Page MenuHomePhabricator

rivanvx (Vedran Miletić)
Doctor, but not that kind of doctor

Projects

User does not belong to any projects.

User Details

User Since
Jan 12 2016, 6:13 AM (255 w, 13 h)

Recent Activity

May 6 2018

rivanvx accepted D46382: [OpenCL] Factor out language version printing.

Nice cleanup.

May 6 2018, 12:51 PM

Oct 8 2017

rivanvx updated the diff for D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.

This should be it.

Oct 8 2017, 5:21 AM

Sep 6 2017

rivanvx added a comment to D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.

looks ok to me. hope you were not waiting for me this whole time.

Sep 6 2017, 2:44 AM

Sep 5 2017

rivanvx updated the diff for D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.

@jvesely I would like to get an ACK before I tackle the tests.

Sep 5 2017, 11:06 AM

Apr 19 2017

rivanvx updated the diff for D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.

That was only a partial diff, should be correct now.

Apr 19 2017, 4:56 AM
rivanvx updated the diff for D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.

This should be it.

Apr 19 2017, 4:54 AM

Apr 18 2017

rivanvx updated the diff for D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.

Sorry for the delay, updated patch, now passes tests, addressed all the comments. I would appreciate if you could check if I got all the R_ and A_ prefixes correctly (it's just variable naming, but there is some logic behind it -- register vs address, I presume?) in f16 tests.

Apr 18 2017, 10:20 AM

Mar 9 2017

rivanvx updated the diff for D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.

TODO: TESTS NEED TO BE UPDATED

Mar 9 2017, 10:28 AM

Feb 22 2017

rivanvx added a comment to D30236: [OpenCL] r600 needs OpenCL kernel calling convention.

@arsenm Can you eventually push this into 4.0 branch as well?

Feb 22 2017, 3:08 AM
rivanvx accepted D30236: [OpenCL] r600 needs OpenCL kernel calling convention.

LGTM

Feb 22 2017, 3:07 AM

Feb 18 2017

rivanvx updated the diff for D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.
Feb 18 2017, 12:04 PM

Feb 15 2017

rivanvx retitled D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics from Add custom lowering for llvm.log.f32 intrinsic to Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.
Feb 15 2017, 9:26 AM
rivanvx updated the diff for D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.

Better version: handle log10 as well, define constants locally, handle f16, do not pretend to handle f64. Tests incoming.

Feb 15 2017, 9:25 AM

Feb 14 2017

rivanvx updated the summary of D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.
Feb 14 2017, 7:43 AM
rivanvx created D29942: Add custom lowering for llvm.log{,10}.{f16,f32} intrinsics.
Feb 14 2017, 7:42 AM

Feb 9 2017

rivanvx accepted D27283: Fix invalid addrspacecast due to combining alloca with global var.

To the extent of my knowledge and testing, this looks good.

Feb 9 2017, 11:22 AM
rivanvx added a comment to D27283: Fix invalid addrspacecast due to combining alloca with global var.

Whoa, this seems to fix the remaining part of https://bugs.freedesktop.org/show_bug.cgi?id=99552
Reviewing.

Feb 9 2017, 7:34 AM

Jan 25 2017

rivanvx created D29147: [master + 4.0 branch] Enable FeatureFlatForGlobal on Volcanic Islands.
Jan 25 2017, 1:50 PM · Restricted Project

Jan 22 2017

rivanvx updated the diff for D28809: [RFC, 4.0 branch] Enable FeatureFlatForGlobal on Volcanic Islands.

Updated per @arsenm's comment.

Jan 22 2017, 12:15 PM · Restricted Project

Jan 20 2017

rivanvx updated the diff for D28809: [RFC, 4.0 branch] Enable FeatureFlatForGlobal on Volcanic Islands.

Addressed all the issues mentioned. Please review.

Jan 20 2017, 11:46 AM · Restricted Project
rivanvx added inline comments to D28809: [RFC, 4.0 branch] Enable FeatureFlatForGlobal on Volcanic Islands.
Jan 20 2017, 5:00 AM · Restricted Project

Jan 18 2017

rivanvx updated the diff for D28809: [RFC, 4.0 branch] Enable FeatureFlatForGlobal on Volcanic Islands.

