DeepGEMM

Clean, high-performance FP8 GEMM kernels with fine-grained scaling for modern inference workloads. Apache 2.0 licensed.

Guide pending

We haven't written this guide yet.

Guides are drafted from the project's own README and metadata. Until one exists, check the product's repository and documentation for authoritative information.

Nearby on the map

Related projects

DeepEP

deepseek-ai/DeepEP

Communication library for efficient expert-parallel training/inference pipelines, reducing MoE cross-device communication overhead. Apache 2.0 licensed.

unknown unknownactivity unknown

FlashKDA

MoonshotAI/FlashKDA

High-performance CUTLASS-based Kimi Delta Attention CUDA kernels for linear attention mechanisms. MIT licensed.

unknown unknownactivity unknown

RAFT

rapidsai/raft

CUDA-accelerated algorithms and ANN building blocks for high-performance similarity search, clustering, and matrix learning workloads.

unknown unknownactivity unknown

ThunderKittens

HazyResearch/ThunderKittens

CUDA tile primitives and kernel templates for accelerating transformer attention blocks. MIT licensed.

unknown unknownactivity unknown

tt-metal

tenstorrent/tt-metal

TT-Metalium + TT-NN operator stack for building and optimizing kernels on Tenstorrent AI accelerators. Apache 2.0 licensed.

unknown unknownactivity unknown

mini-sglang

sgl-project/mini-sglang

Compact implementation of SGLang designed to demystify modern LLM serving systems. Educational yet production-quality with RadixAttention, continuous batching, and speculative decoding. MIT licensed.

unknown unknownactivity unknown