Everything seems to be very Cuda centric, has anyone tried training or other ml projects using AMD GPUs? How was it?

  • lzbz@programming.dev
    link
    fedilink
    arrow-up
    2
    ·
    1 year ago

    I used ROCm on my Vega 64 for my bachelor thesis. Once I got it to run, it was a huge improvemnt over training on CPU in terms of time. Didn’t run into any problems, after the setup. Support for ROCm was very hit andmiss last time I checked though, so ymmv.