So, I learned that #
SBC #
ComputeModule newest models tend to be "#
MachineLearning optimized". If I am going to try to build a personal equivalent of #
StarTrek #
Tricorder, or #
Fallout #
PipBoy, (a) is it going to be some sensible use for this feature, (b) I should just ignore it, or (c) is it better to get older version, without it?
I'd say b)
a) depends on the amount of RAM accessible to the GPU. An LLM like llama 3.1 (ChatGPT-like) requires a GPU with 24GB of RAM.
But if it had a camera it could be useful to have the "MachineLearning optimized" SBC. But mostly b)