[PATCH] D89372: [OpenCL] Remove unused extensions

jvesely added a comment. In D89372#2335027 , @Anastasia wrote: > In D89372#2334728 , @jvesely wrote: > >> In D89372#2334225 , @Anastasia >> wrote: >> >>> >>> Ok, so th

[PATCH] D89372: [OpenCL] Remove unused extensions

jvesely added a comment. In D89372#2334225 , @Anastasia wrote: >> > > Ok, so the pragma is supposed to control the pointer dereferencing which > seems like a valid case but however, it is not implemented now. Do we know of > any immediate plans from any

[PATCH] D89372: [OpenCL] Remove unused extensions

jvesely added a comment. In D89372#2332997 , @Anastasia wrote: > In D89372#2332853 , @jvesely wrote: > >> `cl_khr_byte_addressable_stores` changes language semantics. without it, >> pointer dereferences of types sm

[PATCH] D89372: [OpenCL] Remove unused extensions

jvesely added a comment. `cl_khr_byte_addressable_stores` changes language semantics. without it, pointer dereferences of types smaller than 32 bits are illegal. Even if all clang targets support this the macro should still be defined for backward compatibility. it would be useful if the commit

[PATCH] D78979: OpenCL: Include builtin header by default

jvesely added a comment. In D78979#2006901 , @yaxunl wrote: > In D78979#2006847 , @arsenm wrote: > > > I'm also wondering if using -nogpulib for the corresponding linker purpose > > was correct, since in the OpenCL

[PATCH] D77923: OpenCL: Fix some missing predefined macros

jvesely added a comment. In D77923#1983787 , @yaxunl wrote: > LGTM. The objective of the change is to simplify offline compilation. We want > to avoid asking users to add -D if a proper macro can be inferred from triple > and cpu. Ideally, users only nee

[PATCH] D77923: OpenCL: Fix some missing predefined macros

jvesely added a comment. In D77923#1981398 , @arsenm wrote: > In D77923#1979286 , @jvesely wrote: > > > > In D77923#1976497 , @jvesely > > > wrote: > > > > > >> OPENCL_VERS

[PATCH] D77923: OpenCL: Fix some missing predefined macros

jvesely added a comment. > In D77923#1976497 , @jvesely wrote: > >> OPENCL_VERSION is something that should be really set by the opencl driver >> rather than the compiler. >> coarse-grained SVM can be done without FEATURE_FLAT_ADDRESS_SPACE, so those >>

[PATCH] D77923: OpenCL: Fix some missing predefined macros

jvesely added a comment. OPENCL_VERSION sounds like something that should be really set by the opencl driver rather than the compiler. coarse-grained SVM can be done without FEATURE_FLAT_ADDRESS_SPACE, so those gpus can get over 1.2, while the compiler can be used in an implementation that does

[PATCH] D66068: cmake: Make building clang-shlib optional

jvesely added a comment. Hi, sorry for the delay. I fully understand the need to reduce the number of options. Having always used SHARED_LIBS build I remember weekly shared build breakages. That said, forcing everyone to build one huge library effectively makes debug builds unusable in any pra

[PATCH] D66068: cmake: Make building clang-shlib optional

jvesely added a comment. In D66068#1627706 , @beanz wrote: > I want to dissect this a bit. > > In D66068#1627451 , @E5ten wrote: > > > I am in favour of adding a user-facing option to disable generating this > > du

[PATCH] D66068: cmake: Make building clang-shlib optional

jvesely added a comment. In D66068#1625995 , @beanz wrote: > I think you and I disagree here. General developer workflows don't need to > include building `all` for every minor change. In my normal workflow I just > re-run `check-llvm` or `check-clang` o

[PATCH] D66068: cmake: Make building clang-shlib optional

jvesely added a comment. In D66068#1625478 , @beanz wrote: > I generally am not a fan of adding more and more options. As long as you're > not linking the library it won't be generated by any of the check targets. > With the llvm dylib we've had many iss

[PATCH] D66068: cmake: Make building clang-shlib optional

jvesely created this revision. jvesely added a reviewer: beanz. Herald added a subscriber: mgorny. Herald added a project: clang. It takes ~5min to link, add an option to avoid that. Repository: rC Clang https://reviews.llvm.org/D66068 Files: CMakeLists.txt tools/CMakeLists.txt Index:

[PATCH] D49650: Targets/AMDGPU: Don't set fp32-denormals feature for r600

jvesely abandoned this revision. jvesely added a comment. In https://reviews.llvm.org/D49650#1177323, @jvesely wrote: > In https://reviews.llvm.org/D49650#1176336, @arsenm wrote: > > > In https://reviews.llvm.org/D49650#1175461, @jvesely wrote: > > > > > In https://reviews.llvm.org/D49650#1175438

[PATCH] D49650: Targets/AMDGPU: Don't set fp32-denormals feature for r600

jvesely added a comment. In https://reviews.llvm.org/D49650#1176336, @arsenm wrote: > In https://reviews.llvm.org/D49650#1175461, @jvesely wrote: > > > In https://reviews.llvm.org/D49650#1175438, @arsenm wrote: > > > > > According to cayman manual, these registers do exist so we should > > > pro

[PATCH] D49650: Targets/AMDGPU: Don't set fp32-denormals feature for r600

jvesely added a comment. In https://reviews.llvm.org/D49650#1175438, @arsenm wrote: > According to cayman manual, these registers do exist so we should probably > just make the feature accepted on r600 as well sure, that's the way it was before r335942. I assumed the removal was intentional.

