Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

nneomindryan about 2 hours ago 22 commentsRead Article on neomindlabs.com

⚡ Community Insights

Discussion Sentiment

50% Positive

Analyzed from 529 words in the discussion.

Discussion (22 Comments)Read Original on HackerNews

dwa3592•6 minutes ago

I have a prediction. By the mid of 2027, we will have >200B MoE models running on basic consumer hardware.

I am running Qwen3.6-35B-A3B locally on my 16GB mac with 7-9 tokens/second. Link - https://github.com/deepanwadhwa/samosa-chat

This is a GPT4 level model running locally on a decent speed.

embedding-shape•3 minutes ago

> I am running Qwen3.6-35B-A3B locally on my 16GB mac with 7-9 tokens/second. Link - https://github.com/deepanwadhwa/samosa-chat

It's quite telling you didn't use Qwen3.6-35B-A3B locally to build that, seems there was another collaborator ;)

Show something you've built with the model+tooling instead, truly dogfood it. I'm sure you'll discover things along the way too!

hparadiz•about 1 hour ago

Here's my report running several different models on a dual Xeon with 256 GB of DDR4 and no GPU.

https://gist.github.com/hparadiz/f3596d00a62d8ebb2dadcc46ee5...

puzzlingcaptcha•29 minutes ago

Have you tried with a single CPU to get rid of the NUMA penalty? I understand this likely means halving the memory but I am interested in how much of a difference it makes

trollbridge•23 minutes ago

I have (192GB machine with two CPUs), pretty much does the trick. It just runs some small models used for embedding, etc. and has those on one CPU / memory node and all the Docker containers on the other one.c

neomindryan•about 1 hour ago

Thank you for sharing!

deltamidway•13 minutes ago

He's shown me his set up in his basement. It's sick! Talk about your 3d printer next!

throwaway2027•about 1 hour ago

That's quite slow I'm getting 8-12 t/s on a 13 year old CPU. (Speed varies by context size and other settings who knows)

https://news.ycombinator.com/item?id=48354801

neomindryan•about 1 hour ago

Thank you for sharing / linking!

neomindryan•about 2 hours ago

Author here, it looks like my original comment was flagged for some reason. The fix is open upstream as PR #2138 (https://github.com/ikawrakow/ik_llama.cpp/pull/2138)

robotswantdata•16 minutes ago

Need to run this on my Xeons with AMX

mmastrac•28 minutes ago

Is it just me or does this post not mention how much RAM they had? I would love to know - I have a dual-Xeon 1U screamer with 96GB of DDR4 RDIMM just sitting around...

FWIW I'm getting a hardware max of 20 tok/s (approx topping out the GPU's compute) on my custom local diffusiongemma port running on an M3.

neomindryan•18 minutes ago

hey, I’m the author. That box has 384gb, but loading the model “only” uses about 80gb.

okokwhatever•13 minutes ago

To me context means everything. Tokens per second is a great metric but in the real world context window is the deal breaker when a real use case is on the table.

ChrisArchitect•about 1 hour ago

A 10 year old Xeon is all you need

https://news.ycombinator.com/item?id=48353348

neomindryan•about 2 hours ago

Author here. The short version: a viral post ran Gemma 4 on a 2016 Xeon; my Xeons are 2013, and the fork it used assumes AVX2, which Ivy Bridge doesn't have. The build failure was easy. The fun bug was the silent one: two MoE graph ops with no dispatch case on non-AVX2 builds, so every expert FFN output was uninitialized memory. Deterministic, NaN-free, fluent-looking multilingual gibberish.

The fix is open upstream as PR #2138 (https://github.com/ikawrakow/ik_llama.cpp/pull/2138), awaiting review. Fair warning on the AI angle: the patch was written by Claude at my direction. The post is explicit about which parts were me and which weren't. Happy to answer questions about either the bug or the workflow.

otherjason•about 1 hour ago

This reads as pretty clearly AI-generated text, which is against HN guidelines.

FL410•23 minutes ago

The PR? He said it was AI in the comment you replied to...

I don't think the post itself reads like AI at all, but that's just me.

logicallee•3 minutes ago

I think "this" refers to its parent comment. Part of it sounds like Claude wrote it. AI-generated comments aren't allowed on HN.

pkghost•about 1 hour ago

Here's the thing: life also imitates art. If you invert your load-bearing assumption, it could be that he just reads too much slop. But my honest take? You might be right.

aniwalunj•about 2 hours ago

Truly amazing. This gives a peek into the future for what's possible.

rvba•about 1 hour ago

Sorry for asking here but literally nobody knows:

Android studio connected to a local model disconnects automatiacally after 10 minutes. How set this limit to 12 hours or remove it completely?

I could run my LM studio model all night... but I cant, since Android studio times out after a hard limit of 10M.

This is not related to number of tokens. I do 130 sec.