Well, good thing you were able to make your IT team wait for TurinMeanwhile, I'm waiting for Dell to offer Turin processors...
EPYC prices in retail eventually go down . . . . . with two big caveats: a) It takes a good while after retail availability began, which itself takes a good while for a new EPYC generation after its introduction. b) They virtually only go down for the highest core count dual-socket models of an EPYC generation. Mid core count (not sure about low core count) and/or single-socket EPYCs hardly ever get discounted at all.And yes it is overpriced - where is me cheap and cheerful 64 single socket cores that are very near same clock as much more expensive dual socket ones...
Because not many have experience on making the most out of dual socket systems. If you are using some custom software, you need to make sure that your threads don't do communication or request data from anything on the other socket otherwise down goes the performance and the typical uninformed IT businessman would then try to sell you on a more expensive platform, saying something like, "That was the best system I could give you in your preferred price range. But hey, this other much more expensive system is being used by dozens of clients right now and they have NO issues!".Rather soon within an EPYC generation, retail prices of dual-socket EPYC models frequently undercut retail prices of their P-suffix twins. It's ridiculous.
Because not many have experience on making the most out of dual socket systems
It is technically possible, and also not illegal, to put dual-socket capable CPUs into single-socket systems. :-) The eventually better price of the unrestricted CPUs is merely because these are the run-of-the-mill ones. Combined with retail pricing making wide swings between being utterly ridiculous¹ by default and sometimes being reasonable.Because not many have experience on making the most out of dual socket systems.
How did you fix it, new BIOS? Which board worked out in the end?Ignore all that. We have Turin !
Here is the weird part. supermicro, h13ssl-NT ,version 2.0 BIOS 3.0. You have to populate all 12 dimms even though memory will not be recognised if above 4800, even though the CPU can handle 6000How did you fix it, new BIOS? Which board worked out in the end?
Pretty sure a lot of ES are locked at low freq (like 2 ghz seems common) on purpose to avoid early leaks, plus some new features not enabled in BIOS provided at the time, but perhaps newer one does enable them hopefully...Maybe some TDP setting in BIOS is holding it back?
I just went into bios. It does not have a lot of the settings as normal. Can't select memory speed or anything about the CPU, including FTP. Pretty well locked in everywhere.Maybe some TDP setting in BIOS is holding it back?
With Genoa, this took a while to happen. But with Turin, this *already* happened here in Germany with the 64 cores model. 9555P is priced higher than 9555. (But in case of the 32c, 48c, and 96c models, the non-P models are priced higher. For now.)Rather soon within an EPYC generation, retail prices of dual-socket EPYC models frequently undercut retail prices of their P-suffix twins. It's ridiculous.
Are they in stock?But with Turin, this *already* happened here in Germany
Maybe a limitation of ES chips? Because Turin is supposed to support 6000 MT/s. It's basically a requirement I think, just like we saw with 9800X3D performing a lot better than 9700X. Zen 5 is very bandwidth starved and really needs it.Also you must populate a 12 memory slots with 4800 ram.
But no motherboard support it yet. I have had all 3, and the specs say 4800 only. One of them has 5600, but not 6000.Maybe a limitation of ES chips? Because Turin is supposed to support 6000 MT/s. It's basically a requirement I think, just like we saw with 9800X3D performing a lot better than 9700X. Zen 5 is very bandwidth starved and really needs it.
Maybe only 8 channel ones support 6000.But no motherboard support it yet. I have had all 3, and the specs say 4800 only. One of them has 5600, but not 6000.