Search results

[Feature Request] <code>trtllm_batch_decode_with_kv_cache_mla</code> Batch Invariant support

needs-triage

Status: Open.

#2107 In flashinfer-ai/flashinfer;

· yewentao256 opened on Nov 18, 2025

[Feature Request] Support all head dims in BatchQKApplyRotaryPosIdsCosSinCache

needs-triage

Status: Open.

#2104 In flashinfer-ai/flashinfer;

· merrymercy opened on Nov 18, 2025

[feature request] add optimal gb200 moe comm kernels

needs-triage

Status: Open.

#2094 In flashinfer-ai/flashinfer;

· nvrohanv opened on Nov 14, 2025

[BUG] Flashinfer with jit-cache and cubin still attempts to jit

needs-triage

Status: Open.

#2093 In flashinfer-ai/flashinfer;

· bbartels opened on Nov 14, 2025

【Bug】Test prefill attention kv_dtype=fp8 error

needs-triage

Status: Open.

#2085 In flashinfer-ai/flashinfer;

· heliqi opened on Nov 13, 2025

MoE autotune print a lot failed kernel on SM120

bugSomething isn't workingSomething isn't working

Status: Open.

#2077 In flashinfer-ai/flashinfer;

· weireweire opened on Nov 11, 2025

[Feature Request] FP8 block quant support for AR+RMSNOrm , AR+RMSNorm+Quant.

feature requestNew feature or requestNew feature or request

Status: Open.

#2073 In flashinfer-ai/flashinfer;

· kmrao-nv opened on Nov 10, 2025

[Installation failed] Is <code>nvidia-cudnn-frontend&gt;=1.13.0</code> required?

needs-triage

Status: Open.

#2068 In flashinfer-ai/flashinfer;

· ziminz19 opened on Nov 10, 2025

trtllm-mla seems to have long idle time in some cases (200us idle for a 650us kernel)

performance

priority: must have (P0)

Status: Open.

#2066 In flashinfer-ai/flashinfer;

· fzyzcjy opened on Nov 8, 2025

[Feature Request] Determinism kernels

feature requestNew feature or requestNew feature or request

Status: Open.

#2059 In flashinfer-ai/flashinfer;

· kmrao-nv opened on Nov 7, 2025

[Bug] Segmentation fault when building flashinfer aot of v0.3.1.

needs-triage

Status: Open.

#2054 In flashinfer-ai/flashinfer;

· XuZhang99 opened on Nov 6, 2025

[cuda graphs] Support variable batch size with cuda graphs

feature requestNew feature or requestNew feature or request

needs-triage

Status: Open.

#2046 In flashinfer-ai/flashinfer;

· AgrawalAmey opened on Nov 5, 2025