Tiny Core Linux

General TC => Programming & Scripting - Unofficial => Topic started by: james.morowin on March 31, 2026, 08:10:33 AM

Title: Possibility of using AI agents to code in Tiny Core Linux
Post by: james.morowin on March 31, 2026, 08:10:33 AM: Hi, I'm James. I wanted to ask the possibility of using local llm AI agents like Ollama, Qwen and many other open source llm AI in this Linux distro (Tiny Core). I want to maximize my RAM capacity (16GB for ram, 8GB for VRAM) on my laptop to only for llm, but still with some level of GUI during the work session and in my opinion this distro fit all the criteria which it has the least RAM usage while also still modern enough to be connected to the internet. Any help is appreciated.

Best regards. :)
Title: Re: Possibility of using AI agents to code in Tiny Core Linux
Post by: Juanito on March 31, 2026, 09:03:48 AM: What hardware is providing the vram? Nvidia is problematic due to its proprietary nature.
Title: Re: Possibility of using AI agents to code in Tiny Core Linux
Post by: james.morowin on March 31, 2026, 11:15:38 PM: Quote from: Juanito on March 31, 2026, 09:03:48 AM
What hardware is providing the vram? Nvidia is problematic due to its proprietary nature.

It is quite indeed Nvidia, unfortunately ;D
My systems has i7-14650HX, RTX 4060 8GB, 16 GB LPDDR5 5600 MT/s. I'm still using Windows 11 for my OS but it still got bloated, it takes 13.1 Gb/s out of 15.7 GB available RAM hence the reason why I want to migrate to Linux with the least amount of resources to run. realistically I could only use 5-6 GB of RAM and 6 GB of VRAM for my AI agents in Windows 11
Title: Re: Possibility of using AI agents to code in Tiny Core Linux
Post by: Juanito on April 01, 2026, 04:50:26 AM: Using CorePure64 you'll certainly have more ram available as compared to windows, but you would probably need to compile an Nvidia driver locally to access the vram.

As an example, booting my old intel laptop with 16gb ram and loading graphics-KERNEL, Xorg-7.7-3d, flwm, aterm, wbar, beaver and emelfm - after starting the gui and opening aterm, beaver and emelfm, "free" shows 15,803,744 bytes available.
Title: Re: Possibility of using AI agents to code in Tiny Core Linux
Post by: james.morowin on April 01, 2026, 06:59:06 AM: Quote from: Juanito on April 01, 2026, 04:50:26 AM
Using CorePure64 you'll certainly have more ram available as compared to windows, but you would probably need to compile an Nvidia driver locally to access the vram.

As an example, booting my old intel laptop with 16gb ram and loading graphics-KERNEL, Xorg-7.7-3d, flwm, aterm, wbar, beaver and emelfm - after starting the gui and opening aterm, beaver and emelfm, "free" shows 15,803,744 bytes available.

I see. I'll look into it more. thanks a lot for the help ;D