[PATCH] D49650: Targets/AMDGPU: Don't set fp32-denormals feature for r600

jvesely created this revision. jvesely added reviewers: arsenm, tstellar. Herald added subscribers: t-tye, tpr, dstuttard, yaxunl, nhaehnle, wdng, kzhuravl. This feature was removed in r335942 Repository: rC Clang https://reviews.llvm.org/D49650 Files: lib/Basic/Targets/AMDGPU.cpp Index

[PATCH] D38667: AMDGPU: Parse r600 CPU name early and expose FMAF capability

This revision was automatically updated to reflect the committed changes. Closed by commit rL316181: AMDGPU: Parse r600 CPU name early and expose FMAF capability (authored by jvesely). Changed prior to commit: https://reviews.llvm.org/D38667?vs=118136&id=119614#toc Repository: rL LLVM https

[PATCH] D38667: AMDGPU: Parse r600 CPU name early and expose FMAF capability

jvesely created this revision. Herald added subscribers: t-tye, tpr, dstuttard, nhaehnle, wdng, kzhuravl. Improve amdgcn macro test Repository: rL LLVM https://reviews.llvm.org/D38667 Files: lib/Basic/Targets/AMDGPU.cpp test/Preprocessor/predefined-arch-macros.c Index: test/Preprocesso

[PATCH] D37231: Add half load and store builtins

This revision was automatically updated to reflect the committed changes. Closed by commit rL312742: [OpenCL] Add half load and store builtins (authored by jvesely). Changed prior to commit: https://reviews.llvm.org/D37231?vs=113624&id=114240#toc Repository: rL LLVM https://reviews.llvm.org

[PATCH] D37231: Add half load and store builtins

jvesely marked 2 inline comments as done. jvesely added inline comments. Comment at: test/CodeGenOpenCL/no-half.cl:27 + foo[0] = __builtin_load_halff(bar); +// CHECK: [[HALF_VAL:%.*]] = load +// CHECK: [[FULL_VAL:%.*]] = fpext half [[HALF_VAL]] to float Ana

[PATCH] D37231: Add half load and store builtins

jvesely requested review of this revision. jvesely added a comment. please let me know if your accept still stands for the modified version. Repository: rL LLVM https://reviews.llvm.org/D37231 ___ cfe-commits mailing list cfe-commits@lists.llvm.o

[PATCH] D37231: Add half load and store builtins

jvesely updated this revision to Diff 113624. jvesely added a comment. mark load pointers const Repository: rL LLVM https://reviews.llvm.org/D37231 Files: include/clang/Basic/Builtins.def include/clang/Basic/Builtins.h lib/Basic/Builtins.cpp lib/CodeGen/CGBuiltin.cpp test/CodeGenOp

[PATCH] D37231: Add half load and store builtins

jvesely added inline comments. Comment at: include/clang/Basic/Builtins.def:1427 +// OpenCL half load/store builtin +BUILTIN(__builtin_store_half, "vdh*", "n") +BUILTIN(__builtin_store_halff, "vfh*", "n") Anastasia wrote: > jvesely wrote: > > Anastasia wrote: > >

[PATCH] D37231: Add half load and store builtins

jvesely updated this revision to Diff 113588. jvesely marked 6 inline comments as done. jvesely edited the summary of this revision. jvesely added a comment. fully check loads in tests Repository: rL LLVM https://reviews.llvm.org/D37231 Files: include/clang/Basic/Builtins.def include/cla

[PATCH] D37231: Add half load and store builtins

jvesely updated this revision to Diff 113190. jvesely added a comment. restrict builtins to OCLC langauges Repository: rL LLVM https://reviews.llvm.org/D37231 Files: include/clang/Basic/Builtins.def include/clang/Basic/Builtins.h lib/Basic/Builtins.cpp lib/CodeGen/CGBuiltin.cpp tes

[PATCH] D37231: Add half load and store builtins

jvesely added inline comments. Comment at: include/clang/Basic/Builtins.def:1427 +// OpenCL half load/store builtin +BUILTIN(__builtin_store_half, "vdh*", "n") +BUILTIN(__builtin_store_halff, "vfh*", "n") Anastasia wrote: > I think this should be a language built