Jlama (Java) outperforms llama.cpp in F32 Llama 7B Model

version_five2y ago· 3 in thread

Where does the performance difference come from? And in what kind of processor & gpu? I didn't even know llama.cpp had a 32 bit option. For now I'm pretty suspicious it's a fair comparison.

tjakeOP2y ago

The default for `convert.py` is F32. This is just SIMD CPU comparison.

Jlama uses the vector api in java20 but also better thread scheduling with work stealing and zero allocation.

belfthrow2y ago

Could you link to some of the examples in your repo where you enforce the zero allocation? I don't see much reuse of the buffers, eg float buffers and there is quite a lot of array based heap allocation. Just for my own interest. Many thanks. Cool to see the use of the new vector api also.

version_five2y ago

Very interesting, I'll watch for the quantized version.

syllogistic2y ago· 2 in thread

Huh, yeah it repros. Java is faster 159s vs 203s for the 256 tokens on my intel i9 12 gen

brucethemoose22y ago

> 159s for the 256

This is still extremly slow for that CPU, compared to the quantized model.

IIRC the llama.cpp f32 code is basically a placeholder.

BUT the threading overhead is a known performance issue, and I'm sure Java handles that better.

version_five2y ago

> threading overhead is a known performance issue

I didn't know about it, I should have... are there any "edge" frameworks as complete as ggml/llama.cpp that you know of that are faster now? Ggml is still very easy to use which I like, but I'd always thought of it as the fastest, in particular for CPU, I hadn't noticed there were known performance issues.

1 more reply

tjakeOP2y ago

GH: https://github.com/tjake/Jlama

Jlama (Java) outperforms llama.cpp in F32 Llama 7B Model (opens in new tab)

Jlama (Java) outperforms llama.cpp in F32 Llama 7B Model (opens in new tab)

9 comments

9 comments