Hi - firstily awesome work with OF3 :)
One question - reading your white paper, I see
We ran all OF3p2 assessments on MI300A APUs. The AMD implementation is
based on a version of OF3p2 leveraging Triton kernels introduced in the MegaFold work, which
we adapted for the OF3 inference stack [16]
Is this implementation (or install instructions) available? I would like to run OF3 on Setonix (https://pawsey.org.au/systems/setonix/) which has AMD MI250x available.
George
Hi - firstily awesome work with OF3 :)
One question - reading your white paper, I see
We ran all OF3p2 assessments on MI300A APUs. The AMD implementation is
based on a version of OF3p2 leveraging Triton kernels introduced in the MegaFold work, which
we adapted for the OF3 inference stack [16]
Is this implementation (or install instructions) available? I would like to run OF3 on Setonix (https://pawsey.org.au/systems/setonix/) which has AMD MI250x available.
George