Discovered that it is still possible to buy 8S systems, and honestly if I ever saw this thing in person I would be scared of it, get a load of this: "be not afraid" etc https://www.supermicro.com/en/products/system/mp/6u/sys-681e-tr
erikk@chaos.social
replied 16 Aug 2024 16:42 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/6B626P1lD5rq46663j
@benjojo you can go ever higher 🙃 But also whyy??? I guess this is just really one strange usecase some has for it
erikk@chaos.social
replied 16 Aug 2024 16:51 +0000
in reply to: https://chaos.social/users/erikk/statuses/112972696123454232
@benjojo https://buy.hpe.com/us/en/compute/mission-critical-x86-servers/compute-scale-up-servers/compute-scale-up-servers/hpe-compute-scale-up-server-3200/p/1014774076 those boxes go up to 32 Sockets (or 16 for 4th gen). The limiting factor on the 8S is only the cache coherency and snooping, but intel does offer the specs so you can build / design your own custom asics to offload this. A bit like NVlink switches.
benjojo
replied 16 Aug 2024 17:01 +0000
in reply to: https://chaos.social/users/erikk/statuses/112972730927707122
@erikk omg. At this point the MTBF of any single CPU/DIMM is not enough to call this a machine that could possibly be reliable
erikk@chaos.social
replied 16 Aug 2024 17:02 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/326VV3657wBGtx7wM9
@benjojo You run RAID on your dimms in those setups, so its fine if one fails or trips, you can just take one offline and swap one back in. Similarly with the CPUs those are all hot swappable and can to auto recovery.
erikk@chaos.social
replied 16 Aug 2024 17:04 +0000
in reply to: https://chaos.social/users/erikk/statuses/112972775046891768
@benjojo But compared to what IBM can do these are all rookie numbers, during hotchips they had a theoritical 4000 socket deployment 😨
ninkosan@infosec.exc..
replied 16 Aug 2024 17:48 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/326VV3657wBGtx7wM9
jalict@mastodon.game..
replied 16 Aug 2024 20:35 +0000
in reply to: https://chaos.social/users/erikk/statuses/112972730927707122
kestral@hackers.town
replied 16 Aug 2024 16:45 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/6B626P1lD5rq46663j
feuerrot@chaos.socia..
replied 16 Aug 2024 16:49 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/6B626P1lD5rq46663j
@benjojo ok, sure.
(I do really like the amount of vector graphics in SuperMicro documentation though)
benjojo
replied 16 Aug 2024 16:50 +0000
in reply to: https://chaos.social/users/feuerrot/statuses/112972725705597255
manawyrm@chaos.socia..
replied 16 Aug 2024 16:54 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/7rT87656G15s3y31t7
kouett@soc.kouett.ne..
replied 16 Aug 2024 16:56 +0000
in reply to: https://chaos.social/users/manawyrm/statuses/112972745804688720
erikk@chaos.social
replied 16 Aug 2024 17:00 +0000
in reply to: https://soc.kouett.net.eu.org/objects/3ec7c548-b540-455f-a9c9-f217a7dc227a
@kouett @benjojo @feuerrot @manawyrm If you want large scale memory footprints for in memory databases going this big is often the only option. @work we have a few usecases for this but could lukly solve it with just a shit ton of SSDs as swap. These boxes are mostly a direct competitor for Power based scale-up systems used for SAP and related large scale databases with the uptime requirments. And having only one box to manage is often espeically in the older style enterprise a big plus
penguin42@mastodon.o..
replied 16 Aug 2024 17:11 +0000
in reply to: https://chaos.social/users/erikk/statuses/112972766541909694
erikk@chaos.social
replied 16 Aug 2024 17:01 +0000
in reply to: https://chaos.social/users/erikk/statuses/112972766541909694
manawyrm@chaos.socia..
replied 16 Aug 2024 17:03 +0000
in reply to: https://chaos.social/users/erikk/statuses/112972771685374454
erikk@chaos.social
replied 16 Aug 2024 17:06 +0000
in reply to: https://chaos.social/users/manawyrm/statuses/112972777640656185
@manawyrm @kouett @benjojo @feuerrot fixing you application is often not really possible or rally expensive to do, or just not yet a solved problem. Often then just trowing money at boxes like this is the cheapest solution, and the failure modes of these boxes are way more predictible then large scale distributed systems. Like most large scale destributed fail just as often, if you see the failure rates most large GPU clusters are dealing with....
erikk@chaos.social
replied 16 Aug 2024 17:44 +0000
in reply to: https://chaos.social/users/erikk/statuses/112972790095585821
dlatchx@fedi.tfnux.o..
replied 16 Aug 2024 20:50 +0000
in reply to: https://chaos.social/users/feuerrot/statuses/112972725705597255
penguin42@mastodon.o..
replied 16 Aug 2024 17:11 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/6B626P1lD5rq46663j
@benjojo Lenovo also do something like this still, I think as a pair of 4U(5U?) chassis that go together.
jamesog@mastodon.soc..
replied 16 Aug 2024 18:17 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/6B626P1lD5rq46663j
@benjojo Looks like a slightly beefier version of some Sun (SPARC) kit I used to run back in the day
Tenzer@s.waq.dk
replied 16 Aug 2024 19:41 +0000
in reply to: https://mastodon.social/users/jamesog/statuses/112973071044264709
jamesog@mastodon.soc..
replied 16 Aug 2024 19:46 +0000
in reply to: https://s.waq.dk/users/Tenzer/statuses/112973400758883198
benjojo
replied 16 Aug 2024 19:54 +0000
in reply to: https://mastodon.social/users/jamesog/statuses/112973420988979352
jamesog@mastodon.soc..
replied 16 Aug 2024 19:59 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/WXx2vd3S9Ry7vbzNlw
@benjojo Those machines were pretty fun. The OOB was probably the best I've used. IIRC it was based on Linux/ARM. And a bit like a mainframe you could do either hardware partitioning or use its own hypervisor for logical partitioning.
mattgrayyes@chaos.so..
replied 16 Aug 2024 20:33 +0000
in reply to: https://benjojo.co.uk/u/benjojo/h/6B626P1lD5rq46663j
@benjojo I think my favourite thing is the lack of handles.
Don’t try moving it without taking all the PCI cards and drives out of it