flashinfer-ai/flashinfer-bench

Building the Virtuous Cycle for AI-driven LLM Systems

Stars: 281
Forks: 49
Open beginner issues: 0
Indexed issues: 20

Dominant language: Python
License: Apache-2.0
Last GitHub push: May 1, 2026
Latest indexed: Sep 16, 2026
Contributing guide: Contributing guide
Code of conduct: No code of conduct
Beginner labels: No beginner labels indexed
PR merge metrics: No merged PRs in 30d

Open All

20 open issues indexed

How to dump the fast_p chart as figure at local? Open

flashinfer-ai/flashinfer-bench#105 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 16, 2026
Is the Kernel in solutions generated by the kernel generator in example? Open

flashinfer-ai/flashinfer-bench#112 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 16, 2026
[Tracking] Reduce apply() overhead & improve Adapter usability Open

enhancement

flashinfer-ai/flashinfer-bench#115 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 16, 2026
Kernel request: Low-precision and mixed-precision kernels for Mixture-of-Experts Open

flashinfer-ai/flashinfer-bench#141 · 2 comments · 0 reactions · 1 assignee · Indexed Sep 16, 2026
Kernel request: Deepseek sparse attention Open

flashinfer-ai/flashinfer-bench#142 · 0 comments · 0 reactions · 1 assignee · Indexed Sep 16, 2026
Kernel request: Gated Delta Net Open

flashinfer-ai/flashinfer-bench#143 · 0 comments · 0 reactions · 1 assignee · Indexed Sep 16, 2026
High Latency in _generate_code_from_prompt Function: Tens of Minutes vs Expected 3 Minutes per Kernel (DeepSeek-v3.2-reasoner) Open

flashinfer-ai/flashinfer-bench#149 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 16, 2026
[Bug Report] High failure rate with gemini-3-pro-preview due to "SyntaxError" (conversational output) and Signature Mismatches Open

flashinfer-ai/flashinfer-bench#151 · 3 comments · 4 reactions · 0 assignees · Indexed Sep 16, 2026
Property rate consult Open

flashinfer-ai/flashinfer-bench#169 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 16, 2026
[Bug Report] flashinfer_moe solution incompatible with flashinfer 0.6.2 (tile_tokens_dim removed) Open

flashinfer-ai/flashinfer-bench#177 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 16, 2026
the dependency installed from pypi doesn`t contain the "agent" module, please update the pip version Open

flashinfer-ai/flashinfer-bench#188 · 0 comments · 4 reactions · 0 assignees · Indexed Sep 16, 2026
Support for CuTe headers in tvm/torch FFI? Open

flashinfer-ai/flashinfer-bench#193 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 16, 2026
How to Get Overall Benchmark Results for New Model Open

flashinfer-ai/flashinfer-bench#198 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 16, 2026
[Bug] Workload blob path format inconsistent Open

flashinfer-ai/flashinfer-bench#201 · 0 comments · 1 reaction · 0 assignees · Indexed Sep 16, 2026
Does _clone_args cause unintended plan() re-execution in FlashInfer baseline solutions benchmarks? Open

flashinfer-ai/flashinfer-bench#211 · 3 comments · 1 reaction · 0 assignees · Indexed Sep 16, 2026
model tracker: improve model/kernel coverage - 20260311 Open

flashinfer-ai/flashinfer-bench#212 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 16, 2026
No healthy persistent workers available` after workload shape mismatch Open

flashinfer-ai/flashinfer-bench#215 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 16, 2026
Bug: Kernels can cheat the benchmark by caching their outputs. Open

flashinfer-ai/flashinfer-bench#228 · 0 comments · 1 reaction · 0 assignees · Indexed Sep 16, 2026
Test @flashinfer-bot -p "" Open

flashinfer-ai/flashinfer-bench#406 · 3 comments · 2 reactions · 0 assignees · Indexed Sep 16, 2026
CuTe-DSL kernels are not captured into torch.cuda.graph, causing silent under-measurement (and stale-output correctness bypass) in flashinfer-bench Open

flashinfer-ai/flashinfer-bench#414 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 16, 2026

A short digest of beginner-friendly GitHub issues.