Search results

support for B200?

Status: Open.

#322 In thu-ml/SageAttention;

· peepeepeepoopoopoo opened on Dec 4, 2025

sage-attention 3 mosaic artifacts and not faster than sage-attention 2.2 on dgx spark

Status: Open.

#321 In thu-ml/SageAttention;

· atipasvanund opened on Dec 4, 2025

SM90 backend broken in current version of Sage Attention

Status: Open.

#320 In thu-ml/SageAttention;

· dylanprins opened on Dec 3, 2025

Windows11 python 3.12 cuda 13.0 torch 2.9.1 Compilation failed

Status: Open.

#319 In thu-ml/SageAttention;

· wzgrx opened on Dec 2, 2025

Request: Add Backward Pass support for SageAttention3 (Training/Fine-tuning)

Status: Open.

#317 In thu-ml/SageAttention;

· xTimeCrystal opened on Nov 29, 2025

Why am I not able to increase the speed of z-image turbo by 20% using Patch Sage Attention? It should be giving me generations approximately every 0.76 seconds

Status: Open.

#316 In thu-ml/SageAttention;

· peepeepeepoopoopoo opened on Nov 27, 2025

compile in window is quite tricky

Status: Open.

#314 In thu-ml/SageAttention;

· li-dahua opened on Nov 22, 2025

Can sage_attn3 be used in SM100？

Status: Open.

#313 In thu-ml/SageAttention;

· huwei02 opened on Nov 21, 2025

Sage Attention does not improve the speed of Qwen Image Workflow

Status: Open.

#312 In thu-ml/SageAttention;

· peepeepeepoopoopoo opened on Nov 19, 2025

help: rtx pro 6000 can't build sageattention blackwall

Status: Open.

#311 In thu-ml/SageAttention;

· asking002 opened on Nov 19, 2025

文章的lambada数据集

Status: Open.

#310 In thu-ml/SageAttention;

· GuoHaoren opened on Nov 19, 2025

FP8 QK kernels unstable on Ada (sm89, RTX 4060 Ti) — BF16/FP8 path causes launch failures, FP16 fallback fully resolves

Status: Open.

#309 In thu-ml/SageAttention;

· NasBuk opened on Nov 18, 2025