[PATCH] D114957: [AMDGPU] Change llvm.amdgcn.image.bvh.intersect.ray to take vec3 args

yaxunl added a comment. In D114957#3166817 , @foad wrote: > This is a flag-day change to the signatures of the LLVM intrinsics and the > OpenCL builtins. Is that OK? This breaks users' code. If we have to do this, at least let clang emit a pre-defined

[PATCH] D114957: [AMDGPU] Change llvm.amdgcn.image.bvh.intersect.ray to take vec3 args

yaxunl added a comment. In D114957#3166861 , @arsenm wrote: > In D114957#3166858 , @yaxunl wrote: > >> In D114957#3166817 , @foad wrote: >> >>> This is a flag-day change t

[PATCH] D114957: [AMDGPU] Change llvm.amdgcn.image.bvh.intersect.ray to take vec3 args

yaxunl added a comment. In D114957#3166974 , @foad wrote: > In D114957#3166948 , @b-sumner > wrote: > >> In D114957#3166936 , @foad wrote: >> >>> In D114957#3166858

[PATCH] D114957: [AMDGPU] Change llvm.amdgcn.image.bvh.intersect.ray to take vec3 args

yaxunl added a comment. In D114957#3166936 , @foad wrote: > In D114957#3166858 , @yaxunl wrote: > >> In D114957#3166817 , @foad wrote: >> >>> This is a flag-day change to

[PATCH] D114957: [AMDGPU] Change llvm.amdgcn.image.bvh.intersect.ray to take vec3 args

yaxunl added a comment. LGTM from clang side. Thanks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D114957/new/ https://reviews.llvm.org/D114957 ___ cfe-commits mailing list cfe-commits@lists.llvm.org ht

[PATCH] D114957: [AMDGPU] Change llvm.amdgcn.image.bvh.intersect.ray to take vec3 args

yaxunl added a comment. In D114957#3167703 , @arsenm wrote: > In D114957#3167700 , @arsenm wrote: > >> I think this macro is purely terrible and should not be added (and at least >> should be all caps?). If we ca

[PATCH] D115039: [HIP] Fix -fgpu-rdc for Windows

yaxunl created this revision. yaxunl added a reviewer: tra. yaxunl requested review of this revision. This patch fixes issues for -fgpu-rdc for Windows MSVC toolchain: Fix COFF specific section flags and remove section types in llvm-mc input file for Windows. Escape fatbin path in llvm-mc input

[PATCH] D115032: [AMDGPU] Change llvm.amdgcn.image.bvh.intersect.ray to take vec3 args

yaxunl accepted this revision. yaxunl added a comment. Ideally, we could let the builtins accept both vec3 and vec4. But I am OK with this for now. I think the overhead may be minimal. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D115032/new/ http

[PATCH] D135269: [AMDGPU] Disable bool range metadata to workaround backend issue

yaxunl added a comment. In D135269#3863470 , @nikic wrote: > Checking back here, have you made any progress on reducing the issue? > > cc @arsenm for awareness No. I am busy with other work and have not got time to get back on it. Repository: rG LLVM

[PATCH] D136311: [CUDA,NVPTX] Implement __bf16 support for NVPTX.

yaxunl accepted this revision. yaxunl added a comment. LGTM. Thanks. Do you plan to support arithmetic operators for bf16 or implement the FMA instruction support? Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D136311/new/ https://reviews.llvm.org

[PATCH] D136701: [LinkerWrapper] Perform device linking steps in parallel

yaxunl added a comment. In general, I think parallelizing the linking workload for multiple GPU's in the linker wrapper is a useful feature. I am not sure whether the workload to be parallelized includes the LLVM passes and codegen, which is usually the bottleneck. Parallelizing this workload w

[PATCH] D136854: [HIP] add -fhiplib-add-rpath

yaxunl created this revision. yaxunl added a reviewer: tra. Herald added a project: All. yaxunl requested review of this revision. Herald added a subscriber: MaskRay. Add an option -f[no-]hiplib-add-rpath to control whether to pass -rpath to linker for HIP runtime library. By default it is off to

[PATCH] D136859: [HIP] add fmax/fmin for fp16

yaxunl created this revision. yaxunl added reviewers: tra, b-sumner. Herald added a project: All. yaxunl requested review of this revision. https://reviews.llvm.org/D136859 Files: clang/lib/Headers/__clang_hip_libdevice_declares.h Index: clang/lib/Headers/__clang_hip_libdevice_declares.h

