Skip to content

Issues: NVIDIA/Fuser

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Clear current search query, filters, and sorts
100 Open 221 Closed
100 Open 221 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Refactor IndexLowering::handle(const LoadStoreOp* ldst) enhancement New feature or request
#4058 opened Mar 11, 2025 by rdspring1
RFE: Take contiguity caching into nvFuser enhancement New feature or request
#4043 opened Mar 7, 2025 by csarofeen
inplace update done via aliased outputs should have more strict checks
#4036 opened Mar 7, 2025 by jjsjann123
1
benchmarking suite should initialize cuda graphs / profiler interaction Python Benchmarks
#4008 opened Mar 4, 2025 by tfogal
checking for compatible allocation domain on Fusion::replaceOutput
#3994 opened Feb 28, 2025 by jjsjann123
Reenable reshape canceling for Mistral backward rope
#3975 opened Feb 26, 2025 by naoyam
Error with driver API's lazy load of cuStream ops
#3907 opened Feb 17, 2025 by samnordmann
2
MarkAliasesPrepare to recognize meta ops with DID loop split. allocation domain issues related to allocation domain support Multi-GPU
#3902 opened Feb 15, 2025 by wujingyue
2
Reuse user process group. Multi-GPU
#3901 opened Feb 15, 2025 by wujingyue
1
Fix ReorderShardedAxis and MakeReshardingContiguous for DID loop split. Multi-GPU
#3900 opened Feb 15, 2025 by wujingyue
1
Feature request: Consider privatization instead of forwarding in fusion segmentation Segmentation Issues related to nvFuser Segmentation
#3832 opened Feb 5, 2025 by naoyam
Feature request: Extend the privatization to improve segmentation Segmentation Issues related to nvFuser Segmentation
#3830 opened Feb 5, 2025 by naoyam
Feature request: Fusing sibling exprs in segmentation Segmentation Issues related to nvFuser Segmentation
#3829 opened Feb 5, 2025 by naoyam
supporting float32 on Scalar values
#3816 opened Feb 3, 2025 by jjsjann123
Make FusionProfile object not a singleton and allow copying
#3771 opened Jan 28, 2025 by kshitij12345
LinearOp with multiple sharded dimensions
#3770 opened Jan 28, 2025 by samnordmann
2
pytest benchmark reporting incorrect benchmark time bug Something isn't working Python Benchmarks
#3753 opened Jan 23, 2025 by jjsjann123
9
IOByte computation in benchmarks
#3721 opened Jan 16, 2025 by Priya2698
3
TensorDomain::flatten should squeeze broadcast IDs as done by the usual reshape transform
#3691 opened Jan 9, 2025 by naoyam
pad on broadcast dimensions hitting assert during transform replay
#3660 opened Dec 31, 2024 by jjsjann123
Feature request: Extend the uop forwarding in the fusion segmenter to include other single-input trivial ops rope
#3647 opened Dec 25, 2024 by naoyam
10
Performance gap between manual nvfuser definition and thunder.jit for rmsnorm
#3629 opened Dec 20, 2024 by Priya2698
21
INTERNAL ASSERT FAILED at "nvfuser/csrc/device_lower/analysis/sync_information.cpp":812 Inconsistent parallelization found between TV48 and TV2 Thunder
#3609 opened Dec 18, 2024 by crcrpar
Implement register stealing Matmuls
#3584 opened Dec 12, 2024 by zasdfgbnm
Elect one thread and stick to it Matmuls
#3582 opened Dec 12, 2024 by zasdfgbnm
ProTip! Type g i on any issue or pull request to go back to the issue listing page.