Better approach. I have hardcode the processor names in AMDGPUSubtarget::initializeSubtargetDependencies() because getGeneration() will return SI before ParseSubtargetFeatures() finishes, and I have to provide Feature String for it to execute.

Jan 18 2017, 4:16 PM · Restricted Project

Jan 17 2017

rivanvx updated the summary of D28809: [RFC, 4.0 branch] Enable FeatureFlatForGlobal on Volcanic Islands.
Jan 17 2017, 8:39 AM · Restricted Project
rivanvx created D28809: [RFC, 4.0 branch] Enable FeatureFlatForGlobal on Volcanic Islands.
Jan 17 2017, 8:38 AM · Restricted Project

Dec 22 2016

rivanvx added a comment to D27844: Fixed '!NodePtr->isKnownSentinel()' assert caused by dereferencing end iterator when trying to const cast the iterator..

You just wait until a maintainer picks it up.

Dec 22 2016, 5:19 AM · Restricted Project

Dec 16 2016

rivanvx accepted D27844: Fixed '!NodePtr->isKnownSentinel()' assert caused by dereferencing end iterator when trying to const cast the iterator..

LGTM as well

Dec 16 2016, 8:34 AM · Restricted Project

Dec 12 2016

rivanvx accepted D18730: Fix compile with GCC 6.

Fine with me.

Dec 12 2016, 6:57 AM

Jul 21 2016

rivanvx added a comment to D20168: [CodeGen] Handle structs directly in AMDGPUABIInfo.

Addressed both concerns.

Jul 21 2016, 10:15 AM
rivanvx updated the diff for D20168: [CodeGen] Handle structs directly in AMDGPUABIInfo.
Jul 21 2016, 10:15 AM

Jul 18 2016

rivanvx updated the diff for D20168: [CodeGen] Handle structs directly in AMDGPUABIInfo.

Specifically handle only kernels.

Jul 18 2016, 4:33 PM

May 29 2016

rivanvx updated the diff for D20168: [CodeGen] Handle structs directly in AMDGPUABIInfo.

Updated patch. Single element structs are coerced to its element, and there are tests for structs of different sizes, structs of arrays, structs containing structs. Arrays of structs are disallowed by clang in kernels.
Non-kernel functions are not specifically handled, should they be? How to decide?

May 29 2016, 3:04 PM

May 27 2016

rivanvx added a comment to D19780: Output OpenCL version in Clang diagnostics.

@Anastasia I looked into introducing a separate getOpenCLVersion() function (or perhaps three - major version, minor version and version string). This would have to be used in lib/CodeGen/TargetInfo.cpp and lib/Parse/ParseDecl.cpp, and I am undecided on where should one put this code. One option would be in Parse/Parser.h inside class Parser, and then TargetInfo.cpp would have to include Parser.h, unless we decide to declare it inside AST/ASTContext.h.

May 27 2016, 8:41 AM

May 14 2016

rivanvx added a comment to D19833: [scan-build] fix dead store warnings emitted on LLVM AMDGPU code base.

@apelete neither do I. @tstellarAMD, can you please commit this?

May 14 2016, 2:11 PM

May 12 2016

rivanvx updated the diff for D20168: [CodeGen] Handle structs directly in AMDGPUABIInfo.

Now with 100% more tests.

May 12 2016, 5:42 AM

May 11 2016

rivanvx added a comment to D19780: Output OpenCL version in Clang diagnostics.

Could we solve that at a later point? There is one more place where such code is already used, but this would enlarge the scope of this patch.

May 11 2016, 12:30 PM
rivanvx retitled D20168: [CodeGen] Handle structs directly in AMDGPUABIInfo from to [CodeGen] Handle structs directly in AMDGPUABIInfo.
May 11 2016, 6:47 AM
rivanvx added a comment to D19780: Output OpenCL version in Clang diagnostics.

Please, can anyone push this?

May 11 2016, 6:35 AM

May 9 2016

rivanvx added a comment to D19780: Output OpenCL version in Clang diagnostics.

Thanks for the reviews!

May 9 2016, 1:00 PM
rivanvx removed a reviewer for D19780: Output OpenCL version in Clang diagnostics: cfe-commits.
May 9 2016, 12:59 PM

May 6 2016

rivanvx updated the diff for D19780: Output OpenCL version in Clang diagnostics.

