Essentially moves the foreach over sm integers into a macro and instantiates it for nvptx.
NFC in that the macro is not presently instantiated for amdgpu as the corresponding code doesn't compile yet.
Differential D111987
[libomptarget][nfc]Generalise DeviceRTL cmake to allow building for amdgpu JonChesterfield on Oct 18 2021, 4:52 AM. Authored by
Details Essentially moves the foreach over sm integers into a macro and instantiates it for nvptx. NFC in that the macro is not presently instantiated for amdgpu as the corresponding code doesn't compile yet.
Diff Detail
Event TimelineComment Actions Can we split this patch to two, one is to generalize and another is to add AMD support? Comment Actions This is the second half of that split, the refactor is in D111983. Unfortunately I was unable to clearly express that in phabricator (worked initially, lost it in an update - if you check the history of this patch the first revision is what I'd have liked this to look like). So my hope is to land D111983 soon, which I don't think has any outstanding problems, and then rebase this to get something more legible.
Comment Actions LGTM
|
Consider
or maybe even cmake_parse_arguments of more options might be added in the future.