• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Discussion Intel current and future Lakes & Rapids thread

Page 692 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
I don't think this is actually SPR-X but rather actual Workstation parts. But alright then.

That's true but for all practical purposes the W-series workstation parts have taken over from "HEDT". That's true of current gen Ice Lake W series as well, as they use a different socket from their server counterparts.

From that angle you can already get a 38 core Ice Lake W3375. It's actually possible to get an Ice Lake Xeon + workstation motherboard for ~$1100 if high core count isn't needed, and you still get all that connectivity. Of course, it's also possible get the 38 core / 76T on a high end motherboard for 12 grand.
 
Are the E-cores 1/4th the size of the P-cores? Are we looking at a 6+8 design?
Roughly seems that way from what I can make out of the diagram, but it's difficult to tell without knowing what is cache and what is core.

This does look like a 6+8 though.
 
That's true but for all practical purposes the W-series workstation parts have taken over from "HEDT". That's true of current gen Ice Lake W series as well, as they use a different socket from their server counterparts.

From that angle you can already get a 38 core Ice Lake W3375. It's actually possible to get an Ice Lake Xeon + workstation motherboard for ~$1100 if high core count isn't needed, and you still get all that connectivity. Of course, it's also possible get the 38 core / 76T on a high end motherboard for 12 grand.
The regular -X series is currently more interesting as it's the only one where Intel will essentially have no competition. The last non-WX products released by AMD are based on Rome.

WX is whatever, will be similar to Milan. But -X will have much more significant advantages.
 
Meh, no CPU with just 8 "big" cores is HEDT, in my books, no amount of little cores on top of these, helping it to do well in CB, is going to change that. The score on par with 2 years old TRs is irrelevant. If CB supported GPU rendering, you could hypothetically run it on iGPU part and maybe do equally well (under assumption the iGPU is not too crappy) - you would not call it HEDT cause of that.
Who cares about the core config if the performance is there? Arguably the best workstation chip would be something like 8-16 big cores and 64-128 Atom. Big cores to handle the serial bottlenecks and a sea of Atom cores to handle highly parallel compute like rendering. And on the topic of accelerators, if they're suitable to the task, then maybe that task isn't a good CPU benchmark at all.
 
There is a non-WS SPR. It's why I was asking 🙂
Do you really think there is going to be a Non-WS SPR? I mean the guys are struggling to release normal vanilla SPR, so far the "Monolithic" (Up to 28 Cores) SPR for Desktop has been a glimpse on sisoftware. Perhaps they just cancelled that all together(Like the defunct Non-Pro Threadripper that was built and tested but never released)
 
Who cares about the core config if the performance is there? Arguably the best workstation chip would be something like 8-16 big cores and 64-128 Atom. Big cores to handle the serial bottlenecks and a sea of Atom cores to handle highly parallel compute like rendering. And on the topic of accelerators, if they're suitable to the task, then maybe that task isn't a good CPU benchmark at all.

But is the performance there? 40000 looks great, its HEDT numbers, until you realize its like 3 years old HEDT numbers. True HEDT cpu of the current generation would be performing far better. If Intel ever manages to release those SPR chips, and they will actually work as intended, then it will be obvious, that HEDT numbers in 2022 are somewhere else.
 
True HEDT cpu of the current generation would be performing far better. If Intel ever manages to release those SPR chips, and they will actually work as intended, then it will be obvious, that HEDT numbers in 2022 are somewhere else.
The thing is... We have yet to se a 2S System break 80,000 in Cinebench and those are 100+ Cores... The Ice Lake W are just Meh because they are not longer "Extreme Edition" and SPR-X will just be another Workstation CPU.
 
But is the performance there? 40000 looks great, its HEDT numbers, until you realize its like 3 years old HEDT numbers. True HEDT cpu of the current generation would be performing far better. If Intel ever manages to release those SPR chips, and they will actually work as intended, then it will be obvious, that HEDT numbers in 2022 are somewhere else.
Agreed, RPL is not an HEDT chip. Was just saying that I think the core configuration is the wrong thing to focus on, and furthermore, that hybrid is actually a good fit for workstation tasks.
 
The thing is... We have yet to se a 2S System break 80,000 in Cinebench and those are 100+ Cores... The Ice Lake W are just Meh because they are not longer "Extreme Edition" and SPR-X will just be another Workstation CPU.

Zen 4 Threadripper will care of that surely. And it wont be too long.
 
Not really. E core spamming doesn't really work there. It's all about per core performance.
Yeah, I'm of the opinion that once you have say 8 E-cores, that's more than enough to take care of all the low priority background tasks. After that, just give me more P cores because if I had a serious multi-threaded workload, I'd want it to run as fast as possible and therefore would like it to run on the fastest cores available. To be fair, this approach is not unlike what Apple does for their workstation SOCs: 2 efficiency cores to handle all of the background stuff, so as to not bog down the performance cores with frivolous tasks, then the rest are the actual performance cores.
 
After that, just give me more P cores because if I had a serious multi-threaded workload, I'd want it to run as fast as possible and therefore would like it to run on the fastest cores available.
The same amount of power and silicon devoted to E cores would give something twice the performance for embarrassingly parallel tasks.
 
The same amount of power and silicon devoted to E cores would give something twice the performance for embarrassingly parallel tasks.
Yeah, I don't doubt that. That conclusion is more a consequence of how large the P cores are and/or how small the E cores are. If that relationship didn't exist, it would be more feasible to do a configuration where you have more P cores than E cores.
 
Back
Top