[PATCH] D136854: [HIP] add -fhiplib-add-rpath

yaxunl added a comment. In D136854#3889141 , @MaskRay wrote: > If an option does not affect compilation, I prefer `--` to `-f` I will rename it as --offload-add-rpath Comment at: clang/include/clang/Driver/Options.td:4158 + HelpText<

[PATCH] D136854: [HIP] add -fhiplib-add-rpath

yaxunl updated this revision to Diff 471289. yaxunl added a comment. rename to --offload-add-rpath CHANGES SINCE LAST ACTION https://reviews.llvm.org/D136854/new/ https://reviews.llvm.org/D136854 Files: clang/include/clang/Driver/Options.td clang/lib/Driver/ToolChains/Linux.cpp clang/t

[PATCH] D136854: [HIP] add --offload-add-rpath

This revision was automatically updated to reflect the committed changes. Closed by commit rG55b43449978c: [HIP] add --offload-add-rpath (authored by yaxunl). Herald added a project: clang. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D136854/new/ h

[PATCH] D136859: [HIP] add fmax/fmin for fp16

This revision was automatically updated to reflect the committed changes. Closed by commit rG319444fcf586: [HIP] add fmax/fmin for fp16 (authored by yaxunl). Herald added a project: clang. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D136859/new/ ht

[PATCH] D136981: [HIP] add float to fp16 convert functions

yaxunl created this revision. yaxunl added reviewers: tra, b-sumner. Herald added a project: All. yaxunl requested review of this revision. https://reviews.llvm.org/D136981 Files: clang/lib/Headers/__clang_hip_libdevice_declares.h Index: clang/lib/Headers/__clang_hip_libdevice_declares.h

[PATCH] D136981: [HIP] add float to fp16 convert functions

This revision was landed with ongoing or failed builds. This revision was automatically updated to reflect the committed changes. Closed by commit rG36a025366215: [HIP] add float to fp16 convert functions (authored by yaxunl). Herald added a project: clang. Repository: rG LLVM Github Monorepo

[PATCH] D136701: [LinkerWrapper] Perform device linking steps in parallel

yaxunl added inline comments. Comment at: clang/tools/clang-linker-wrapper/ClangLinkerWrapper.cpp:1211-1212 +llvm::sort(Input, [](OffloadingImage &A, OffloadingImage &B) { + return A.StringData["triple"].compare(B.StringData["triple"]) == 1 || + A.StringData[

[PATCH] D132140: [AMDGPU] Add builtin s_sendmsg_rtn_b{32|64}

yaxunl created this revision. yaxunl added reviewers: b-sumner, arsenm, foad, kzhuravl, bcahoon. Herald added subscribers: kosarev, kerbowa, t-tye, tpr, dstuttard, jvesely. Herald added a project: All. yaxunl requested review of this revision. Herald added a subscriber: wdng. https://reviews.llvm.

[PATCH] D132140: [AMDGPU] Add builtin s_sendmsg_rtn_b{32|64}

yaxunl added a comment. In D132140#3732262 , @b-sumner wrote: > Following existing naming, it might make sense to rename "rtn_b32" --> "rtn" > and "rtn_b64" --> "rtnl". will modify. thanks. CHANGES SINCE LAST ACTION https://reviews.llvm.org/D132140/

[PATCH] D132140: [AMDGPU] Add builtin s_sendmsg_rtn_b{32|64}

yaxunl updated this revision to Diff 453681. yaxunl added a comment. revised by Brian's comments CHANGES SINCE LAST ACTION https://reviews.llvm.org/D132140/new/ https://reviews.llvm.org/D132140 Files: clang/include/clang/Basic/BuiltinsAMDGPU.def clang/lib/CodeGen/CGBuiltin.cpp clang/te

[PATCH] D132248: [CUDA][OpenMP] Fix the new driver crashing on multiple device-only outputs

yaxunl added a comment. In D132248#3736295 , @tra wrote: > I'm OK with that. > > @yaxunl -- what are your thoughts on whether this approach would work for > HIP? On one hand HIP already has a lot of features that the new driver is > intended to provide,

[PATCH] D132140: [AMDGPU] Add builtin s_sendmsg_rtn

yaxunl added inline comments. Comment at: clang/test/CodeGenOpenCL/builtins-amdgcn-gfx11.cl:23 + +// Test mismatched argument and return types are handled. + tra wrote: > Is there a particular reason for this test? > > Argument and return value type checks shoul