Make that int const as well.

May 6 2016, 4:17 PM
rivanvx updated the diff for D19780: Output OpenCL version in Clang diagnostics.

I am neither aware how to convert ints to StringRef nor how to concatenate StringRefs. Apologies if I missed something in the API.

May 6 2016, 3:55 PM
rivanvx added a comment to D19780: Output OpenCL version in Clang diagnostics.

@Anastasia would you still prefer to make VerSpec a std::string?

May 6 2016, 9:35 AM
rivanvx added inline comments to D19780: Output OpenCL version in Clang diagnostics.
May 6 2016, 8:01 AM
rivanvx added a reviewer for D19780: Output OpenCL version in Clang diagnostics: cfe-commits.
May 6 2016, 6:23 AM
rivanvx updated the diff for D19780: Output OpenCL version in Clang diagnostics.

Replace Twine with std::string and use llvm::to_string().

May 6 2016, 6:23 AM

May 3 2016

rivanvx accepted D19833: [scan-build] fix dead store warnings emitted on LLVM AMDGPU code base.

LGTM

May 3 2016, 2:22 PM
rivanvx accepted D19837: AMDGPU: Simplify control flow / conditions.

LGTM

May 3 2016, 2:15 PM
rivanvx added a reviewer for D19780: Output OpenCL version in Clang diagnostics: pxli168.
May 3 2016, 2:12 PM

May 1 2016

rivanvx updated the diff for D19780: Output OpenCL version in Clang diagnostics.

Fix formatting.

May 1 2016, 10:55 PM
rivanvx updated the diff for D19780: Output OpenCL version in Clang diagnostics.

Fixed wrong search and replace in tests.

May 1 2016, 4:08 PM
rivanvx retitled D19780: Output OpenCL version in Clang diagnostics from to Output OpenCL version in Clang diagnostics.
May 1 2016, 3:42 PM

Apr 27 2016

rivanvx abandoned D18730: Fix compile with GCC 6.

GCC 6.1 (GCC 6 final release in GCC's funky release numbering scheme) does not have this issue anymore: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70528

Apr 27 2016, 1:17 PM

Apr 4 2016

rivanvx requested review of D17046: AMDGPU: Document address space mapping.
Apr 4 2016, 8:08 AM

Apr 2 2016

rivanvx added a comment to D18730: Fix compile with GCC 6.

Is GCC 6 being stricter than older versions? Is this behavior standard-compliant?

Apr 2 2016, 3:28 PM
rivanvx added a comment to D18730: Fix compile with GCC 6.

I'm using

Apr 2 2016, 2:09 PM
rivanvx retitled D18730: Fix compile with GCC 6 from to Fix compile with GCC 6.
Apr 2 2016, 2:03 PM

Mar 31 2016

rivanvx accepted D17280: AMDGPU: Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2}.

LGTM and 32-bit version of the instruction tested on Tonga, works.

Mar 31 2016, 6:08 PM

Mar 29 2016

rivanvx updated the diff for D17280: AMDGPU: Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2}.

Works now. Since there were no changes to MUBUF_Atomic multiclass, this addresses the concerns by @nhaehnle.

Mar 29 2016, 4:45 PM
rivanvx updated the diff for D17280: AMDGPU: Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2}.

Merged Jan's approach. It would be nice if this works, because it's a much smaller diff.

Mar 29 2016, 8:45 AM

Mar 14 2016

rivanvx added a comment to D17300: llvm-config: fix --libs on Linux.

Thanks!

Mar 14 2016, 2:58 PM

Mar 10 2016

rivanvx updated the diff for D17280: AMDGPU: Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2}.

Updated per comments by @arsenm.

Mar 10 2016, 3:36 PM

Mar 8 2016

rivanvx added a comment to D17300: llvm-config: fix --libs on Linux.

@rnk can you re-review?

Mar 8 2016, 5:04 AM

Mar 2 2016

rivanvx added a comment to D17300: llvm-config: fix --libs on Linux.

Sorry for the delay, I only read "ehsan accepted this revision" and did not notice there were comments.

