projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: ce8245b) | patch
Remove use of llvm.codegen intrinsic for GPGPU codegen
authorYabin Hu <yabin.hwu@gmail.com>
Tue, 24 Jun 2014 08:11:36 +0000 (08:11 +0000)
committerYabin Hu <yabin.hwu@gmail.com>
Tue, 24 Jun 2014 08:11:36 +0000 (08:11 +0000)
commitcc91169fd7e683711528401ca3350b205721148e
tree8fc51bbc67999fb09da789127b13856ef82a963btree | snapshot
parentce8245b5fdfe700ec0fdda13a2e8257de60d70eccommit | diff
Remove use of llvm.codegen intrinsic for GPGPU codegen

We use llvm.codegen intrinsic to generate code for embedded LLVM-IR
strings. The reason we introduce such a intrinsic is that previous
clang/opt tools was NOT linked with various LLVM targets and their
AsmParsers and AsmPrinters. Since clang/opt been linked with all the
needed libraries, we no longer need the llvm.codegen intrinsic.

llvm-svn: 211573
polly/include/polly/CodeGen/PTXGenerator.h diff | blob | history
polly/lib/CodeGen/PTXGenerator.cpp diff | blob | history
polly/test/Cloog/CodeGen/GPGPU/1d_parallel.ll [new file with mode: 0644] blob
polly/test/Cloog/CodeGen/GPGPU/gpu___%1---%5.jscop [new file with mode: 0644] blob
polly/test/Cloog/CodeGen/GPGPU/gpu___%1---%5.jscop.transformed+gpu [new file with mode: 0644] blob
polly/utils/0001-Add-llvm.codegen-intrinsic.patch [deleted file] blob | history
Domain: System / Toolchain;