[PATCH] D132140: [AMDGPU] Add builtin s_sendmsg_rtn

yaxunl updated this revision to Diff 454607. yaxunl added a comment. remove unnecessary tests CHANGES SINCE LAST ACTION https://reviews.llvm.org/D132140/new/ https://reviews.llvm.org/D132140 Files: clang/include/clang/Basic/BuiltinsAMDGPU.def clang/lib/CodeGen/CGBuiltin.cpp clang/test/

[PATCH] D132140: [AMDGPU] Add builtin s_sendmsg_rtn

This revision was automatically updated to reflect the committed changes. Closed by commit rG9f6cb3e9fdb4: [AMDGPU] Add builtin s_sendmsg_rtn (authored by yaxunl). Herald added a project: clang. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D132140/ne

[PATCH] D130096: [Clang][AMDGPU] Emit AMDGPU library control constants in clang

yaxunl added inline comments. Comment at: clang/lib/CodeGen/TargetInfo.cpp:9436 +CGM.getModule(), Type, true, +llvm::GlobalValue::LinkageTypes::LinkOnceODRLinkage, +llvm::ConstantInt::get(Type, Value), Name, nullptr, This does not support

[PATCH] D132248: [CUDA][OpenMP] Fix the new driver crashing on multiple device-only outputs

yaxunl accepted this revision. yaxunl added a comment. This revision is now accepted and ready to land. LGTM. Thanks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D132248/new/ https://reviews.llvm.org/D132248 __

[PATCH] D132607: [OffloadPackager] Add ability to extract mages from other file types

yaxunl added inline comments. Comment at: clang/tools/clang-offload-packager/ClangOffloadPackager.cpp:17-21 +#include "llvm/IR/Constants.h" +#include "llvm/IR/Module.h" +#include "llvm/IRReader/IRReader.h" +#include "llvm/Object/Archive.h" +#include "llvm/Object/ArchiveWriter.h"

[PATCH] D133705: [HIP] Fix unbundling archive

yaxunl added a comment. In D133705#3793702 , @tra wrote: > In D133705#3785470 , @yaxunl wrote: > >>> Also, using `lib*.a` as pattern to tell device libraries from the host-ony >>> one will be insufficient. There

[PATCH] D133705: [HIP] Fix unbundling archive

yaxunl added a comment. In D133705#3793931 , @MaskRay wrote: > I know very little about HIP, but I am concerned with relying on extensions > as well. For example, I've seen `libc++.a.1` (we use this for the real > archive while `libc++.a` is a linker sc

[PATCH] D130096: [Clang][AMDGPU] Emit AMDGPU library control constants in clang

yaxunl added inline comments. Comment at: clang/lib/CodeGen/TargetInfo.cpp:9468-9472 + AddGlobal("__oclc_wavefrontsize64", Wavefront64, /*Size=*/8); + AddGlobal("__oclc_daz_opt", DenormAreZero, /*Size=*/8); + AddGlobal("__oclc_finite_only_opt", FiniteOnly || RelaxedMath, /*Siz

[PATCH] D130096: [Clang][AMDGPU] Emit AMDGPU library control constants in clang

yaxunl added inline comments. Comment at: clang/lib/CodeGen/TargetInfo.cpp:9468 + // Control constants for math operations. + AddGlobal("__oclc_wavefrontsize64", Wavefront64, /*Size=*/8); + AddGlobal("__oclc_daz_opt", DenormAreZero, /*Size=*/8); jhuber6 wrote:

[PATCH] D130096: [Clang][AMDGPU] Emit AMDGPU library control constants in clang

yaxunl added inline comments. Comment at: clang/lib/CodeGen/TargetInfo.cpp:9468 + // Control constants for math operations. + AddGlobal("__oclc_wavefrontsize64", Wavefront64, /*Size=*/8); + AddGlobal("__oclc_daz_opt", DenormAreZero, /*Size=*/8); jhuber6 wrote:

[PATCH] D134355: [AMDGPU] Emit module flag for all code object versions

yaxunl created this revision. yaxunl added reviewers: b-sumner, cfang. Herald added subscribers: kosarev, kerbowa, t-tye, tpr, dstuttard, jvesely, kzhuravl. Herald added a project: All. yaxunl requested review of this revision. Herald added a subscriber: wdng. https://reviews.llvm.org/D134355 Fi

