Skip to content

Pull requests: thu-ml/SageAttention

New pull request New
28 Open 42 Closed
28 Open 42 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

Patch kernels launch - launch them in the current CUDA stream
#380 opened Jul 28, 2026 by Abusagit Loading…
Fix hardcoded SM count (170) in Blackwell FP4 persistent kernel launch
#379 opened Jul 22, 2026 by sowmy86 Loading…
1
1
fix: prevent NaN outputs for short/non-multiple sequence lengths via CUDA-level zero-fill
#377 opened Jul 17, 2026 by Onkitova Loading…
feat: add Intel GPU (XPU) support via auto-round-lib dispatch
#375 opened Jul 6, 2026 by luoyu-intel Loading…
Fix Sage2 causal attention for unequal qkv lengths
#372 opened Jun 15, 2026 by qqq-tao Loading…
Launch sm90 attention and fused quant kernels on the current torch stream (CUDA-graph capture correctness)
#371 opened Jun 11, 2026 by shanai13 Loading…
Add Windows support for SageAttention3 Blackwell
#370 opened May 29, 2026 by sepcnt Loading…
1
Add native gfx12 attention backend
#368 opened May 13, 2026 by jammm Loading…
82
fix: use num_stages=1 on ROCm to avoid Triton pipelining crash on RDNA4 (gfx12xx)
#365 opened Apr 29, 2026 by brosequist Loading…
1
Comprehensive fix for Windows MSVC build errors (C2872 std, LNK2001) and thread-safety
#355 opened Mar 20, 2026 by munder-sa Loading…
Windows MSVC compile fix for SA2, SA3
#352 opened Feb 26, 2026 by loscrossos Loading…
1
add support of sparse:
#350 opened Feb 11, 2026 by STwangyingrui Loading…
Add Windows build workflow for GitHub Actions
#335 opened Dec 25, 2025 by hust-wx Loading…
SageAttention AMD version
#332 opened Dec 18, 2025 by eliotwang Loading…
27
Support head dim 256
#329 opened Dec 13, 2025 by woct0rdho Loading…
Sage Attention supports minimal modifications when compiling MS Visual Studio on Windows.
#323 opened Dec 8, 2025 by mengqin Loading…
5
Add MSVC support to setup.py for Windows builds
#293 opened Nov 2, 2025 by sfinktah Loading…
support TORCH_CUDA_ARCH_LIST and avoid link against libcuda.so at compile time
#245 opened Aug 24, 2025 by winggan Loading…
2
Fix windows build for pytorch 2.9
#243 opened Aug 21, 2025 by StrongerXi Loading…
1
2
resolve multi-CUDA_ARCHITECTURES compilation conflicts
#241 opened Aug 14, 2025 by GACLove Loading…
6
Fix setup.py for machines without GPUs
#239 opened Aug 13, 2025 by nedos Loading…
2
Update core.py Fix for CUDA <12.8
#202 opened Jul 2, 2025 by umerkay Loading…
2
support TORCH_CUDA_ARCH_LIST
#189 opened Jun 24, 2025 by shiyemin Loading…
1
Run on the correct CUDA stream
#175 opened May 19, 2025 by xrq-phys Loading…
1
Add support more NVIDIA devices
#160 opened Apr 22, 2025 by johnnynunez Loading…
11
ProTip! Exclude everything labeled bug with -label:bug.