In general you’re mem bandwidth constrained so cpu vs gpu often ends up similar on APUs
replyThere are ways to trade off compute power for memory bandwidth (like MTP and other speculative decoding approaches). The CPU and GPU would need to be able to share the same cache for this to work. In the Strix Halo case the GPU has a private cache on the GPU die I think, which is the snag.
replyIf you get the inference engine to route the heavy matrix math to the GPU and the speculative drafting to the CPU without choking on latency it's probably gonna be very fast.
replyWould love to see the benchmarks if someone actually pulls something like that off.