tye added a comment. I vaguely remember that clang needed to know what code object it was going to request as it used that to either validate other options, or change the format of other passed cc1 options. If that is true, then I am not sure the defaulting approach works as clang will not kn

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

JonChesterfield added inline comments. Comment at: clang/lib/Driver/ToolChains/Clang.cpp:1115-1124 + // Currently defaults to 3 in AMDGPUBaseInfo.cpp + // Using that default lets clang emit IR for amdgcn when llvm has been built + // without that target, provided the user want

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

yaxunl added inline comments. Comment at: clang/lib/Driver/ToolChains/Clang.cpp:1115-1124 + // Currently defaults to 3 in AMDGPUBaseInfo.cpp + // Using that default lets clang emit IR for amdgcn when llvm has been built + // without that target, provided the user wants this co

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

t-tye added a comment. I have no opinion, just making an observation and defer to @kzhuravl . Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D98746/new/ https://reviews.llvm.org/D98746 ___ cfe-commits mail

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

jdoerfert added inline comments. Comment at: clang/lib/Driver/ToolChains/Clang.cpp:1115-1124 + // Currently defaults to 3 in AMDGPUBaseInfo.cpp + // Using that default lets clang emit IR for amdgcn when llvm has been built + // without that target, provided the user wants this

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

JonChesterfield added inline comments. Comment at: clang/lib/Driver/ToolChains/Clang.cpp:1115-1124 + // Currently defaults to 3 in AMDGPUBaseInfo.cpp + // Using that default lets clang emit IR for amdgcn when llvm has been built + // without that target, provided the user want

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

t-tye added inline comments. Comment at: clang/lib/Driver/ToolChains/Clang.cpp:1115-1124 + // Currently defaults to 3 in AMDGPUBaseInfo.cpp + // Using that default lets clang emit IR for amdgcn when llvm has been built + // without that target, provided the user wants this cod

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

JonChesterfield added a comment. Thanks. I'm going to wait for some of the rocm people to pass judgement too as this code path is shared with hip / opencl etc. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D98746/new/ https://reviews.llvm.org/D9874

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

jdoerfert accepted this revision. jdoerfert added a comment. This revision is now accepted and ready to land. LG, it unbreaks the build and is not totally bananas. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D98746/new/ https://reviews.llvm.org/D9

[PATCH] D98746: [clang][amdgpu] Use implicit code object default

JonChesterfield created this revision. JonChesterfield added reviewers: yaxunl, t-tye, kzhuravl, ronlieb, b-sumner. Herald added subscribers: kerbowa, pengfei, tpr, dstuttard, nhaehnle, jvesely. JonChesterfield requested review of this revision. Herald added subscribers: cfe-commits, sstefan1, wdng