256
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 05 May 2024
256 points (99.6% liked)
Technology
35127 readers
116 users here now
This is the official technology community of Lemmy.ml for all news related to creation and use of technology, and to facilitate civil, meaningful discussion around it.
Ask in DM before posting product reviews or ads. All such posts otherwise are subject to removal.
Rules:
1: All Lemmy rules apply
2: Do not post low effort posts
3: NEVER post naziped*gore stuff
4: Always post article URLs or their archived version URLs as sources, NOT screenshots. Help the blind users.
5: personal rants of Big Tech CEOs like Elon Musk are unwelcome (does not include posts about their companies affecting wide range of people)
6: no advertisement posts unless verified as legitimate and non-exploitative/non-consumerist
7: crypto related posts, unless essential, are disallowed
founded 5 years ago
MODERATORS
Does the hardware being all so arranged as it is in this manner to create a supercomputer make any difference to that evaluation? Like does the work of putting all the outdated hardware together in the complex way needed to make it functional for supercomputing make it potentially cheaper than buying more modern hardware but having to build it all yourself?
Some of those nodes are broken, it’s got coolant leaks and the processors are 8 years old. I read in another article they meant to replace it a couple years ago but couldn’t due to COVID supply chain problems (makes sense tbh).
What I’m saying is, it might be “assembled” already but due to the power draw, you definitely don’t want to run something this out of date. It’s huge. Might be a logistical problem to even get enough juice for this thing. Another poster pointed out it might need its own electrical substation. IMHO it’s only worth parting out. Far too expensive to run.
Those cabinets look nice though, with the Cheyanne graphic on the front.
The storage drives for all of this have been stripped. You can't just run commands on the hardware... you have to figure out how to cluster things with software, buy drives for it all, have it all installed in a datacenter somewhere which is going to cost way more than the purchase price.
The labor costs for the technical people required to do this are way more than half a million a year.
Didn't know they'd taken out the storage drives but I was aware despite my general ignorance that it's not turnkey ready to go. I guess what I'm wondering is, is there any part of the of the process involved in designing and building such a supercomputing cluster that is already taken care by buying it in the manner that it has been sold and could that in any way offset the increased costs of trying to bring such a cluster online rather than starting from scratch? I'm not saying it is the case, so much as wondering aloud for anyone with expertise to chime in, to see if that's a way it could make sense.
I understand there's a mountain to climb to bring this thing in to a usable state for anyone, but could it maybe get you to base camp more quickly?
I'm guessing that due to the nature of this server (scientific calculations by scientists and students who sent jobs remotely, if I understood right) everything was done on RAM and then stored on SAN (separate servers) afterwards
Otherwise taking out all the storage from each single blade AND putting it back nicely in the racks would have been a massive job, if they did that I would expect to have it half dismantled on the pictured