[ Gpu ]

Discover insights, tutorials, and thoughts on technology, homelab, and development.

Dec 16, 2025 5 min read

Writing High-Performance AI Kernels in Mojo: 10x Faster Than PyTorch

How I built custom Mojo kernels for AI research that outperform PyTorch on consumer hardware

mojo performance simd

system-status.sh

root@lab-stack:~# uptime

up 247 days, 18:42, 1 user, load average: 0.52, 0.58, 0.59

root@lab-stack:~# systemctl status lab-stack.target

� lab-stack.target - Lab Stack Services
Loaded: loaded (/etc/systemd/system/lab-stack.target)
Active: active since 2026-01-08

root@lab-stack:~# _

Last build: 2026-01-08 | Status: Operational

Install Lab Stack Add to home screen for offline access

Offline

Lab Stack Assistant

Local

Hi! I'm the Lab Stack assistant. I can help you find information about this homelab, search blog posts, or answer questions about the infrastructure.

Try asking: "What services are running?" or "Find posts about Docker"

Ready Privacy-first

Claude Code Assistant

Connecting to server...

PWA & Cache

Cache Status

SW Version: --

Caches: --

Status: --

Topic

Style Length

Hi! I'm connected to Claude Code with full access to your homelab infrastructure.

Use the tabs above to access agents, commands, and services. Or just ask me anything!

Ready Session: --