[PATCH] D134314: [HIP] stop forcing the lang std in the driver

yaxunl accepted this revision. yaxunl added a comment. This revision is now accepted and ready to land. LGTM. Thanks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D134314/new/ https://reviews.llvm.org/D134314 __

[PATCH] D134355: [AMDGPU] Emit module flag for all code object versions

yaxunl marked an inline comment as done. yaxunl added inline comments. Comment at: clang/lib/CodeGen/CodeGenModule.cpp:590-591 // times 100. -// ToDo: Enable module flag for all code object version when ROCm device -// library is ready. -if (getTarget().getTarget

[PATCH] D133705: [HIP] Fix unbundling archive

yaxunl updated this revision to Diff 462189. yaxunl added a comment. allow archive files to have unknown extension CHANGES SINCE LAST ACTION https://reviews.llvm.org/D133705/new/ https://reviews.llvm.org/D133705 Files: clang/lib/Driver/Driver.cpp clang/lib/Driver/ToolChains/CommonArgs.cp

[PATCH] D134355: [AMDGPU] Emit module flag for all code object versions

yaxunl marked an inline comment as done. yaxunl added a comment. In D134355#3807435 , @cfang wrote: > LGTM > > Should the module flag name be amdgpu_code_object_version or > amdhsa_code_object_version? Good question. @b-sumner Does code object version

[PATCH] D134355: [AMDGPU] Emit module flag for all code object versions

yaxunl added a comment. In D134355#3809471 , @b-sumner wrote: > In D134355#3809294 , @yaxunl wrote: > >> In D134355#3807435 , @cfang wrote: >> >>> LGTM >>> >>> Should the

[PATCH] D134355: [AMDGPU] Emit module flag for all code object versions

This revision was landed with ongoing or failed builds. This revision was automatically updated to reflect the committed changes. Closed by commit rG5e25284dbc94: [AMDGPU] Emit module flag for all code object versions (authored by yaxunl). Herald added a project: clang. Repository: rG LLVM Gith

[PATCH] D133705: [HIP] Fix unbundling archive

yaxunl marked 4 inline comments as done. yaxunl added inline comments. Comment at: clang/lib/Driver/Driver.cpp:2907-2908 +// which are not object files. Files with extension ".lib" is classified +// as TY_Object but they are actually archives, therefore should no

[PATCH] D133705: [HIP] Fix unbundling archive

yaxunl updated this revision to Diff 462376. yaxunl marked 4 inline comments as done. yaxunl added a comment. revised by Artem's comments CHANGES SINCE LAST ACTION https://reviews.llvm.org/D133705/new/ https://reviews.llvm.org/D133705 Files: clang/lib/Driver/Driver.cpp clang/lib/Driver/T

[PATCH] D134546: [clang-offload-bundler] extracting compatible bundle entry

yaxunl created this revision. yaxunl added reviewers: tra, saiislam, lamb-j. Herald added a project: All. yaxunl requested review of this revision. Herald added subscribers: sstefan1, MaskRay. Herald added a reviewer: jdoerfert. In HIP a library is usually compiled with default target ID e.g. gfx9

[PATCH] D133705: [HIP] Fix unbundling archive

yaxunl updated this revision to Diff 462673. yaxunl added a comment. I just found clang-offload-bundler reports an error when trying to unbundle an archive but the input file is not an archive. This update let clang-offload-bundler to extract empty archives when the input file is not an archive

[PATCH] D133705: [HIP] Fix unbundling archive

yaxunl marked an inline comment as done. yaxunl added inline comments. Comment at: clang/test/Driver/clang-offload-bundler.c:410-412 +// Check clang-offload-bundler extracts empty archives if the input file +// is not an archive when --allow-missing-bundles is specified, otherwis

[PATCH] D133705: [HIP] Fix unbundling archive

yaxunl updated this revision to Diff 463016. yaxunl marked an inline comment as done. yaxunl added a comment. check file magic and only unbundle real archives CHANGES SINCE LAST ACTION https://reviews.llvm.org/D133705/new/ https://reviews.llvm.org/D133705 Files: clang/lib/Driver/Driver.cpp

[PATCH] D130096: [Clang][AMDGPU] Emit AMDGPU library control constants in clang

yaxunl added inline comments. Comment at: clang/lib/CodeGen/CodeGenAction.cpp:299-308 + if (!LinkModules.empty() && Gen->CGM().getTriple().isAMDGCN() && + !Gen->CGM().getLangOpts().GPURelocatableDeviceCode) { +const StringRef GVS[] = {"__oclc_daz_opt", "__oc

