This is an archive of the discontinued LLVM Phabricator instance.

[CUDA] Added support for CUDA-8
ClosedPublic

Authored by tra on Sep 26 2016, 5:03 PM.

Diff Detail

Repository
rL LLVM

Event Timeline

tra updated this revision to Diff 72586.Sep 26 2016, 5:03 PM
tra retitled this revision from to [CUDA] Added support for CUDA-8.
tra updated this object.
tra added a reviewer: jlebar.
tra added a subscriber: cfe-commits.
jlebar added inline comments.Sep 27 2016, 1:29 PM
lib/Headers/__clang_cuda_runtime_wrapper.h
139 ↗(On Diff #72586)

Nit, missing period.

140 ↗(On Diff #72586)

typo

156 ↗(On Diff #72586)

I don't understand what we are doing here...

We're saying, if USE_FAST_MATH is defined, and if it's not equal to 0, then redefine it equal to 1? Isn't that a compile error?

tra marked an inline comment as done.Sep 27 2016, 2:03 PM
tra added inline comments.
lib/Headers/__clang_cuda_runtime_wrapper.h
156 ↗(On Diff #72586)

Not if it happens in system headers.
That said, I can eliminate true branch of the conditional and only set those macros to 0 if they are undefined.
It will be up to whoever sets those macros to set them correctly otherwise.

tra updated this revision to Diff 72707.Sep 27 2016, 2:05 PM

addressed Justin's comments.

jlebar accepted this revision.Sep 27 2016, 2:06 PM
jlebar edited edge metadata.
This revision is now accepted and ready to land.Sep 27 2016, 2:06 PM
This revision was automatically updated to reflect the committed changes.