On June 20, TapTechNews reported that AMD officially launched the latest ROCm (Radeon Open Compute) platform and announced support for multi-GPU clusters, such as dual RX7900XTX or W7900 graphics cards, and 2 or 4 PROW7900 dual-slot workstation graphics cards.
In addition, ROCm also supports more RDNA3 graphics cards, including the Radeon RX7900GRE, and the newly launched PROW7900 dual-slot workstation graphics card this month, the latter priced at $3499 (TapTechNews note: currently about 25,433 RMB).
In addition, the most important point of the new ROCm also lies in the increased support for WSL2 (Windows Subsystem for Linux), so that various Linux-based AI tools can be initially supported on the Windows system.
AMD also announced that the TensorFlow framework has been ROCm certified. Along with PyTorch and ONNX, this means that the ROCm library has been significantly improved for the most popular AI acceleration platforms.
This major function enhancement of this release focuses on improving compatibility, accessibility and scalability, including:
Multi-GPU support: Allows for building AI workstations based on multiple graphics cards, suitable for multi-service and multi-user solutions.
Beta support for Microsoft WSL2: Allows WSL to work with ROCm to run AI programs on Windows.
TensorFlow framework support: Provides more choices for AI developers.
AMD Radeon Software for Linux with ROCm 6.1.3 is compatible with the following AMD products:
AMD Radeon RX7900XTX
AMD Radeon RX7900XT
AMD Radeon RX7900GRE
AMD Radeon PRO W7900
AMD Radeon PRO W7900 Dual-Slot
AMD Radeon PRO W7800
Ubuntu 22.04.3 HWE