[PATCH] D133705: [HIP] Fix unbundling archive

yaxunl closed this revision. yaxunl added a comment. committed by 1172bdecfab364579d90e6aa5ba7fc64a5b96786 CHANGES SINCE LAST ACTION https://reviews.llvm.org/D133705/new/ https://reviews.llvm.org/D133705 _

[PATCH] D134872: AMDGPU: Add __builtin_amdgcn_permlane64

yaxunl accepted this revision as: yaxunl. yaxunl added a comment. This revision is now accepted and ready to land. LGTM. Thanks. CHANGES SINCE LAST ACTION https://reviews.llvm.org/D134872/new/ https://reviews.llvm.org/D134872 ___ cfe-commits mailin

[PATCH] D88976: [clang] Use correct address space for global variable debug info

yaxunl accepted this revision. yaxunl added a comment. LGTM. Thanks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D88976/new/ https://reviews.llvm.org/D88976 ___ cfe-commits mailing list cfe-commits@list

[PATCH] D130096: [Clang][AMDGPU] Emit AMDGPU library control constants in clang

yaxunl added inline comments. Comment at: clang/test/CodeGen/amdgcn-link-control-constants.c:2-3 +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --check-globals --include-generated-funcs --global-value-regex "__oclc_

[PATCH] D130096: [Clang][AMDGPU] Emit AMDGPU library control constants in clang

yaxunl added inline comments. Comment at: clang/test/CodeGen/amdgcn-link-control-constants.c:2-3 +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --check-globals --include-generated-funcs --global-value-regex "__oclc_

[PATCH] D130096: [Clang][AMDGPU] Emit AMDGPU library control constants in clang

yaxunl added inline comments. Comment at: clang/test/CodeGen/amdgcn-control-constants.c:8 + +// GFX90A: @__oclc_daz_opt = linkonce_odr hidden local_unnamed_addr addrspace(4) constant i8 0, align 1 +// GFX90A: @__oclc_wavefrontsize64 = linkonce_odr hidden local_unnamed_addr addr

[PATCH] D130096: [Clang][AMDGPU] Emit AMDGPU library control constants in clang

yaxunl added inline comments. Comment at: clang/test/CodeGen/amdgcn-control-constants.c:8 + +// GFX90A: @__oclc_daz_opt = linkonce_odr hidden local_unnamed_addr addrspace(4) constant i8 0, align 1 +// GFX90A: @__oclc_wavefrontsize64 = linkonce_odr hidden local_unnamed_addr addr

[PATCH] D134546: [clang-offload-bundler] extracting compatible bundle entry

yaxunl marked 3 inline comments as done. yaxunl added inline comments. Comment at: clang/lib/Driver/OffloadBundler.cpp:1008 +auto Output = Worklist.begin(); +for (auto E = Worklist.end(); Output != E; Output++) { + if (isCodeObjectCompatible( tra wro

[PATCH] D134546: [clang-offload-bundler] extracting compatible bundle entry

yaxunl updated this revision to Diff 465265. yaxunl marked an inline comment as done. yaxunl added a comment. check bundle entry ID compatibility when bundling CHANGES SINCE LAST ACTION https://reviews.llvm.org/D134546/new/ https://reviews.llvm.org/D134546 Files: clang/include/clang/Basic/

[PATCH] D135269: [AMDGPU] Disable bool range metadata to workaround backend issue

yaxunl created this revision. yaxunl added reviewers: tra, ronl. Herald added subscribers: kosarev, t-tye, tpr, dstuttard, kzhuravl. Herald added a project: All. yaxunl requested review of this revision. Herald added a subscriber: wdng. Currently there is some backend issue which causes values loa

[PATCH] D135269: [AMDGPU] Disable bool range metadata to workaround backend issue

yaxunl updated this revision to Diff 465403. yaxunl added a comment. fix comments in test CHANGES SINCE LAST ACTION https://reviews.llvm.org/D135269/new/ https://reviews.llvm.org/D135269 Files: clang/lib/CodeGen/CGExpr.cpp clang/test/CodeGenCUDA/bool-range.cu Index: clang/test/CodeGenC

[PATCH] D135305: [Clang] Fix using LTO with the new driver in RDC-mode

yaxunl added a comment. should we test with -ccc-print-phases instead? It is not clear what actions are produced by driver. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D135305/new/ https://reviews.llvm.org/D135305 ___

