[PATCH] D92483: AMDGPU - Use MUBUF instructions for global address space access

Stanislav Mekhanoshin via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Dec 21 14:40:38 PST 2020


rampitec added a comment.

In D92483#2466959 <https://reviews.llvm.org/D92483#2466959>, @t-tye wrote:

> I updated AMDGPUUsage to reflect the current state which is that gfx6* does not support amdhsa. But if this patch allows gfx6* to be supported without generic addresses then that table should also be updated as part of the patch.

It does not allow for generic pointers to be used on SI.

> We did discuss this further and decided that gfx6* should report it supports amdhsa with the restriction that it does not support generic addresses. AMDGPUUsage already notes the restriction in all relevant sections. The LIT tests would also need updating accordingly. The COMgr tests would also need updating.

AMDGPUUsage does not list amdhsa for all 3 SI targets, yet it is going to be accepted with this patch. If that is so then amdhsa has to be listed in the table for these targets.

> The alternative would be that the compiler reports an error message for gfx6* if the triple specified amdhsa (rather than crash as it does currently). But it was felt that the amdhsa ABI can allow targets that do not support generic to be included provided it is documented. There is no need to artificially restrict them as a user may want to try using the target on an HSA complaint runtime to implement OpenCL 1.2.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D92483/new/

https://reviews.llvm.org/D92483



More information about the llvm-commits mailing list