AMD ROCm 6.1.3 Unleashed: Multi-GPU, WSL, TensorFlow Support, and More

Recently, AMD updated the ROCm software stack, releasing the new iteration, ROCm 6.1.3. This version enhances support for multiple GPUs, allowing up to four Radeon RX/PRO graphics cards to be installed in a single machine. It also adds beta-level support for the Windows Subsystem for Linux and support for the TensorFlow framework. Additionally, the new version supports the Radeon PRO W7900 Dual Slot professional graphics card, unveiled at COMPUTEX 2024.

The most significant update in ROCm 6.1.3 is the enhanced multi-GPU support, enabling users of the ROCm software stack to utilize up to four GPUs, thereby greatly reducing processing times. The graphics cards that support this multi-GPU functionality include: Radeon RX 7900 XTX, RX 7900 XT, RX 7900 GRE, and Radeon PRO W7900, W7800, W7900 Dual Slot. This means users can choose from high-memory professional graphics cards or more cost-effective consumer gaming graphics cards, offering greater flexibility.

AMD stated that the enhanced multi-GPU support aims to build scalable AI desktops for hardware-accelerated AI workflows. GPUs are the cornerstone of high-performance AI, so adding up to four high-performance GPUs in a single workstation will significantly boost the system’s AI performance compared to running a single GPU.

The new Radeon PRO W7900 Dual Slot is optimized for artificial intelligence (AI) applications and is a more compact version of the Radeon PRO W7900, with its thickness reduced to dual slots. It continues the turbo cooling design without compromising performance, allowing users to more easily install four Radeon PRO graphics cards in a single PC system, further enhancing AI performance.