https://jonready.com/blog/feed.xml
Thoughts on building, exploring, and technology.
Aug 24, 2026
vllm-ios: 88% Faster Multi-Agent Inference on iOS
You get about 15-20 seconds of GPU per minute before an iPhone overheats, so every second has to do maximum work. Batched inference on iOS is surprisingly bad at that, so I rewrote vLLM's continuous
None found.
If you are the site owner, you can add <link rel="me" ...> elements to link to your other sites or profile pages.
These will appear after the next crawl.
None found.
Web feeds (RSS feeds or Atom feeds) will appear here. If you are the site owner you can add an OPML blogroll to recommend other blogs or websites.
None found.
If you'd like to promote this feed you can add it to your OPML blogroll. You'll appear here as a subscriber after the next crawl if your site is part of the network.
This feed scored 17 point in our rating.
A feed can increase its score (and appear higher on our list) by implementing an OPML blogroll, adding a title, description, categories, and more.
cats - 0See criteria
description - 3
feedlangs - 1
hascontent - 0
hasposts - 3
postcats - 0
promoted - 5
promotes - 0
relme - 0
title - 3
website - 2