Announcement

Collapse
No announcement yet.

AMD Has Open-Source Ryzen AI Demo Code - But Only For Windows

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • AMD Has Open-Source Ryzen AI Demo Code - But Only For Windows

    Phoronix: AMD Has Open-Source Ryzen AI Demo Code - But Only For Windows

    One of the most interesting aspects of the new AMD Ryzen 7040 series laptop processors is the new "Ryzen AI" capabilities with the new XDNA AI engine capabilities built into the SoC, leveraging IP from their Xilinx acquisition. Linux support details remain scarce but at least one of their (Windows) demos for showcasing Ryzen AI is open-source...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    https://www.phoronix.com/news/AMD-Ryzen-AI-Open-Source-Demo AMD Has Open-Source Ryzen AI Demo Code - But Only For Windows

    Comment


    • #3
      But why Windows when the majority of such loads happen on Linux........................

      Comment


      • #4
        Originally posted by tildearrow View Post
        But why Windows when the majority of such loads happen on Linux........................
        I suspect AMD actual Linux capable personnel is very small and mostly focused on AMDGPU kernel coding which would help explain RoCm horror show and AMDVLK which i assume are done by teams on Windows and then kinda learning Linux enough to migrate it some time later.

        Comment


        • #5
          Somebody please use chatgpt and rewrite the python to be linux only with rust

          Comment


          • #6
            Michael, if you are looking for an "AI benchmark," try running mlc-llm and/or llama.cpp. MLC will run an LLM on highly optimized Vulkan, and Llama.cpp will run it on CPU with SIMD intrinsics (with optional OpenCL offload that is, TBH, not good on IGPs). MLC in particular is comparable to the CUDA kernals I've seen on Nvidia GPUs, which is unusual for Vulkan inference.


            There are other potential candidates, but few are so easy to set up and so "apples to apples" across vendors.
            Last edited by brucethemoose; 04 June 2023, 12:48 AM.

            Comment


            • #7
              Originally posted by tildearrow View Post
              But why Windows when the majority of such loads happen on Linux........................
              This is not necessarily true. If you look at the communities running generative AI locally, most users (and some devs) are running Windows, though WSL is very common.


              AMD is demoing more "Windowsy" things with Ryzen AI anyway, like face detection (face unlock) and object classification (sorting photos).
              Last edited by brucethemoose; 04 June 2023, 12:55 AM.

              Comment


              • #8
                The linux driver is missing, but the backend looks like its "just" Vitis AI running an onnx model.

                If anyone has a Xilinx FPGA, maybe they could change the config and try it out?

                Comment


                • #9
                  Originally posted by tildearrow View Post
                  But why Windows when the majority of such loads happen on Linux........................
                  Simple, they observed the blind devotion that Linux users have for nvidia even when they are hostile towards FOSS, so they are assuming that we respond better when ignored/abused.

                  Comment


                  • #10
                    Meanwhile Google has presented a new JS engine for Chrome called Maglev which will make it a whole lot faster:



                    Michael

                    Maybe mention this on the front page, no?

                    Comment

                    Working...
                    X