- User Since
- Oct 1 2016, 8:22 AM (172 w, 17 h)
Aug 27 2017
Aug 22 2017
The novice reader would associate shape with only two of the DerivedTypes: VectorType and ArrayType. How could a FunctionType or an IntegerType have a shape ?
Could you suggest a paper I could read to get to know this Zone Algorithm ?
Aug 14 2017
Sorry about the delayed response.
Aug 8 2017
Aug 3 2017
This breaks my Julia build.
@bollu what is this patch about ?
Aug 2 2017
Aug 1 2017
Simpler check to determine the destination Module of the cloned instruction.
Jul 31 2017
@grosser I was able to commit it.
Jul 30 2017
I'm comitting on the release_50 branch only.
I got the following output on git svn dcommit -n,
Jul 29 2017
@grosser could you please approve ? Also
restrict to 80 columns.
I'm using git-svn. Would doing a arc patch D36050 on the release_50 branch followed by a merge and then git svn dcommit push it to llvm.org/git/polly.git ?
remove the llvm.lifetime* declarations
Jul 24 2017
How are invariant loads to pointers turned into scalar accesses ? Do we need -polly-invariant-load-hoisting for that ?
Jul 23 2017
Jul 22 2017
Simpler, Unoptimized and readable testcase.
Jul 21 2017
Added pseudo code to test case.
@PhilippSchaad Please go ahead.
Added test case.
Jul 20 2017
Jul 19 2017
Jul 17 2017
Jul 16 2017
@grosser Thanks for adding me as a reviewer ! It's helped me acquaint myself with the SPIR V codegen and understand GPUJIT better.
Jul 15 2017
Sure. Will do in sometime.
Jul 12 2017
Yes. I'll drop it.
- Instructions and basic blocks of new test case have been named.
- Removed format error with unittests/ScopPassManager/PassManagerTest.cpp
Jul 9 2017
What do you mean by killing a scalar ?
@grosser I tried what you suggessted,
Jul 5 2017
Jul 4 2017
Drop Scop name from ptx_kernel's name.
Jul 3 2017
- Moved building the name into a routine.
- ScopNameofIR is const now.
- Rebased to trunk @ 306540.
Jul 1 2017
The name of the Scop is built from the boundary basic blocks of the region in Scop::getNameStr. How does that depend on the build type ? Are extra basic blocks inserted in debug builds ?
Jun 30 2017
Jun 29 2017
Jun 28 2017
Update diff to include changes till https://llvm.org/svn/llvm-project/polly/trunk@306540
Why wasn't it supported previously ?
Jun 23 2017
This patch corrects the semantics of the accessor methods of the Scop class.
- Clearly mention the source FUNCtion and SCOP and KERNEL number.
- Changed name of SCOP prefixed to kernel to follow PTX identifier syntax acc. to http://docs.nvidia.com/cuda/parallel-thread-execution/index.html#identifiers (Raised by @philip.pfaffe )
Jun 22 2017
Jun 21 2017
Prefix SCoP name as well to uniquely identify a kernel, since a function can have more than one SCoP.
Jun 18 2017
Yes, we can handle the metadata problem in a different patch.
Jun 17 2017
@bollu Please refer this comment on my patch. I observed a pattern with functions that produced the buggy metadata, simple copy kernels (A[i][j][k] = B[i][j][k]) didn't produce buggy metadata but kernels that had computation (A[i][j][k] +=/-=/*= B[i][j][k]) resulted in Modules that didn't list the DICompileUnit in llvm.dbg.cu.
Jun 16 2017
- Rename TARGET_Hybrid to TARGET_HYBRID
- Reduce code redundancy.
Jun 15 2017
@bollu The patch did prevent passing the address of the debug intrinsic to the kernel, but couldn't guard against buggy metadata,
Jun 14 2017
Jun 13 2017
Jun 12 2017
- Punctuation consistency in comments.
- Align code to LLVM's C/C++ format.
- Initialized SkipScop at constructor like all other members.
- Include code for GPU targeting only when GPU_CODEGEN is defined.
Jun 10 2017
I'm still waiting for my commit access 😁.