Skip to content

Pull requests: NVIDIA/TransformerEngine

New pull request New
160 Open 2,360 Closed
160 Open 2,360 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
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[Common] Fix pointer arithmatic for MXFP8 to generate correct LDS/STS instructions
#3291 opened Jul 30, 2026 by kainzhong Collaborator Loading…
8 of 13 tasks
1
Enable THD with hierarchical context parallelism
#3290 opened Jul 30, 2026 by sudhakarsingh27 Member • Draft
2 of 13 tasks
@sudhakarsingh27
Fix cuDNN SDPA score_mod cache collision, sm_12x arch gates, per-device plan cache community-contribution PRs from external contributor outside the core maintainers, representing community-driven work.
#3289 opened Jul 30, 2026 by YangXu1990uiuc Loading…
1
Add opt-in reduced precision output for cuDNN MXFP8 norm community-contribution PRs from external contributor outside the core maintainers, representing community-driven work.
#3288 opened Jul 30, 2026 by sraman-rgb Collaborator Loading…
13 tasks
2
[PyTorch][torch.compile] Support for DotProductAttention on flash and unfused backends
#3286 opened Jul 30, 2026 by pggPL Collaborator Loading…
9 tasks done
1
[PyTorch] Preserve FP8 recompute state for inner autocast
#3284 opened Jul 30, 2026 by nvegesna-netizen Loading…
@pggPL
4
[JAX] Allow input tensors to be sharded with TP in EP dispatch/combine ops
#3283 opened Jul 30, 2026 by phu0ngng Collaborator • Draft
13 tasks
[Common][PyTorch] EP Fused prepare and dispatch for non-eager mode
#3282 opened Jul 30, 2026 by phu0ngng Collaborator • Draft
13 tasks
[Common] Fix NVFP4 stochastic rounding on architectures without cvt.rs community-contribution PRs from external contributor outside the core maintainers, representing community-driven work.
#3281 opened Jul 29, 2026 by davidkny22 Loading…
6 of 13 tasks
@ptrendx
2
[Common] Remove nv-internal-* comments 2.18
#3280 opened Jul 29, 2026 by ksivaman Member Loading…
5 of 13 tasks
1
[JAX] EP Dispatch with overflow detection option
#3277 opened Jul 29, 2026 by phu0ngng Collaborator Loading…
8 of 13 tasks
6
[Pytorch] [NCCL EP] Allow zero tokens for an EP rank in eager mode
#3276 opened Jul 29, 2026 by YangFei1990 Collaborator Loading…
7 of 13 tasks
2
[Common] Enable arch support for Rubin
#3275 opened Jul 29, 2026 by denera Collaborator Loading…
9 of 13 tasks
@denera
2
Add per-sequence causal policy to packed THD attention community-contribution PRs from external contributor outside the core maintainers, representing community-driven work.
#3274 opened Jul 29, 2026 by desh2608 Loading…
7 of 13 tasks
2
Minimize the memory usage of the fused cross entropy kernel
#3273 opened Jul 29, 2026 by ptrendx Member • Draft
1 of 13 tasks
[Common] Ensure quantization kernels handle noop properly
#3271 opened Jul 28, 2026 by kainzhong Collaborator Loading…
1 of 13 tasks
2
[Common][PyTorch] EP dispatch with unfused MXFP8 quantization
#3270 opened Jul 28, 2026 by phu0ngng Collaborator Loading…
8 of 13 tasks
3
[PyTorch] Add architecture gate to NVFP4 split_quantize RHT path community-contribution PRs from external contributor outside the core maintainers, representing community-driven work.
#3265 opened Jul 27, 2026 by davidkny22 Loading…
6 of 13 tasks
1
5
[PyTorch] MXFP4 weight QAT on MXFP8 and FP8 block-scaling recipes community-contribution PRs from external contributor outside the core maintainers, representing community-driven work.
#3264 opened Jul 27, 2026 by xiuhu17 Contributor Loading…
8 tasks done
2
[common] Fix UE8M0 code 0 (2^-127) and code 255 (NaN) expansion in ptx::exp2f community-contribution PRs from external contributor outside the core maintainers, representing community-driven work.
#3262 opened Jul 25, 2026 by xiuhu17 Contributor Loading…
8 tasks done
@ptrendx
10
[Common][PyTorch] Fuse NVFP4 quantization launches for MoE experts community-contribution PRs from external contributor outside the core maintainers, representing community-driven work.
#3261 opened Jul 25, 2026 by liujshi Loading…
13 tasks done
1
Use a temp directory for CMake build
#3254 opened Jul 24, 2026 by fheinecke Collaborator • Draft
3 of 13 tasks
1
Add CUDA wheels as build-time requirements for build isolation
#3253 opened Jul 24, 2026 by fheinecke Collaborator • Draft
3 of 13 tasks
1
Enable runtime resolution of CUDA header path for NVRTC
#3252 opened Jul 24, 2026 by fheinecke Collaborator • Draft
3 of 13 tasks
1
[CI] Improve build time dependency resolution
#3251 opened Jul 24, 2026 by fheinecke Collaborator • Draft
4 of 13 tasks
1
ProTip! Type g p on any issue or pull request to go back to the pull request listing page.