[libclc] r315235 - Implement mem_fence on ptx

Author: jketema Date: Mon Oct 9 12:43:04 2017 New Revision: 315235 URL: http://llvm.org/viewvc/llvm-project?rev=315235&view=rev Log: Implement mem_fence on ptx PTX does not differentiate between read and write fences. Hence, these a lowered to a mem_fence call. The mem_fence function compiles to

[libclc] r315228 - Make ptx barrier work irrespective of the cl_mem_fence_flags

Author: jketema Date: Mon Oct 9 11:36:48 2017 New Revision: 315228 URL: http://llvm.org/viewvc/llvm-project?rev=315228&view=rev Log: Make ptx barrier work irrespective of the cl_mem_fence_flags This generates a "bar.sync 0” instruction, which not only causes the threads to wait, but does acts as

[libclc] r315147 - Delete empy directory

Author: jketema Date: Sat Oct 7 02:10:44 2017 New Revision: 315147 URL: http://llvm.org/viewvc/llvm-project?rev=315147&view=rev Log: Delete empy directory Removed: libclc/trunk/ptx/lib/integer/ ___ cfe-commits mailing list cfe-commits@lists.llvm.o

[libclc] r314998 - Add vload_half helpers for ptx

Author: jketema Date: Thu Oct 5 11:17:40 2017 New Revision: 314998 URL: http://llvm.org/viewvc/llvm-project?rev=314998&view=rev Log: Add vload_half helpers for ptx The removes the vload_half unresolved calls from the nvptx libraries. Reviewed-by: Jan Vesely Added: libclc/trunk/ptx/lib/sha

[libclc] r314925 - Add vstore_half helpers for ptx

Author: jketema Date: Wed Oct 4 12:07:48 2017 New Revision: 314925 URL: http://llvm.org/viewvc/llvm-project?rev=314925&view=rev Log: Add vstore_half helpers for ptx Reviewed-by: Jan Vesely Added: libclc/trunk/ptx/lib/SOURCES_3.9 libclc/trunk/ptx/lib/SOURCES_4.0 libclc/trunk/ptx/lib

[libclc] r314633 - Do no circularly define NULL

Author: jketema Date: Sun Oct 1 13:10:14 2017 New Revision: 314633 URL: http://llvm.org/viewvc/llvm-project?rev=314633&view=rev Log: Do no circularly define NULL Reviewed-by: Jan Vesely Modified: libclc/trunk/generic/include/clc/clcmacros.h Modified: libclc/trunk/generic/include/clc/clcma

[libclc] r314634 - Let get_work_dim take exactly 0 arguments

Author: jketema Date: Sun Oct 1 13:11:46 2017 New Revision: 314634 URL: http://llvm.org/viewvc/llvm-project?rev=314634&view=rev Log: Let get_work_dim take exactly 0 arguments Reviewed-by: Jan Vesely Modified: libclc/trunk/amdgcn/lib/workitem/get_work_dim.cl libclc/trunk/generic/include

[libclc] r294916 - Move BufferPtr into the block where it it being used

Author: jketema Date: Sun Feb 12 15:33:49 2017 New Revision: 294916 URL: http://llvm.org/viewvc/llvm-project?rev=294916&view=rev Log: Move BufferPtr into the block where it it being used The previous location outside the block would crash prepare-builtins when no the builtins file accidentially n

[libclc] r294915 - Add the correct prefixes to the cl_khr_fp64 pragma

Author: jketema Date: Sun Feb 12 15:31:41 2017 New Revision: 294915 URL: http://llvm.org/viewvc/llvm-project?rev=294915&view=rev Log: Add the correct prefixes to the cl_khr_fp64 pragma Modified: libclc/trunk/generic/include/clc/shared/vstore.h Modified: libclc/trunk/generic/include/clc/share

Re: r269670 - [OpenCL] Add supported OpenCL extensions to target info.

Hi Sam, Thanks baring with me, and thanks a lot for your explanation! I indeed now believe that this is correct; I got somewhat confused by all the macro magic that is going on, and had missed that the enabled extensions are actually stored in a class instance separate from the one with availab

Re: r269670 - [OpenCL] Add supported OpenCL extensions to target info.

Hi Sam, > This commit does not change the initial state of the extensions. An extension > is supported is not the same as enabled. At the beginning all extensions are > disabled. I do not see this reflected in the code at all. Could you please: a. Point me to the location where this distinctio

Re: r269670 - [OpenCL] Add supported OpenCL extensions to target info.

