Skip to content

Conversation

@alphalm4
Copy link
Contributor

@alphalm4 alphalm4 commented Jan 19, 2026

Fixes #260

D3Calculator compile issue: CUDA architecture detection and environment setup:

  • Updated sevenn/calculator.py to dynamically detect the GPU's CUDA architecture using torch.cuda.get_device_capability() when TORCH_CUDA_ARCH_LIST is absent. If TORCH_CUDA_ARCH_LIST is set but does not match the detected CUDA architecture version, it warns the user if there is a mismatch or missing configuration.

LAMMPS patch script error handling:

  • Modified sevenn/pair_e3gnn/patch_lammps.sh to no longer exit with an error when the flashTP_so shared object is missing or invalid. Instead, it logs a message and continues.

@alphalm4 alphalm4 requested a review from YutackPark January 19, 2026 18:40
@YutackPark YutackPark merged commit 4af45c1 into MDIL-SNU:main Jan 19, 2026
0 of 2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Minor issues

2 participants