[PATCH] D135305: [Clang] Fix using LTO with the new driver in RDC-mode

yaxunl added a comment. In D135305#3838435 , @jhuber6 wrote: > In D135305#3838412 , @yaxunl wrote: > >> should we test with -ccc-print-phases instead? It is not clear what actions >> are produced by driver. > > A

[PATCH] D135269: [AMDGPU] Disable bool range metadata to workaround backend issue

yaxunl added a comment. In D135269#3837394 , @tra wrote: > Is there more info about the issue? What does AMDGPU currently emit for the > test case? > > AFAICT from running it on CE (https://godbolt.org/z/ccq3vnbrM) llvm optimizes > it to essentially `*y

[PATCH] D134546: [clang-offload-bundler] extracting compatible bundle entry

This revision was landed with ongoing or failed builds. This revision was automatically updated to reflect the committed changes. Closed by commit rG844b84af20c7: [clang-offload-bundler] extracting compatible bundle entry (authored by yaxunl). Herald added a project: clang. Repository: rG LLVM

[PATCH] D135328: [CUDA] Refactored CUDA version housekeeping to use less boilerplate.

yaxunl added inline comments. Comment at: clang/lib/Basic/Cuda.cpp:59 +CudaVersion ToCudaVersion(llvm::VersionTuple Version) { + int IVer = Version.getMajor() * 10 + Version.getMinor().value_or(0); + for (auto *I = CudaNameVersionMap; I->Version != CudaVersion::UNKNOWN; ++I) --

[PATCH] D135305: [Clang] Fix using LTO with the new driver in RDC-mode

yaxunl added inline comments. Comment at: clang/test/Driver/hip-phases.hip:553 +// +// RUN: %clang -### --target=x86_64-linux-gnu --offload-new-driver -ccc-print-phases \ +// RUN:--offload-arch=gfx90a --offload-arch=gfx908 -foffload-lto -fgpu-rdc -c %s 2>&1 \ --

[PATCH] D135306: [CUDA] Add support for CUDA-11.8 and sm_{87,89,90} GPUs.

yaxunl added inline comments. Comment at: clang/include/clang/Basic/BuiltinsNVPTX.def:30-32 +#define SM_89 "sm_87|" SM_90 +#define SM_87 "sm_89|" SM_89 +#define SM_86 "sm_89|" SM_87 typo? Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https:

[PATCH] D135269: [AMDGPU] Disable bool range metadata to workaround backend issue

yaxunl marked an inline comment as done. yaxunl added inline comments. Comment at: clang/lib/CodeGen/CGExpr.cpp:1792 // attach range metadata to the load. - } else if (CGM.getCodeGenOpts().OptimizationLevel > 0) +// TODO: Enable range metadata for AMDGCN after backend i

[PATCH] D135305: [Clang] Fix using LTO with the new driver in RDC-mode

yaxunl accepted this revision. yaxunl added a comment. This revision is now accepted and ready to land. LGTM. Thanks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D135305/new/ https://reviews.llvm.org/D135305 __

[PATCH] D135306: [CUDA] Add support for CUDA-11.8 and sm_{87,89,90} GPUs.

yaxunl accepted this revision. yaxunl added a comment. This revision is now accepted and ready to land. LGTM. Thanks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D135306/new/ https://reviews.llvm.org/D135306 __

[PATCH] D135269: [AMDGPU] Disable bool range metadata to workaround backend issue

yaxunl updated this revision to Diff 465731. yaxunl marked an inline comment as done. yaxunl edited the summary of this revision. yaxunl added a comment. update comments with issue link CHANGES SINCE LAST ACTION https://reviews.llvm.org/D135269/new/ https://reviews.llvm.org/D135269 Files:

[PATCH] D135269: [AMDGPU] Disable bool range metadata to workaround backend issue

This revision was automatically updated to reflect the committed changes. Closed by commit rG107ee2613063: [AMDGPU] Disable bool range metadata to workaround backend issue (authored by yaxunl). Herald added a project: clang. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https

[PATCH] D135328: [CUDA] Refactored CUDA version housekeeping to use less boilerplate.

yaxunl accepted this revision. yaxunl added a comment. This revision is now accepted and ready to land. LGTM. Thanks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D135328/new/ https://reviews.llvm.org/D135328 __

[PATCH] D135614: [OpenMP][CUDA][AMDGPU] Accept case insensitive subarchitecture names

yaxunl added a comment. I am not sure whether it is a good idea to allow gfx90A in `--offload-arch`, since it is not documented in LLVM AMDGPU usage. @b-sumner @arsenm Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D135614/new/ https://reviews.llvm.

