Author: jeffstokes2014

Architecting a Local, Headless AI Inference Lab

When engineering a localized, bare-metal AI lab environment, combining cloud-first container ecosystems with highly specific on-premises hardware introduces unique structural friction. In this guide, I document the deployment of a unified text-and-image inference stack utilizing OpenWebUI and ComfyUI on a headless Ubuntu Server LTS platform. Crucially, I trace the end-to-end journey of isolating container execution… Continue reading Architecting a Local, Headless AI Inference Lab

VectorDB tuning for ZFS

Part 2: Storage Layer Optimization — How I Made ZFS Play Nice with My VectorDB In the world of Retrieval-Augmented Generation (RAG), I’ve learned that while my GPU handles the “thinking,” my filesystem handles the retrieval. Even though my ZFS RAIDZ1 array is built for high-speed sequential throughput, I realized the default “one size fits… Continue reading VectorDB tuning for ZFS

AMD User Experience Proxy and handle leaks

high handles in AUEPMaster

This is an odd fellow here. As you can see in the image below, AUEPMaster has 157k handles. Yesterday it had over 1 million but I failed to capture a screenshot before reboot. This hasn’t impacted game performance as far as I can tell, just an annoyance. Wondering if it’s actually working as intended, does… Continue reading AMD User Experience Proxy and handle leaks

Win11 and Edge – I zip bombed myself

Scenario: I had the pleasure recently of downloading two memory dumps on a Windows 11 machine using normal channel Edge. The ‘problem’ was that these two zip files (both 11 GB in size) were compressed memory dumps (uncompressed size 110GB and 153GB respectively). I started downloading both at the same time, not thinking of it,… Continue reading Win11 and Edge – I zip bombed myself

Exit mobile version