Maybe. But I would consider ~$500 far too expensive, except as 'because I can' kudos value and clicks, unless it could be proven to be useful. Initial development, and determining how useful an RP2040 cluster could be, can be achieved using fewer and cheaper Pico.A 4x4 array of the PGA boards stacked 4 high for 64 RP2040s
I would start with a stack of three Pico (~$20), then turn that into a ring or flatten it out if I saw potential, before spending a lot of money on a potential white elephant.
I am intrigued what people think a Pico cluster would be useful for. I can understand people who just want to get some experience of clusters ( where performance isn't a goal ) choosing Pico over a Pi SBC but can't see that offering something a Pi SBC wouldn't be better for at lower cost.
I am not sure how one would compare '128 cores of 128KB at 250MHz' to '4 cores of 1GB at 2.5GHz' at ten times the cost. It's an adventure - and I am never against those - but I can't see it as being a useful or worthwhile one. But I always welcome being proven wrong.
Statistics: Posted by hippy — Sat May 11, 2024 11:32 am