[PATCH] D135724: [HIP] Fix unbundling archive

yaxunl created this revision. yaxunl added reviewers: tra, saiislam. Herald added a project: All. yaxunl requested review of this revision. Herald added a subscriber: MaskRay. When `-lxxx` is specified, if there happens to have a directory or file with name `xxx`, clang will not look up `libxxx.a`

[PATCH] D135724: [HIP] Fix unbundling archive

yaxunl added inline comments. Comment at: clang/lib/Driver/ToolChains/CommonArgs.cpp:1839 auto Ext = IsMSVC ? ".lib" : ".a"; - if (!Lib.startswith(":") && llvm::sys::fs::exists(Lib)) { -ArchiveOfBundles = Lib; -FoundAOB = true; + if (!Lib.startswith(":") && !Lib.star

[PATCH] D135796: [HIP] Detect HIP for Debian/Fedora

yaxunl created this revision. yaxunl added a reviewer: tra. Herald added subscribers: kosarev, kerbowa, jvesely. Herald added a project: All. yaxunl requested review of this revision. Herald added a subscriber: MaskRay. HIP is installed at /usr or /usr/local on Debin/Fedora, and the version file i

[PATCH] D135796: [HIP] Detect HIP for Debian/Fedora

yaxunl added inline comments. Comment at: clang/lib/Driver/ToolChains/AMDGPU.cpp:309 + ROCmSearchDirs.emplace_back(D.SysRoot + "/usr/local", + /*StrictChecking=*/true); tra wrote: > Should it be done for Debian/Fedora only? See >

[PATCH] D135724: [HIP] Fix unbundling archive

This revision was automatically updated to reflect the committed changes. Closed by commit rG12c6a41f5249: [HIP] Fix unbundling archive (authored by yaxunl). Herald added a project: clang. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D135724/new/ ht

[PATCH] D135796: [HIP] Detect HIP for Debian/Fedora

yaxunl marked an inline comment as done. yaxunl added inline comments. Comment at: clang/lib/Driver/ToolChains/AMDGPU.cpp:309 + ROCmSearchDirs.emplace_back(D.SysRoot + "/usr/local", + /*StrictChecking=*/true); tra wrote: > yaxunl w

[PATCH] D135796: [HIP] Detect HIP for Debian/Fedora

yaxunl updated this revision to Diff 467282. yaxunl marked an inline comment as done. yaxunl added a comment. only check /usr and/usr/local for debian and redhat CHANGES SINCE LAST ACTION https://reviews.llvm.org/D135796/new/ https://reviews.llvm.org/D135796 Files: clang/lib/Driver/ToolCha

[PATCH] D135796: [HIP] Detect HIP for Debian/Fedora

This revision was automatically updated to reflect the committed changes. Closed by commit rG082593ff7aff: [HIP] Detect HIP for Debian/Fedora (authored by yaxunl). Herald added a project: clang. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D135796/ne

[PATCH] D136036: [Clang] Add __has_constexpr_builtin support

yaxunl added a comment. need some Sema tests to verify a constexpr builtin is const evaluated. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D136036/new/ https://reviews.llvm.org/D136036 ___ cfe-commits m

[PATCH] D135832: Do not append terminating NUL to the string with embedded GPU binary.

yaxunl accepted this revision. yaxunl added a comment. This revision is now accepted and ready to land. LGTM. Thanks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D135832/new/ https://reviews.llvm.org/D135832 __

[PATCH] D140315: [AMDGCN] Update search path for device libraries

yaxunl added a comment. files under clang/test/Driver/Inputs/rocm_resource_dir/lib/amdgcn/bitcode-no-abi-ver seem not used. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D140315/new/ https://reviews.llvm.org/D140315 __

[PATCH] D138392: clang/HIP: Fix broken implementations of __make_mantissa* functions

yaxunl accepted this revision. yaxunl added a comment. This revision is now accepted and ready to land. LGTM. Thanks. It seems gcc assumes the argument to nan is nonnull (https://godbolt.org/z/xzb8T6Gon), so we can assume that too. CHANGES SINCE LAST ACTION https://reviews.llvm.org/D138392/n

[PATCH] D141008: [Clang][SPIR-V] Emit target extension types for OpenCL types on SPIR-V.

