Add grid-stride + ROCm cap to split optimizer update kernel - #6086
Open
q10 wants to merge 1 commit into
Open
Conversation
Summary:
The standalone split optimizer path launches split_{optimizer}_update_kernel
with grid = div_round_up(grad_dev_indices.numel(), kMaxThreads / kThreadGroupSize)
and block = dim3(kThreadGroupSize, kMaxThreads / kThreadGroupSize, 1), so total
threads ~= num_unique_indices * kThreadGroupSize exceeds the HIP 2^32
threads-per-launch limit on ROCm for large index counts.
Cap the launch with utils::cuda::cap_grid_dim_x(..., OverflowOnly) and add a
ROCm grid-stride loop over run_id to the kernel (single guard converted to the
loop bound; no internal early-returns). No-op on CUDA.
Reviewed By: henrylhtsang
Differential Revision: D113351687
Contributor
|
@q10 has exported this pull request. If you are a Meta employee, you can view the originating Diff in D113351687. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary:
The standalone split optimizer path launches split_{optimizer}_update_kernel
with grid = div_round_up(grad_dev_indices.numel(), kMaxThreads / kThreadGroupSize)
and block = dim3(kThreadGroupSize, kMaxThreads / kThreadGroupSize, 1), so total
threads ~= num_unique_indices * kThreadGroupSize exceeds the HIP 2^32
threads-per-launch limit on ROCm for large index counts.
Cap the launch with utils::cuda::cap_grid_dim_x(..., OverflowOnly) and add a
ROCm grid-stride loop over run_id to the kernel (single guard converted to the
loop bound; no internal early-returns). No-op on CUDA.
Reviewed By: henrylhtsang
Differential Revision: D113351687