rdna

The rDNA 2 dGPU support patch kext.

RDNA-native LLM inference engine in Rust.

RadeonTuner is an easy to use alternative for the AMD Adrenalin Software for users that just want the basics and use the driver only software install type.

Windows-only version of ComfyUI which uses AMD's official ROCm and PyTorch libraries to get better performance with AMD GPUs. [auto-installation and popular performance enhancing packages like triton * sage-attention * flash-attention * bitsandbytes included ]

Utilities for accessing AMD's Machine-Readable GPU ISA Specifications.

AMPtk: Amplicon ToolKit for NGS data (formally UFITS)

A working Comfy UI with AMD GPUs. Works with RDNA 3/3.5/4

Reclaiming the ASRock BC-250 as a real, tunable PC - Cachenetics' open-source firmware liberation + tuning suite.

Human and mouse genome assemblies customized for rDNA mapping

Windows platform manager for ephemeral ComfyUI+ROCm instances optimized for AMD GPUs via PowerShell.

AMD RDNA 7000/9000 GPU Passthrough

Deploy AI Models On Any Hardware Train, fine-tune, convert, quantize, serve, and benchmark models for NPU, TPU, GPU, and CPU. One toolkit — every accelerator. OpenAI-compatible API included.

This repository contains 18S reference sets that can be used with the RDP Classifier to classify 18S rDNA metabarcode sequences.

A high-performance, Ring-3 system agent for AMD RDNA architectures designed to mitigate DPC latency spikes and dynamically optimize GPU thermal limits for pristine frame pacing.

Reproducible RDNA reference for Meta Muse-Glimmer-30B — adapting MI-series ROCm recipes to Ryzen AI and Radeon with vLLM, llama.cpp, DFlash and auditable benchmarks.

A machine learning pipeline for comparative analysis of 45S rRNA genes from plant NOR regions

Pipeline to process RNAseq and ChIPseq data. Outputs include gene counts, transcripts per million, alternate splicing events, SNP calls, and bigwigs. Supports both Docker and Singularity for all dependencies.

Run JAX on AMD GPU — Windows DirectML + Linux/WSL2 ROCm. The missing setup guide for RX 5000/6000/7000 series. vmap, grad, jit tested on real hardware.

Native Windows AMD multi-GPU bridge for vLLM and llama.cpp/ROCmFPX via Direct-RCCL, D3D12, and hybrid routing

RNS8 explores exact integer matrix multiplication on AMD GPU matrix engines.