I patched llama.cpp to gain 20% prompt processing TPS. Help me make a PR
6 points - last Saturday at 7:03 PM
I've been running Qwen3.6-35B-A3B locally on llama.cpp and noticed that prompt processing throughput gets too low with MTP. I got nerd-sniped.
What started as curiosity turned into a two-week rabbit hole of experiments and ended with a PoC that fully recovers the MTP PP overhead on GPU, above any expectation I had.
TL;DR: instead of processing the last layer MoE FFN for the entire ubatch tokens (usually 512-2048 tokens), this PoC processes only the output row (usually 1 token during prefill). The result is PP TPS is back to the same as with MTP disabled, keeping most of MTP's benefits to TG TPS, even with a slight drop in draft acceptance rate in one of the benchs.
I'm not opening a PR to llama.cpp because this is AI-generated code, which goes against their contribution policy, which I support. If you know C++ and llama.cpp internals, I invite to work together with me to open a PR with a more mature implementation.
potus_kushner
last Saturday at 7:25 PM
you have better chances opening an issue and explaining your findings. lots of people watch the repo so probably someone will chime in.
i_am_rocoe
last Saturday at 7:34 PM
Thank you!
i_am_rocoe
last Saturday at 7:14 PM
[dead]