Mar 2 2016, 8:08 AM
rivanvx updated the diff for D17300: llvm-config: fix --libs on Linux.
Mar 2 2016, 8:07 AM

Feb 25 2016

rivanvx added a comment to D17046: AMDGPU: Document address space mapping.

@arsenm is this mergeable? Any other changes I should make?

Feb 25 2016, 1:51 PM
rivanvx updated the diff for D17280: AMDGPU: Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2}.

Posting updated version.

Feb 25 2016, 1:12 PM

Feb 24 2016

rivanvx added a comment to D17280: AMDGPU: Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2}.

Yes, I will post the updated patch per arsenm's suggestions tomorrow and one with tests over the weekend. I didn't find time to write the tests, sorry.

Feb 24 2016, 3:15 PM

Feb 22 2016

rivanvx added a comment to D17300: llvm-config: fix --libs on Linux.

rnk, ehsan, any comments? Anything else I should do?

Feb 22 2016, 3:22 AM

Feb 18 2016

rivanvx added a comment to D17281: [Hexagon] Fix compilation error with GCC 6.

Is there anything else I need to do for this to land?

Feb 18 2016, 7:52 AM

Feb 17 2016

rivanvx updated the diff for D17300: llvm-config: fix --libs on Linux.

Handle Windows first, output full path. Elsewhere, output -l and library name.

Feb 17 2016, 5:18 AM
rivanvx retitled D17280: AMDGPU: Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2} from Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2} to AMDGPU: Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2}.
Feb 17 2016, 4:42 AM

Feb 16 2016

rivanvx added a comment to D17300: llvm-config: fix --libs on Linux.

It does not do anything special on any OS, but I guess it should. I'm happy to adjust it for Windows if you give me the specification.

Feb 16 2016, 1:02 PM
rivanvx retitled D17300: llvm-config: fix --libs on Linux from to llvm-config: fix --libs on Linux.
Feb 16 2016, 11:24 AM
rivanvx added a comment to D17280: AMDGPU: Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2}.

Fixed all comments except intrinsics, will upload a new revision with tests and without intrinsics chages when I have tests ready.

Feb 16 2016, 6:39 AM
rivanvx added a reviewer for D17281: [Hexagon] Fix compilation error with GCC 6: colinl.
Feb 16 2016, 2:51 AM

Feb 15 2016

rivanvx retitled D17281: [Hexagon] Fix compilation error with GCC 6 from to [Hexagon] Fix compilation error with GCC 6.
Feb 15 2016, 3:25 PM
rivanvx added a comment to D17280: AMDGPU: Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2}.

Please ignore changes in lib/Target/AMDGPU/SIIntrinsics.td, that's unrelated and was uploaded by accident. Will fix on reupload.

Feb 15 2016, 3:12 PM
rivanvx retitled D17280: AMDGPU: Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2} from to Implement {BUFFER,FLAT}_ATOMIC_CMPSWAP{,_X2}.
Feb 15 2016, 3:10 PM

Feb 11 2016

rivanvx added a comment to D17046: AMDGPU: Document address space mapping.

Mark as done.

Feb 11 2016, 3:13 AM
rivanvx updated the diff for D17046: AMDGPU: Document address space mapping.

Updated per arsenm's comments.

Feb 11 2016, 3:12 AM

Feb 10 2016

rivanvx added a comment to D17046: AMDGPU: Document address space mapping.

Cool, I can mark comments as done.

Feb 10 2016, 10:32 AM
rivanvx abandoned D17071: [NVPTX] Update address space mapping documentation.

I see. Thanks for the info.

Feb 10 2016, 8:05 AM
rivanvx retitled D17071: [NVPTX] Update address space mapping documentation from to [NVPTX] Update address space mapping documentation.
Feb 10 2016, 5:21 AM
rivanvx updated the diff for D17046: AMDGPU: Document address space mapping.

Updated per Matt's suggestions.

Feb 10 2016, 3:59 AM

Feb 9 2016

rivanvx retitled D17046: AMDGPU: Document address space mapping from to AMDGPU: Document address space mapping.
Feb 9 2016, 3:12 PM

Jan 12 2016

rivanvx accepted D15986: llvm-config: handle BUILD_SHARED_LIBS correctly.

Works for me. Needs rebase to current master.

Jan 12 2016, 6:14 AM