microsoft / microsoft/TRELLIS.2
Silent failure in SparseConvNeXtBlock3d in RTX 5080
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 11.3k
- Forks
- 1.4k
- PR merge metrics
- No merged PRs in 30d
Description
I was trying to use the ComfyUI Trellis2 setup, but my generation keeps failing on the sample shape slat function. I debug it down to the upsample block failing. This module specifically SparseConvNeXtBlock3d, dies silently on the forward(). That's in main Trellis2 repo but I assume that is calling the cuda kernels defined for the sparse convs from flexgemm.
I installed flexgemm as per the README.
Any ideas why and whether special considerations must be made for sm120?
Running on Windows
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with the SparseConvNeXtBlock3d forward path used by the sample shape slat function, then review the flexgemm README and its sparse-convolution setup. Reproduce the silent failure on Windows with the RTX 5080 and check whether the sm120 consideration is documented. Done means identifying the compatibility cause and documenting or resolving the required setup.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- computer-graphics, machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100