←back to thread

I regret building this $3000 Pi AI cluster

(www.jeffgeerling.com)
468 points speckx | 2 comments | 19 Sep 25 14:28 UTC | HN request time: 0.4s | source
Show context
nromiun ◴[19 Sep 25 15:31 UTC] No.45302752[source]▶
There is a reason all the big supercomputers have started using GPUs in the last decade. They are much more efficient. If you want 32bit parallel performance just buy some consumer GPUs and hook them up. If you need 64bit buy some prosumer GPUs like the RTX 6000 Pro and you are done.

Nobody is really building CPU clusters these days.

replies(2): >>45304577 #>>45305705 #
1. ted_dunning ◴[19 Sep 25 18:02 UTC] No.45304577[source]▶
Well, El Capitan uses AMD CPUs (which have integrated GPU capabilities) and it is right on top of the rankings lately.

Frontier is right behind it with the same arrangement.

Having honest to god dedicated GPUs on their own data bus with their own memory isn't necessarily the fastest way to roll.

replies(1): >>45304802 #
2. nromiun ◴[19 Sep 25 18:24 UTC] No.45304802[source]▶
They do not. The CPUs are only there to support and push data to the GPUs. Much like Nvidia GH200 systems. Nobody buys these APU chips for their CPU parts.

For comparison there are 9,988,224 GPU compute units in El Capitan and only 1,051,392 CPU cores. Roughly one CPU core to push data to 10 GPU CUs.