Repository navigation
Add Intel GPU (XPU) support for inference - #358
Open
rajeshgangireddy wants to merge 8 commits into
Open
rajeshgangireddy wants to merge 8 commits into
rajeshgangireddy wants to merge 8 commits into
Conversation
- Updated kernel documentation to clarify behavior on non-CUDA runtimes and added notes for Intel XPU support. - Introduced a portable installation guide for Intel XPU, emphasizing single-device inference and kernel defaults. - Refactored device management in the Protenix model to utilize a new utility module for better device type handling. - Implemented device-aware autocasting and cache management across various model components. - Enhanced tests to dynamically adapt to the available device type, ensuring compatibility with both CUDA and XPU.
…on in Protenix initialization
|
|
rajeshgangireddy
changed the title
rajeshgangireddy
marked this pull request as ready for review
October 20, 2026 07:27
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Hi
Thank you for the great work. I would like to add XPU support to this project.
This PR adds single-device Intel XPU inference using portable PyTorch kernels. On XPU, triangle attention and triangle multiplication default to the
torchbackends; CUDA-only kernel choices fail early. Device placement, autocast, and cache handling now support XPU, while CUDA keeps its existing defaults. The PR also includes XPU installation instructions and a portable requirements file.We validated the tiny v0.5.0 checkpoint in bf16 on an Intel Arc Pro B70 and NVIDIA RTX 3090 side by side in the same machine. Earlier full-sampling runs covered protein-only, mixed protein/DNA/ligand, and precomputed-MSA inputs with
torchtriangle kernels. For protein-only input, matched-seed aligned all-atom RMSD was 0.23 and 0.44 Å (seeds 101 and 102); complex inputs showed larger differences. Latest test suites: XPU 125 passed / 10 skipped; CUDA 129 passed / 6 skipped.Please let me know if you need any further validation/experimentation numbers.
This PR covers inference only.
If it goes well, we can follow up with training support in a separate PR.