design

Speed AI with aiter

Name: Speed AI with aiter
Author: ROCm

Get accelerated AI with aiter, a tensor engine for founders using ROCm.

477 stars386 forksPythonQuality 8/10Updated 7/5/2026100% free · open source

What it does

Aiter accelerates AI computations by providing a tensor engine optimized for the ROCm platform, allowing founders to run AI workloads on AMD GPUs

Install / run

git clone https://github.com/ROCm/aiter.git && cd aiter

When to use it

•When you need to deploy AI models on AMD hardware
•When you're working with large tensors and need optimized performance
•When you're looking for an alternative to NVIDIA CUDA for AI computations

Quick start

1Build the aiter library by running the command `mkdir build && cd build && cmake .. && make`
2Use the `aiter_example` application in the `examples` directory to test the installation
3Modify the `CMakeLists.txt` file to customize the build configuration for your specific use case
4Run the `aiter_benchmark` application to evaluate the performance of aiter on your system
5Integrate aiter into your Python application using the provided Python bindings

Ready-to-paste prompt

python examples/aiter_example.py --rocmlib /path/to/rocmlib --input-size 1024 --batch-size 32

Heads up: Aiter requires a compatible version of the ROCm platform to be installed on your system, and you need to ensure that the ROCm version is compatible with your AMD GPU

Saves to your device

What's inside — free to inspect

No purchase needed

Read the entire source before you build — unlike paid marketplaces that hide it behind a buy button.

top-level files

folders

348.2M

repo size

MIT

license

Key files

README.md