Find tools › Open-source projects › Write code faster

DeepGEMM

Open-source project · by deepseek-ai

A high-performance library for accelerating AI matrix math on NVIDIA GPUs.

Some setupWell known

What it costs

Free and open source; you may pay for the AI model it uses.

Use it to

  • Accelerate tensor core operations on NVIDIA GPUs
  • Run FP8 and FP4 large language model computations

How to start

  1. Clone the repository with submodules using git clone --recursive git@github.com:deepseek-ai/DeepGEMM.git.

  2. Navigate into the DeepGEMM directory.

  3. Run the development build script via ./develop.sh.

Good to know. Requires an NVIDIA SM90 or SM100 architecture GPU (such as H800) and specific CUDA toolkit versions.

Before you install

  • Found automaticallyNot reviewed by a person yet. Check it before installing.
  • License: MITYou are allowed to use it under these terms.
  • Last updated 2026-10-20

More detail

DeepGEMM is a high-performance tensor core kernel library providing computation primitives for large language models on NVIDIA GPUs.

It optimizes matrix multiplication and mixture-of-experts operations to accelerate large language model training and inference on modern hardware.

Similar tools

Descriptions are written in plain English from each project's own documentation and refreshed by our agent; this one was last checked 2026-10-06. Spotted a mistake? · Open in Find tools · © 2026 Which AI Map