http://franfabrizio.dev/tags/tesla-p40/ Recent content in tesla-p40 on franfabrizio.dev Hugo en-us Wed, 03 Jun 2026 22:00:13 -0500 http://franfabrizio.dev/posts/am4-tesla-p40-buildout/ Wed, 03 Jun 2026 22:00:13 -0500 http://franfabrizio.dev/posts/am4-tesla-p40-buildout/ <p>This is Part 2 in my series about building a cheap VRAM AI machine with used Tesla P40 GPUs. If you haven&rsquo;t already, check out <a href="http://franfabrizio.dev/posts/tesla-p40-ai-server-project/">Part 1: When Cheap VRAM Gets Complicated</a> for the backstory on why I needed P40s, and optionally, <a href="http://franfabrizio.dev/posts/troubleshooting-tesla-p40-dell-precision-t7820/">the troubleshooting post</a> for the technical details of the dead end that led to this build.</p> <p>In the first post, I explained why I went chasing cheap VRAM in the first place. The short version: local LLMs love VRAM, modern high-VRAM GPUs are painfully expensive, and used Tesla P40s offer 24GB of CUDA-compatible VRAM for a price that justifies using 10-year-old datacenter hardware for this project.</p>