SoylentNews is people
SoylentNews
Submitted via IRC for TheMightyBuzzard
Creating Raspberry Pi clusters is a popular hacker activity. Bitscope has been commercializing these clusters for a bit now and last year they created a cluster of 750 Pis for Los Alamos National Labs. You might wonder what an institution know for supercomputers wants with a cluster of Raspberry Pis. Turns out it is tough to justify taking a real high-speed cluster down just to test software. Now developers can run small test programs with a large number of CPU cores without requiring time on the big iron.
[...] The system is modular with each module holding 144 active nodes, 6 spares, and a single cluster manager. This all fits in a 6U rack enclosure. Bitscope points out that you could field 1,000 nodes in 42U and the power draw — including network fabric and cooling — would be about 6 kilowatts. That sounds like a lot, but for a 1,000 node device, that's pretty economical. The cost isn't bad, either, running about $150,000 for 1,000 nodes. Sure, that's a lot too but not compared to the alternatives.
Huh. That's actually not a bad idea for sounding so silly at face value.
Source: https://hackaday.com/2018/01/24/firing-up-750-raspberry-pis/
(Score: 4, Touché) by Azuma Hazuki on Thursday January 25 2018, @03:59AM (1 child)
...someone imagined a Beowulf cluster of these? :)
I am "that girl" your mother warned you about...
(Score: 3, Informative) by Anonymous Coward on Thursday January 25 2018, @05:03AM
I had thought these things operated as some kind of Single System Image backplane, just extremely slow, due to the Pi I/O bus.
It turns out all they fucking are is a power/ground plane PCB with a bucking power supply on the end for connecting 4 RPis side by side on.
The Uno has i/o header access and a protoboard on the other side, but the Duo and Quattro models are just power backplanes for the units, making their costs seem far less worthwhile compared to the 'cheaper' models.
The RPi's biggest issue has been its lack of memory, so an i/o backplane, even a slow one, would make sense, while allowing the network i/o to be fully used for web hosting or whatever else. As this thing is, literally all it is good for is low bandwidth data logging from usb connected peripherals, or slow data processing where the workflow won't saturate 100 megabit per node.
I hope the switches these things come with are gigabit, or 4x100+1gbit uplink port, because otherwise they are being bandwidth constrained.
The websites 'flash' factor really was frustrating me until I unravelled what they really were and why there wasn't any SSI cluster image software for them.