# MoonshotAI/FlashKDA

FlashKDA: high-performance Kimi Delta Attention kernels

Repository: https://github.com/MoonshotAI/FlashKDA
Canonical: https://ross.abutalabs.com/products/flashkda
Language: Cuda
License: MIT
License Family: permissive
Last push: 2026-07-30T07:14:39+00:00

## Health v2 (maintenance only)
Score: 57/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 95, release rhythm 35, longevity 9
- inputs: {"age_days": 135, "days_push": 34, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1229, forks 118 (observed 2026-08-28T04:04:03.581495+00:00)

## What it is
FlashKDA is a set of high-performance CUDA kernels (built on CUTLASS) implementing Kimi Delta Attention, a linear attention mechanism, for NVIDIA SM90+ GPUs. It integrates as an auto-dispatched backend for the flash-linear-attention library's chunk_kda operation.

## Use cases
- run kimi delta attention fast on gpu
- speed up linear attention kernels for llm training
- flash attention style kernel for delta rule attention
- backend for flash-linear-attention chunk_kda
- benchmark attention kernel performance on H100
- train models with gated linear attention efficiently

## When to choose
- you need fast KDA forward passes on NVIDIA Hopper (SM90+) GPUs
- you already use flash-linear-attention and want a faster backend
- you train or serve models using Kimi Delta Attention with PyTorch 2.4+

## When to avoid
- you have pre-Hopper GPUs (SM80 or earlier)
- you need a pure Triton or CPU implementation
- you don't use PyTorch or CUDA 12.9+

## Facets
- artifact type: library
- maturity: active
- function: llm-inference, gpu-computing, deep-learning, machine-learning
- domain: deep-learning, large-language-models, gpu-computing
- platform: python
- tags: cuda-kernels, cutlass, linear-attention, flash-attention, pytorch, kimi-delta-attention, hopper, linux, gpu, cuda

## Member repositories
- MoonshotAI/FlashKDA (main) score 57

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:03.581495+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:15:05.079810+00:00, confidence not recorded.
  - readme: https://github.com/MoonshotAI/FlashKDA (fetched 2026-08-28T04:04:03.581495+00:00, sha 78f67f526f91)
- Data as of 2026-08-30T08:39:29.467469+00:00.