projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 1291166) | patch
[SE] KernelSpec return best PTX
authorJason Henline <jhen@google.com>
Tue, 13 Sep 2016 23:29:25 +0000 (23:29 +0000)
committerJason Henline <jhen@google.com>
Tue, 13 Sep 2016 23:29:25 +0000 (23:29 +0000)
commitb459eb352986891bb0ec4c146954c2627ed6fc8e
tree0abd56e6e3175221fff80b632f0d895d93108fa2tree | snapshot
parent1291166c110ecda7efe7be9a36aff04f6bc46611commit | diff
[SE] KernelSpec return best PTX

Summary:
Before, the kernel spec would only return PTX for exactly the requested
compute capability. With this patch it will now return the PTX with the
largest compute capability that does not exceed that requested compute
capability.

Reviewers: jlebar

Subscribers: jprice, jlebar, parallel_libs-commits

Differential Revision: https://reviews.llvm.org/D24531

llvm-svn: 281417
parallel-libs/streamexecutor/include/streamexecutor/KernelSpec.h diff | blob | history
parallel-libs/streamexecutor/lib/KernelSpec.cpp diff | blob | history
parallel-libs/streamexecutor/unittests/CoreTests/KernelSpecTest.cpp diff | blob | history
Domain: System / Toolchain;