Switch from HIP_PLATFORM_HCC to HIP_PLATFORM_AMD #4539
Conversation
loadams
requested review from
RezaYazdaniAminabadi,
arashb,
awan-10,
cmikeh2 and
jeffra
as code owners
|
@jithunnair-amd if you could review this to make sure I'm implementing this correctly, since I believe this is deprecated: https://rocm.docs.amd.com/projects/HIP/en/latest/user_guide/hip_porting_guide.html#identifying-hip-target-platform |
|
Need to wait for the additional one added in #4538. |
|
Looks like this was added in ROCM 4.1 release (which is 2+ years old). So we shouldn't run into any issues with backwards compatibility, right? I presume most if not all AMD users are on ROCM4.1+ |
That's what it looks like to me, and it will be fully deprecated in ROCm 6+ coming later this year, it also looks like that means maybe only the MI25s would be impacted? So I think we should be fine. |
* Switch from HIP_PLATFORM_HCC to HIP_PLATFORM_AMD * Merge changes and fix from deepspeedai#4528
This PR is required in addition to #4539 to define HIP_PLATFORM_AMD on ROCm. This is required for DeepSpeed non-JIT build. For JIT build we have #4570. This is required for DeepSpeed extensions build in those docker images with PyTorch built before pytorch/pytorch#111975. cc: @jeffdaily @jithunnair-amd Co-authored-by: Logan Adams <114770087+loadams@users.noreply.github.com>
* Switch from HIP_PLATFORM_HCC to HIP_PLATFORM_AMD * Merge changes and fix from deepspeedai#4528
This PR is required in addition to deepspeedai#4539 to define HIP_PLATFORM_AMD on ROCm. This is required for DeepSpeed non-JIT build. For JIT build we have deepspeedai#4570. This is required for DeepSpeed extensions build in those docker images with PyTorch built before pytorch/pytorch#111975. cc: @jeffdaily @jithunnair-amd Co-authored-by: Logan Adams <114770087+loadams@users.noreply.github.com>
This appears to have been deprecated a while ago. AMD docs here and here.