Leaked 12-core Ice Lake server ES nearly 100 percent faster than Cascade Lake in Geekbench, takes the fight to EPYC Milan

erek

[H]F Junkie
Joined
Dec 19, 2005
Messages
10,875
Looks like it's all over for AMD now with the Ice Lake-SP dominatin' even in ES form !!!!!!!

"AMD's EPYC Rome server parts have been making steady inroads in the server market. According to Q4 2019 figures, Team Red saw a substantial increase in year-on-year server market share. This is in large part because Intel's current 14nm Xeon lineup consumes more power and costs more per core than AMD alternatives.

Intel's monolithic die design means that more cores cost exponentially more to manufacture. If Intel can bring Ice Lake Xeons to the market, delivering more performance with fewer cores, it could mean the end for AMD's server ambitions, even as Team Red preps EPYC Milan. "


https://www.notebookcheck.net/Leake...h-takes-the-fight-to-EPYC-Milan.453582.0.html
 
AMD has been throwing too much heat at Intel for them not to respond hard, I look forward to the show they provide. Still bought my Epyc's for my latest builds but 5 years from now when I revisit them again let's see what they can offer.
 
I mean, Intel has some good stuff. But they don't bring that "stuff" down to the desktop enthusiast market. Granted, AMD's higher end has become higher priced (not as attractive as it once was), but Intel.. sheesh...
 
Looks like it's all over for AMD now with the Ice Lake-SP dominatin' even in ES form !!!!!!!

News forums, news implies neutral and objective reporting. Granted this sub-forum is open for practically anyone to post items as news but it's not good to come off as a screaming fan boy (for any company). I'm only saying this to be constructive in a positive fashion.

That being said I would like to see AMD continue to absolutely dominate as they can't contineously be forced to play catch-up one or two years after dominating or Intel will always go back to it's complacency. While the numbers for AMD's Q4 2019 were good, their best ever actually I was surprised they weren't higher. We enthusiasts only are able to contribute a small portion, a drop in the bucket and AMD needs to continue to dominate in order to have a fair chance to financially grow. It's literally in everyone's best interests if Intel takes it's sweet time.
 
Is there any Epyc chip that sports less than 16 cores? I guess Intel's new processor might compete with the 16- and 24- core Epycs but the rest of the AMD lineup is going to ROFL-stomp the Intels.

RE: News editorial tone... News posts are required to be accompanied by commentary; objectivity isn't needed.
 
News forums, news implies neutral and objective reporting. Granted this sub-forum is open for practically anyone to post items as news but it's not good to come off as a screaming fan boy (for any company). I'm only saying this to be constructive in a positive fashion.

That being said I would like to see AMD continue to absolutely dominate as they can't contineously be forced to play catch-up one or two years after dominating or Intel will always go back to it's complacency. While the numbers for AMD's Q4 2019 were good, their best ever actually I was surprised they weren't higher. We enthusiasts only are able to contribute a small portion, a drop in the bucket and AMD needs to continue to dominate in order to have a fair chance to financially grow. It's literally in everyone's best interests if Intel takes it's sweet time.
I own an AMD tho
 
Quote from a comment on the article, as I was about to post something very similar here:
Just for a quick comparision, my 12-core 3900x ( which is admittedly a consumer and not a server part ) scores 48,000 to 50,000 on Geekbench 4.3 for the multi-core test.
This score at 27k, even for a 12-core efficiency-focused / engineering part doesn't seem that good at all. Interesting to see the final results after release.

I'm glad that Intel is making performance gains, but these don't seem that amazing to me for a 12-core CPU.
Even with the score adjusted from 2.7GHz to 3.0GHz (assuming this is the production clock speed), it still would only score around ~31,000 on Geekbench.

Am I missing something here with these results that make them so amazing? (legitimately asking)
 
Looking through the Geekbench v4 top results for single core runs...there appears to be a lot of false positives or outright purposeful manipulations to the results. Or am I missing something?
 
