Hacker News: i386

New comment by i386 in "We finally learned to center a div, then browsers added sidebars"

i386 — Wed, 05 Aug 2026 10:35:51 +0000

Its an entire misunderstanding of how the layout system works.

New comment by i386 in "BMW Is Showing Commercials on Their Car's Dash Screens as a Treat"

i386 — Sat, 01 Aug 2026 20:23:50 +0000

Most people working at the top of these large organizations don’t know how to make anything net new anymore. They pushed all the people that do out when padding their nests.

New comment by i386 in "Mesh LLM: distributed AI computing on iroh"

i386 — Sun, 12 Jul 2026 20:56:40 +0000

exo is Mac only.

New comment by i386 in "Mesh LLM: distributed AI computing on iroh"

i386 — Sun, 12 Jul 2026 09:25:15 +0000

I’d love a bug report - we can get it working for you!

New comment by i386 in "Mesh LLM: distributed AI computing on iroh"

i386 — Sun, 12 Jul 2026 09:24:20 +0000

Our skippy library is a patch queue on top of llama that allows us to access internal information, such as activations, and filter tensors on model load.

New comment by i386 in "Mesh LLM: distributed AI computing on iroh"

i386 — Sun, 12 Jul 2026 02:54:22 +0000

To be honest, both are very tough problems we don't have a good answer for yet. If that is something that concerns you, look into building a private mesh with trusted peers.

New comment by i386 in "Mesh LLM: distributed AI computing on iroh"

i386 — Sun, 12 Jul 2026 02:52:14 +0000

Each stage has its own KV for the layers it hosts. You are on the money there, when one stage is waiting it's free for more parallelism. I am planning on exploiting this for more token verification through ngram spec decoding.

New comment by i386 in "Mesh LLM: distributed AI computing on iroh"

i386 — Sun, 12 Jul 2026 02:39:22 +0000

It can dynamically route. If a machine drops out of split, the topology is recalculated and the request is automatically retried.

New comment by i386 in "Mesh LLM: distributed AI computing on iroh"

i386 — Sun, 12 Jul 2026 01:22:45 +0000

Thank you! AMD is a weak spot in our testing right now. If you’re willing to contribute or let us borrow some compute time, drop in on the Discord.

New comment by i386 in "Mesh LLM: distributed AI computing on iroh"

i386 — Sun, 12 Jul 2026 01:12:44 +0000

The lab features two Mac Studios: an Apple M3 Ultra (32 CPU cores, 80 GPU cores, 256 GB unified memory) and an Apple M1 Ultra (20 CPU cores, 48 GPU cores, 128 GB unified memory), both connected via 1Gbit Ethernet.

We use a customized Q2 quantization that preserves sensitive tensors at Q8.

To reduce compute time per layer, we are developing a custom GLM DSA Metal graph.

While we are not yet approaching MTP, we plan to port our existing MTP implementations from versions 4.7 and 5.1 to 5.2.

Since GLM's MTP acceptance rate is very high for a single predicted token, we are exploring token prediction techniques to widen the predicted tokens and utilize parallelism for verification.

New comment by i386 in "Mesh LLM: distributed AI computing on iroh"

i386 — Sun, 12 Jul 2026 00:38:03 +0000

That isn’t true. llama RPC is incredibly slow but staged splits in skippy are orders of magnitude faster.

New comment by i386 in "Mesh LLM: distributed AI computing on iroh"

i386 — Sun, 12 Jul 2026 00:36:37 +0000

This was done on my home lab simulating 5ms latency and jitter between machines. Splits work quite well if you your nodes are over WAN at metro latency’s but not super fast on global WAN.

The idea is that you could take several machines without dedicated RDMA or NVLINK fabric and use them to serve a large model on hardware you own then share it with others.

I’m currently working on GLM 5.2 on my lab environment with around 10 tok/s on the same split.

New comment by i386 in "Mesh LLM: distributed AI computing on iroh"

i386 — Sun, 12 Jul 2026 00:21:48 +0000

I’m one of the contributors to Mesh LLM and happy to answer any questions. I authored the skippy engine that allows you to split large models across nodes.

New comment by i386 in "VS Code inserting 'Co-Authored-by Copilot' into commits regardless of usage"

i386 — Sun, 03 May 2026 00:09:49 +0000

PMs at Microsoft have incredibly bad taste

New comment by i386 in "I bought Friendster for $30k – Here's what I'm doing with it"

i386 — Mon, 27 Apr 2026 13:17:32 +0000

I kinda fucking love it. He’s doing something. It’s interesting. There’s no scam. Let live!

New comment by i386 in "Google releases Gemma 4 open models"

i386 — Fri, 03 Apr 2026 02:18:20 +0000

You can try this new model live using mesh-llm right now: https://www.anarchai.org/dashboard

New comment by i386 in "Signing data structures the wrong way"

i386 — Thu, 02 Apr 2026 13:13:46 +0000

switch on version, then messageId…

New comment by i386 in "Signing data structures the wrong way"

i386 — Thu, 02 Apr 2026 13:12:04 +0000

> I didn't know that Protobuf wasn't canonical but even without this knowledge, there are many other factors which make it an inferior format to JSON.

If strict typing and versioning are useless to you, sure.

Pool spare GPU capacity to run LLMs at larger scale

i386 — Tue, 24 Mar 2026 05:57:00 +0000

Article URL: https://github.com/michaelneale/mesh-llm

Comments URL: https://news.ycombinator.com/item?id=47499033

Points: 11

# Comments: 3

New comment by i386 in "Apple's latest attempt to launch the new Siri runs into snags"

i386 — Thu, 12 Feb 2026 21:08:15 +0000

yeah this product does not exist