feat: unify edge_degree + layer radial MLPs into single batched GEMM#1849
Draft
feat: unify edge_degree + layer radial MLPs into single batched GEMM#1849
Conversation
UnifiedRadialMLP consolidates edge_degree_embedding.rad_func and all layer rad_funcs into a single first-layer GEMM, reducing kernel launches and improving GPU utilization. Key changes: - UnifiedRadialMLP: batches edge_degree + layer first linear layers into single GEMM, processes tails separately - get_unified_radial_emb: returns [edge_degree_out, layer_0_out, ...] - rad_func=None sentinel: signals precomputed radials in EdgeDegreeEmbedding - Fast backends create UnifiedRadialMLP at prepare_model_for_inference time Also includes torch.compile compatibility fix: - ChgSpinEmbedding: replaced dict lookup with tensor arithmetic to avoid graph break from x.tolist() Performance: ~16 QPS on 2000 atoms (H200), forces match baseline.
The rescale_factor position change (dividing wigner_inv before bmm instead of after) introduces small floating-point variations. Increase absolute tolerance from 1e-6 to 2e-6 to accommodate this.
misko
force-pushed
the
unified_radial_edge_degree
branch
from
7c12230 to
ebf7b78
Compare
|
This PR has been marked as stale because it has been open for 30 days with no activity. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
UnifiedRadialMLP consolidates edge_degree_embedding.rad_func and all layer rad_funcs into a single first-layer GEMM, reducing kernel launches and improving GPU utilization.
Key changes:
Also includes torch.compile compatibility fixes: