Old AirPort Extreme time capsule

I was just curious if I could use my time capsule as an AI extension. Basically I’d like to load up llama and then have it use the capsule as a wife based AI for the house. I’m just not sure how to set it all up.


I plan to hardwire it into my network, I use star link at the farm - then loading an AI llm onto it with agents and having it kind of run some small things plus keep track of things. I know it may be easier to buy somethings else to set it up but I already have this and thought- well it’s a Terabyte of hard drive space ¯\_(ツ)_/¯ and its here in the shelf……

Posted on Nov 2, 2024 2:15 PM

Reply
Question marked as Top-ranking reply

Posted on Nov 2, 2024 8:56 PM

Any recent Raspberry Pi will seriously outrun the embedded processor* found in a Time Capsule, and the Pi with far better development support, connectivity, and documentation.


Particularly given the TC has no support and no available documentation for your proposed endeavor, and nowhere near the selection of available mounts and cases, HATs, shields, displays, GPIO, and other such. Like, TC has none of that. (Info)


Not only will you need to port your target OS and toolchain onto the TC, and integrate whatever expansions and GPIO connections you might want or need, you’ll also need to contend with the lack of compute resources for the ML or LLM algorithms you plan to run.


*The ARMv5 Marvell Ferocean SoC was a common choice for TC, though there may well be other SoCs used. That’s really old, really slow, and limited. (ARMv8 and ARMv9 designs are common choices now, and can be rather better equipped for running ML, or LLMs)

5 replies
Question marked as Top-ranking reply

Nov 2, 2024 8:56 PM in response to LuckyLine

Any recent Raspberry Pi will seriously outrun the embedded processor* found in a Time Capsule, and the Pi with far better development support, connectivity, and documentation.


Particularly given the TC has no support and no available documentation for your proposed endeavor, and nowhere near the selection of available mounts and cases, HATs, shields, displays, GPIO, and other such. Like, TC has none of that. (Info)


Not only will you need to port your target OS and toolchain onto the TC, and integrate whatever expansions and GPIO connections you might want or need, you’ll also need to contend with the lack of compute resources for the ML or LLM algorithms you plan to run.


*The ARMv5 Marvell Ferocean SoC was a common choice for TC, though there may well be other SoCs used. That’s really old, really slow, and limited. (ARMv8 and ARMv9 designs are common choices now, and can be rather better equipped for running ML, or LLMs)

This thread has been closed by the system or the community team. You may vote for any posts you find helpful, or search the Community for additional answers.

Old AirPort Extreme time capsule

Welcome to Apple Support Community
A forum where Apple customers help each other with their products. Get started with your Apple Account.