yaxunl added inline comments. Comment at: clang/test/CodeGenOpenCL/cl20-device-side-enqueue.cl:6 // RUN: %clang_cc1 -no-opaque-pointers -no-enable-noundef-analysis %s -cl-std=CL3.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --c

[PATCH] D141051: [CUDA][HIP] Add support for `--offload-arch=native` to CUDA and refactor

yaxunl added inline comments. Comment at: clang/test/Driver/amdgpu-hip-system-arch.c:24 + +// case when amdgpu_arch does not return anything with successful execution +// RUN: %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-arch=native --amdgpu-arch-tool=%t/a

[PATCH] D140663: CUDA/HIP: Use kernel name to map to symbol

yaxunl added inline comments. Comment at: clang/test/CodeGenCUDA/incomplete-func-ptr-type.cu:2 +// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -x hip %s -o - \ +// RUN: | FileCheck %s + need to check `_Z19__device_stub__kern7TempValIjE` generates

[PATCH] D140315: [AMDGCN] Update search path for device libraries

yaxunl accepted this revision. yaxunl added a comment. This revision is now accepted and ready to land. LGTM. Thanks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D140315/new/ https://reviews.llvm.org/D140315 __

[PATCH] D141051: [CUDA][HIP] Add support for `--offload-arch=native` to CUDA and refactor

yaxunl accepted this revision. yaxunl added a comment. This revision is now accepted and ready to land. LGTM. Thanks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D141051/new/ https://reviews.llvm.org/D141051 __

[PATCH] D141437: [HIP] Use .hipi as preprocessor output extension

yaxunl created this revision. yaxunl added a reviewer: tra. Herald added a project: All. yaxunl requested review of this revision. Herald added a subscriber: MaskRay. so that clang can recognize it and handle it automatically without -x hip-cpp-output https://reviews.llvm.org/D141437 Files:

[PATCH] D141449: clang/OpenCL: Fix not setting convergent on block invoke kernels

yaxunl added inline comments. Comment at: clang/lib/CodeGen/TargetInfo.cpp:12428 F->addFnAttr(llvm::Attribute::NoUnwind); + F->addFnAttr(llvm::Attribute::Convergent); How about using CGF.CGM.addDefaultFunctionDefinitionAttributes? same as below CHANGES SI

[PATCH] D141447: clang/OpenCL: Don't use a Function for the block type

yaxunl added a comment. need a test CHANGES SINCE LAST ACTION https://reviews.llvm.org/D141447/new/ https://reviews.llvm.org/D141447 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D141437: [HIP] Use .hipi as preprocessor output extension

yaxunl marked an inline comment as done. yaxunl added inline comments. Comment at: clang/tools/clang-offload-bundler/ClangOffloadBundler.cpp:105 + " cui - cuda-cpp-output\n" + " hipi - hip-cpp-outpu\n" + " d- dependency\n" ---

[PATCH] D137251: [clang][cuda/hip] Allow `__noinline__` lambdas

yaxunl added a comment. need a CodeGenCUDA test Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D137251/new/ https://reviews.llvm.org/D137251 ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://

[PATCH] D137275: [Driver][test] Fix test by creating empty archive instead of empty file

yaxunl accepted this revision. yaxunl added a comment. This revision is now accepted and ready to land. LGTM. Thanks. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D137275/new/ https://reviews.llvm.org/D137275 __

[PATCH] D137154: Adding nvvm_reflect clang builtin

yaxunl added a comment. In D137154#3904810 , @hdelan wrote: > In DPC++ for CUDA we use libclc as a wrapper around CUDA SDK's libdevice. > Like libdevice we want to precompile libclc to bc for the CUDA backend > without specializing for a particular arch

[PATCH] D137251: [clang][cuda/hip] Allow `__noinline__` lambdas

yaxunl added inline comments. Comment at: clang/lib/Parse/ParseExprCXX.cpp:1300 +ParseGNUAttributes(Attr, nullptr, &D); + } else if (Tok.is(tok::kw___noinline__)) { +IdentifierInfo *AttrName = Tok.getIdentifierInfo(); Pierre-vh wrote: > aaron

[PATCH] D137251: [clang][cuda/hip] Allow `__noinline__` lambdas

yaxunl added a comment. In D137251#3904618 , @Pierre-vh wrote: > Comments > > Not sure if the release note is in the right place though. > As for the test, I did something quite targeted/minimal, hope it's fine? LGTM. Thanks. Repository: rG LLVM Gith

<    13   14   15   16   17   18   19   20   21   22   >