Search results

Sageattention 3 reduce the generated video quality (Wan2.2 model) but yes it is much faster

Status: Open.

#369 In thu-ml/SageAttention;

· kananaestate opened on May 19, 2026

[Bug Report] Video Temporal Flickering when using SageAttention in ComfyUI

Status: Open.

#367 In thu-ml/SageAttention;

· voidyear0320 opened on May 12, 2026

Private security contact for SageAttention vulnerability report

Status: Open.

#366 In thu-ml/SageAttention;

· igorsantosgrueiro opened on May 5, 2026

Can we get modern install?

Status: Open.

#363 In thu-ml/SageAttention;

· Dunc4n1dah0 opened on Apr 22, 2026

How to compile with Torch 2.11.0, Python 3.14, Linux?

Status: Open.

#361 In thu-ml/SageAttention;

· noobmaster777 opened on Apr 11, 2026

_qattn_sm80 extension fails on Ada Lovelace (sm89/L40S) — illegal instruction at runtime or compilation failure

Status: Open.

#360 In thu-ml/SageAttention;

· ShaiDiamant opened on Apr 10, 2026

pytorch2.11.0+cuda13.0 5090无法编译

Status: Open.

#359 In thu-ml/SageAttention;

· tonylee2022 opened on Apr 7, 2026

Supporting B300 sm103

Status: Open.

#358 In thu-ml/SageAttention;

· alexbevz opened on Mar 29, 2026

[Bug] sageattn3_blackwell: CUDA error: misaligned address on RTX 5060 Ti (Blackwell)

Status: Open.

#357 In thu-ml/SageAttention;

· gzsiang opened on Mar 29, 2026

implementation for sageattn_qk_int8_pv_int8_cuda

Status: Open.

#356 In thu-ml/SageAttention;

· AIgiraffe opened on Mar 26, 2026

Is there a plan to refacotr SageAttention2.2 following the warp specialization style?

Status: Open.

#354 In thu-ml/SageAttention;

· Deeperfinder opened on Mar 17, 2026

Failed to run SageAttention in a Docker Container on Ubuntu 24.02 with an RTX A4000 SFF

Status: Open.

#353 In thu-ml/SageAttention;

· ctschach opened on Mar 4, 2026