AI and a Host


The project Goals

We were gifted an Nvidia Telsa K80. With 4995 Nvidia Cuda Cores and 24GB of GDDR5 memory, it is quite the card and can give us the ability to start some AI work in our lab. The difficulty is making it work with our equipment.

We first installed it into one of our newer systems, used primarily for gaming. We thought that the extra power would offer us extra performance. We had planned this as an opprotunity to explore the Windows Subsystem for Linux and that it could host Open-Webui with a Docker instance. We ran into trouble with Ollama‘s Windows support for the card

We needed a Linux system that we could install the card into, and then compile support for the older CUDA cores into Ollama.

We thought this could become a solid series of updates



Project Updates

  • Install Docker on Linux Mint

    While Linux Mint is one of the leanest, most efficient, highest performing distributions we’ve worked with, it does feature an awkward quirk in that while Linux Mint is Ubuntu based, and each version features a different “Code Name” Linux Mint features a different “Code Name” than the current Ubuntu Linux. As such, when we attempt…

  • Setup Linux Mint for SSH and Microsoft Remote Desktop Connection

    Update apt with You will be prompted to enter the user with sudo privileges the first time sudo is invoked The following set of commands will install, verify that the openssh server is running, and ensure access through the ( UFW ) firewall The following set of commands will install xrdp and xrdp for the…

  • Installing Linux Mint

    Ai and a Host We are recycling a system that was built for Bitcoin mining, named Moria, that was built with an ASUS B250 Mining Motherboard with a 7th Generation Celeron Processor, and 16GB of RAM It was later recycled as a file server system running FreeBSD Preparing the installation Media We downloaded a fresh…

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *