is HX370 supported by rocm 7.2

[dusk igloo]

is HX370 supported by rocm 7.2

[magic cairn]

Yes, the AMD Ryzen™ AI 9 HX 370 (part of the Ryzen AI 300 Series) is supported in ROCm 7.2.0. Our latest release expands platform support to include select Ryzen AI processors for local AI development and inference. Here are the specific details regarding support for this hardware: [rocm.docs...]

• Supported Frameworks: Currently, support is focused on PyTorch. [rocm.docs...]
• Operating Systems:
• Windows: PyTorch is supported on AMD Ryzen AI processors. [rocm.docs...]
• Linux: Initial PyTorch support is available as a technology preview. [rocm.docs...]

[magic cairn]

• Memory: These APUs can utilize up to 128GB of shared memory, making them efficient for developing machine learning workflows locally. [rocm.docs...]

For the best experience on Windows, we recommend using the AMD Software: Adrenalin Edition 26.1.1 driver. [rocm.docs...]

[magic cairn]

Did this help? Let us know below 👇
Use /ask to continue the conversation — I'll pick up where we left off

[magic cairn]

@tawny snow asked: "what is rocm"

📄 Your question has 3 messages below. Please check all!

[magic cairn]

[Message 1/3]
ROCm is our open-source software stack that provides the tools for programming AMD Graphics Processing Units (GPUs), from low-level kernels to high-level end-user applications. It is an open, modular, and high‑performance GPU software ecosystem optimized for consistent, scalable performance across data centers, workstations, and edge devices. [rocm.docs...] [rocm.docs...]

Our platform supports various programming models, including HIP (Heterogeneous-computing Interface for Portability), OpenCL, and OpenMP. The following image illustrates our software stack and the technologies it enables: [rocm.docs...]

_images/rocm-software-stack-7_0_0.jpg

AMD's ROCm software stack and enabling technologies.

ROCm includes a comprehensive set of components to support your development:

[magic cairn]

[Message 2/3]

• Libraries: Optimized math libraries (like rocBLAS and hipFFT), machine learning frameworks (PyTorch, JAX, TensorFlow), and communication primitives (RCCL). [rocm.docs...] [rocm.docs...] [rocm.docs...] [rocm.docs...]
• Tools: Powerful development tools such as the ROCm Debugger (ROCgdb), ROCProfiler, and system management interfaces like AMD SMI. [rocm.docs...] [rocm.docs...] [rocm.docs...]
• Compilers and Runtimes: Our LLVM-based compiler infrastructure and runtimes like ROCR-Runtime. [rocm.docs...] [rocm.docs...]

[magic cairn]

[Message 3/3]
By using ROCm, you can accelerate AI workloads—including training, fine-tuning, and inference—while benefiting from a seamless migration path that allows you to develop applications locally on Radeon GPUs and deploy them at scale on AMD Instinct accelerators. I am currently trained on ROCm version 7.2.0, which is our current production release. [rocm.docs...] [rocm.docs...]

References:
• What is ROCm? — ROCm Documentation
• AMD ROCm 7.12.0 preview — AMD ROCm 7.12.0 preview
• Use ROCm for AI — ROCm Documentation