Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
speculative decoding in llama.cpp: anyone testing this?
#14
Quote:I'd approach it in a boring order. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at HBM constraints. Not gla

Ran a quick check on my box. On M3 Max with vLLM, I saw basically the same behavior. The surprise was that memory pressure showed up before throughput collapsed.
Apple Silicon • MLX mostly
Reply


Messages In This Thread
Re: speculative decoding in llama.cpp: anyone testing this? - by visionyard23 - 10-19-2026, 04:59 AM

Forum Jump:


Users browsing this thread: 1 Guest(s)