>> >>> I'd say Clang currently miss a feature to detect used extensions and emit >>> the metadata. >>> >>> Sam >>> >>> -Original Message- >>> From: Anastasia Stulova [mailto:anastasia.stul...@arm.com] >>&

Re: r269670 - [OpenCL] Add supported OpenCL extensions to target info.

Yaxun (Sam) ; Jeroen Ketema >> >> Cc: Clang Commits ; nd >> Subject: Re: r269670 - [OpenCL] Add supported OpenCL extensions to target >> info. >> >> Hi Sam, >> >> Has this been addressed? >> >> @Jeroen, as far as I am aware adding supported extensions is completely new >

Re: r269670 - [OpenCL] Add supported OpenCL extensions to target info.

27;t be braking any existing functionality that are related > to that. Could you elaborate on the problem. Would creating new Clang target > help? Otherwise, do you have any other proposal for the solution? > > Thanks, > Anastasia > > > >

Re: r263795 - Add -fnative-half-arguments-and-returns

Hi, The below commit is, as it enables half arguments and returns by default for OpenCL, while this is actually an extension that might not be supported on all platforms. I think that this part of the commit should be reverted: > --- cfe/trunk/lib/Frontend/CompilerInvocation.cpp (original) > ++

Re: r269670 - [OpenCL] Add supported OpenCL extensions to target info.

Hi, The below commit enables all OpenCL extensions for the SPIR target by default. This incorrect, as SPIR allows you to specify which extensions are enabled/disabled its metadata. This means that any SPIR generated by Clang may now be rejected by specific OpenCL platforms, because they might

r248888 - [ARM][NEON] Use address space in vld([1234]|[234]lane) and vst([1234]|[234]lane) instructions

Author: jketema Date: Wed Sep 30 05:56:56 2015 New Revision: 24 URL: http://llvm.org/viewvc/llvm-project?rev=24&view=rev Log: [ARM][NEON] Use address space in vld([1234]|[234]lane) and vst([1234]|[234]lane) instructions This is the clang commit associated with llvm r248887. This commit

Re: [PATCH] D13127: [ARM] Upgrade codegen for vld[234] and vst[234] to to communicate a 0 address space

This revision was automatically updated to reflect the committed changes. Closed by commit rL24: [ARM][NEON] Use address space in vld([1234]|[234]lane) and vst([1234]|… (authored by jketema). Changed prior to commit: http://reviews.llvm.org/D13127?vs=35871&id=36082#toc Repository: rL LLV

Re: [PATCH] D13127: [ARM] Upgrade codegen for vld[234] and vst[234] to to communicate a 0 address space

jketema updated this revision to Diff 35871. jketema added a comment. Sync with updates to the llvm part of this patch http://reviews.llvm.org/D13127 Files: lib/CodeGen/CGBuiltin.cpp test/CodeGen/arm-neon-misc.c test/CodeGen/arm-vector-align.c test/CodeGen/vld_dup.c Index: test/CodeGen

[PATCH] D13127: [ARM] Upgrade codegen for vld[234] and vst[234] to to communicate a 0 address space

jketema created this revision. jketema added a reviewer: sbaranga. jketema added a subscriber: cfe-commits. Herald added subscribers: rengolin, aemerson. This if the clang companion patch for http://reviews.llvm.org/D12985, which upgrades the vld[234] and vst[234] to take pointers annotated with

[libclc] r244987 - Remove files accidentally not removed in r244310

Author: jketema Date: Thu Aug 13 18:43:12 2015 New Revision: 244987 URL: http://llvm.org/viewvc/llvm-project?rev=244987&view=rev Log: Remove files accidentally not removed in r244310 Removed: libclc/trunk/generic/lib/LLVM3.6/ libclc/trunk/generic/lib/SOURCES_LLVM3.6 Removed: libclc/trunk

[libclc] r244310 - Require LLVM >=3.7 and bump version to 0.2.0

Author: jketema Date: Fri Aug 7 03:31:37 2015 New Revision: 244310 URL: http://llvm.org/viewvc/llvm-project?rev=244310&view=rev Log: Require LLVM >=3.7 and bump version to 0.2.0 v2: Also remove LLVM 3.6 traces from prepare-builtins.cpp Patch by: EdB Modified: libclc/trunk/configure.py