AMD has recently announced the release of ROCm 6.1.3, the latest version of its open-source GPU compute stack. This new update brings significant improvements, including enhanced multi-GPU support, compatibility with up to four Radeon RX or Radeon PRO graphics cards, and optimization for AI workloads.

One of the key features of ROCm 6.1.3 is its improved multi-GPU support, which is specifically optimized for AI tasks. By leveraging the AMDGPU and AMDKFD kernel drivers, this update aims to enhance the performance of workloads across multiple GPUs simultaneously, with a focus on benefiting AI inference workloads.

In addition to multi-GPU support, AMD has also introduced beta-level support for the Windows Subsystem for Linux (WSL2) in ROCm 6.1.3. This new feature allows users to utilize ROCm on Windows 11, opening up new possibilities for GPU compute on different operating systems.

Another highlight of the ROCm 6.1.3 release is the TensorFlow Framework qualification support. AMD has officially added support for the new Radeon PRO W7900 Dual Slot graphics card with this update, further expanding the range of compatible hardware.

Despite being a point release, ROCm 6.1.3 is a significant update that follows closely on the heels of ROCm 6.1.2. However, it is worth noting that while Ubuntu 22.04 LTS is the primary Linux support target, there is currently no official support for Ubuntu 24.04 LTS.

The announcement of ROCm 6.1.3 was made on community.amd.com, although the software release on GitHub is still pointing to v6.1.2 as the latest version. It is expected that the official release of ROCm 6.1.3 will be available soon, possibly within the week.

Related News:

• Fedora 41 Aims To Ship AMD SEV-SNP Confidential Virtualization Host Support
• New Linux Change Helps Ensure AMD Ryzen With NVMe Works After Resuming From Suspend
• AMD Fast CPPC To Be Merged For Linux 6.11
• AMD P-State Linux Patches Updated For Heterogeneous CPUs
• AMD Posts New Linux Patches For Per-Core CPU Energy Counters
• AMD’s Newest Open-Source Surprise: ‘Peano’ – An LLVM Compiler For Ryzen AI NPUs