SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : AMD, ARMH, INTC, NVDA -- Ignore unavailable to you. Want to Upgrade?


To: rzborusa who wrote (43246)10/21/2021 11:44:08 PM
From: Joe NYCRead Replies (1) | Respond to of 56646
 
Here is another supercomputer in Sweden being installed.

65,000 Epyc Cores and future Graphics modules will have:

------------------------------------
GPU partitionThis partition will be equipped with GPU nodes, each of which will be configured with:

one AMD EPYC™ processor with 64 cores (which is currently under development)512 GB memoryfour AMD Instinct™ GPUs (which are currently being developed)

-----------------------------

Which looks like another supercomputer win for Mi200

Dardel | KTH



To: rzborusa who wrote (43246)10/21/2021 11:58:41 PM
From: Joe NYCRead Replies (1) | Respond to of 56646
 
If AMD had unlimited wafers, I wonder if any of the cloud / hyperscalers would even consider Ice Lake. Intel market share would collapse. It might be that the demand supplied by Intel to Hyperscalers is the demand that AMD can't satisfy (if the lead times are really ~20+ weeks for Milan).

Looking further, Sapphire Rapids will have the same cores as Alder Lake big cores, and these cores can be quite power hungry when pushed. I can't find the link, but all core clock speed of 5.0 - 5.2 GHz on 8 of those cores burned 400 Watts.

Since SPR will have some 56 of these cores, they will have to be clocked way down. It seems Intel is still fiddling with the design to squeeze something competitive out of it.



To: rzborusa who wrote (43246)10/22/2021 2:26:38 PM
From: Joe NYCRead Replies (1) | Respond to of 56646
 
BTW, another gem from the call was that the reason for decreasing ASPs in servers is that the mix of the low core count to high core count is shifting to low core count. Which coincides with Enterprise growing and Cloud shrinking..

That means that the high core count is shifting to AMD.