← Feed

Posts

• Post #1439389

  My #SelfHosted LLM machine is ready. Two P40's with a total of 48GB VRAM allow for running 70b parameter models. #HomeLab

• Post #1439388

  @chad dual Nvidia Tesla P40’s that allow for running models up to 70b params. Check my feed for a pic if you are interested.

• Post #1439387

  @ceresbzns So far I am running Ollama with the Ollama WebUI and just having a fun time testing out different models. I haven't started using it for anything specific yet other than messing around but I do intend to start integrating it into things which is why I set this server up in the first place. Coding integration, integration with search and so on. @technotim provided me with some inspiration!

• Post #1439385

  Added a 2.5Gbit NIC to my Lenovo ThinkCentre M920q Tiny 1L box that currently is a Proxmox node. The NIC is a M.2 A+E key based, Intel i226 unit for under $25 so I bought one for test. First picture shows the cover off and in the lower right corner you see where the port can go, next to the VGA output. Second pic shows the SSD removed and the metal plate for the eth port removed. The M.2 port is further up in the picture. Pic 3, the little M.2 card is installed and the eth port is screwed to th...

• Post #1439384

  @pair12 Yeah, so the size of it is mostly for connectors like VGA, HDMI and DP ones, but as you can also see , the RJ45 did fit, even though it is a snug fit from a vertical perspective. Snug but doable.

• Post #1439383

  Enjoying Datadog #Dash2025!