Issues
Open
flashinfer.nvfp4_quantize: make CuTe-DSL backend's `a_global_sf` arg a host side arg (to align CuTe-DSL backend with CUDA backend)
good first issueop: misc
Has a beginner-friendly labelContributing guide available
2 comments0 reactions1 assignee
Open
[Compilation] GDN decode cold kernel compilation time is too long on SM90/100/120/121
good first issueop: gdn
Why recommendedNo assignee yet · Has a beginner-friendly label
No assignee yetHas a beginner-friendly labelContributing guide available
3 comments0 reactions0 assignees
Open
DGX Spark (SM121) Current Support Audit
arch: sm12xgood first issueop: gemmwip
Why recommendedHas a beginner-friendly label · Contributing guide available
Has a beginner-friendly labelContributing guide available
14 comments1 reaction1 assignee
Open
BF16 hidden_states for trtllm_fp4_block_scale_moe
good first issueop: moe
Why recommendedHas a beginner-friendly label · Contributing guide available
Has a beginner-friendly labelContributing guide available
5 comments0 reactions1 assignee
Open
Next fix for gemm-allreduce two-shot
cute-dslgood first issueop: commop: gemmrefactor
Why recommendedHas a beginner-friendly label · Contributing guide available
Has a beginner-friendly labelContributing guide available
10 comments0 reactions1 assignee