Issues

• fla backend support

  SandAI-org/MagiAttention#350 · Filter by author liujia-cc opened on Aug 19, 2026

• [native CuteDSL FFA] Long Term native CuteDSL FFA dev Progress Tracker

  SandAI-org/MagiAttention#349 · Filter by author cherhh opened on Aug 18, 2026
  

• really need warmhearted provide a complied wheel

  SandAI-org/MagiAttention#340 · Filter by author songqikong opened on Jul 29, 2026

• Feature request: support NVIDIA GB10 (sm121) with FFA_FA4 backend

  SandAI-org/MagiAttention#321 · Filter by author ZYC99 opened on Jun 6, 2026

• Feature request: chunk-causal mask type for Flex Flash Attention

  SandAI-org/MagiAttention#316 · Filter by author xin-w8023 opened on May 15, 2026

• Windows checkout fails: invalid path with<and>in filenames

  SandAI-org/MagiAttention#311 · Filter by author zhuochenKIDD opened on Apr 26, 2026

• ssm support

  SandAI-org/MagiAttention#310 · Filter by author jordane95 opened on Apr 23, 2026
  1

• [Feature Request] Support head_dim=256 on SM100 (Blackwell)

  SandAI-org/MagiAttention#309 · Filter by author LucienXian opened on Apr 7, 2026
  2

• Does magi support zigzag dispatch pattern?

  SandAI-org/MagiAttention#196 · Filter by author jordane95 opened on Dec 29, 2025
  1

  

• TypeError: flex_flash_attn_func() got an unexpected keyword argument 'max_seqlen_q'

  SandAI-org/MagiAttention#193 · Filter by author hengrui0516 opened on Dec 16, 2025
  2

• Are there plans to support sparse attention inference on 4090 and 5090?

  SandAI-org/MagiAttention#184 · Filter by author helloyongyang opened on Dec 8, 2025
  2

  

• Using dummy/padded attention slices with flex_flash_attn_func for torch.compile static shapes

  SandAI-org/MagiAttention#180 · Filter by author keunhong opened on Dec 2, 2025
  2

  

• Support 256 Head Dim

  SandAI-org/MagiAttention#165 · Filter by author lin-ht opened on Nov 5, 2025
  4

• [Support for Cross-Attention] How to use cross-attention with magiattention

  SandAI-org/MagiAttention#123 · Filter by author zelingao98 opened on Aug 26, 2025
  2

  

• NSA Support

  SandAI-org/MagiAttention#114 · Filter by author jensenzhoujh opened on Aug 11, 2025
  2

• Blackwell support

  SandAI-org/MagiAttention#105 · Filter by author littsk opened on Aug 2, 2025
  

• Dynamic Load Blancer

  SandAI-org/MagiAttention#104 · Filter by author littsk opened on Aug 2, 2025
  

• Nvshmem Based GroupCast & GroupReduce

  SandAI-org/MagiAttention#103 · Filter by author littsk opened on Aug 2, 2025
  

• Performance Optimize For Sparse Attention

  SandAI-org/MagiAttention#99 · Filter by author littsk opened on Aug 1, 2025
  2

  

• feature request: support for MLA

  SandAI-org/MagiAttention#64 · Filter by author vorobyov01 opened on Jul 10, 2025
  2