AMD Preparing ROCm 6.1 For Release With New Features

It looks like AMD will soon be announcing the ROCm 6.1 update to its open-source GPU compute stack.

ROCm 6.0 debuted back in December with official support for the AMD Instinct MI300A/MI300X, more AI capabilities, and other improvements as it aims to better compete with NVIDIA's CUDA. ROCm 6.1 should be released soon as the next step in the GPU compute battle.

There has been a number of ROCm 6.1 related commits hitting AMD's public GitHub repositories in recent days, pointing to what's likely to be an imminent release.

For MIPOpen 3.1 with ROCm 6.1 there's been many additions including new solvers, an AI-based parameter prediction model for the conv_hip_igemm_group_fwd_xdlops solver, numerous fixes, and other updates.

AMD MIGraphX will see an important update with ROCm 6.1. For the next ROCm release, MIGraphX 2.9 brings FP8 support, support for more operators, documentation examples for Whisper / Llama-2 / Stable Diffusion 2.1, new ONNX examples, BLAS auto-tuning for GEMMs, and initial code for MIGraphX running on Microsoft Windows.

AMD's HIP for ROCm 6.1 is rolling out a "hipother" repository for holding HIP back-end implementations for non-AMD platforms. HIP is AMD's C++ Runtime API and kernel language for portable AMD/NVIDIA GPU apps and with the HIPIFY tools to help in converting CUDA sources to HIP.

HIPIFY for ROCm 6.1 meanwhile is adding CUDA 12.3.2 support, full hipSOLVER support, full rocSPARSE support, and other improvements.

Given the recent uptick in ROCm 6.1 mentioning Git commits and especially with change-logs/documentation being updated, look for ROCm 6.1 to be releasing soon.