Quote from a comment on the article, as I was about to post something very similar here:


I'm glad that Intel is making performance gains, but these don't seem that amazing to me for a 12-core CPU.
Even with the score adjusted from 2.7GHz to 3.0GHz (assuming this is the production clock speed), it still would only score around ~31,000 on Geekbench.

Am I missing something here with these results that make them so amazing? (legitimately asking)

Thanks for clearing that up. Was going to say this doesn't really seem possible to jump so far unless it's like 800W 12 core part or some shit (wouldn't put it past them).

Looking through the Geekbench v4 top results for single core runs...there appears to be a lot of false positives or outright purposeful manipulations to the results. Or am I missing something?
rEaL wOrlD wOrKLoaDs
 
rEaL wOrlD wOrKLoaDs

Can you please elaborate?

Second place is a OnePlus 7 Pro sporting a Qualcomm processor.


https://browser.geekbench.com/v4/cpu/singlecore

Screenshot_20200211-072052.png
 
Looking through the Geekbench v4 top results for single core runs...there appears to be a lot of false positives or outright purposeful manipulations to the results. Or am I missing something?

I wouldn't hunt for greatness among the dead. Try Geekbench v5 please.
 
And this is assuming Intel can solve their 10nm yield issues anytime soon. Otherwise, this won't see release for another year.

It's taken Intel 6 months to make Ice Lake widely available - the fact that these aren't scheduled for desktop anytime soon is all you need to know :D
 
Last edited:
Of course, not necessarily yield related, but certainly supply related.... does the remind anyone of when new AMD CPUs (talking before Ryzen) came out, but they were nowhere to be found (bought)? Of course, in those days, AMD, it was all about price. But what price can you put on something that can't be bought? (priceless?)

Just an observation.
 
I mean, Intel has some good stuff. But they don't bring that "stuff" down to the desktop enthusiast market.

They don't?

Granted, AMD's higher end has become higher priced (not as attractive as it once was), but Intel.. sheesh...

Neither have reached AMDs heights during their last performance lead ;)

That being said I would like to see AMD continue to absolutely dominate as they can't contineously be forced to play catch-up one or two years after dominating or Intel will always go back to it's complacency.

We can criticize Intel for their fab fumbles, but the reality is up to that point, they weren't being complacent. They had no competition from AMD and they continued to push up IPC and clockspeeds as well as iterate their platforms, while they kept pricing generation to generation more or less the same.

AMD needs to continue to dominate in order to have a fair chance to financially grow.

This is their 'fair chance'; Intel's fab operations being five or six years behind plan. Absent that opening, the Zen line would have remained uninspiring. AMD now has cash, products that are competitive for much of the market, and good platforms to deploy them on -- that's the best situation they've been in for at least ten years.

It's literally in everyone's best interests if Intel takes it's sweet time.

Companies choosing to not compete is not in anyone's best interest, quite the opposite! You don't think AMD would get complacent?

Take your worst accusations for Intel's product development and pricing strategies -- AMD one-upped them the last time they felt they were ahead.

And this is assuming Intel can solve their 10nm yield issues anytime soon. Otherwise, this won't see release for another year.

It's taken Intel 6 months to make Ice Lake widely available.

That question of 'when' is still hanging in the air for Intel. At least they've had the good sense to put a muzzle on their marketing.
 
They do? Seriously. I guess you're not familiar with the vast difference between the server side of Intel vs the desktop side? It's pretty big.
Where? What are you talking about? Not trolling just point me to the server parts that are so much faster than their enthusiast cpus.
 
They do? Seriously. I guess you're not familiar with the vast difference between the server side of Intel vs the desktop side? It's pretty big.

I'd like to see an example for a common point of comparison -- both Intel and AMD serve the desktop and enthusiast markets with consumer and HEDT products, and both have dedicated server lines, with 'workstation' blending in between all three.

As an enthusiast, I'm not really interested in Intel's server products, unless I'm putting them in... a server. Same for AMD, but to an even lesser extent.
 
I wouldn't hunt for greatness among the dead. Try Geekbench v5 please.

No. The "source" link in the article shows GB v4.3.3 was used, so I was trying to see comparisons and correlations within the same v4 benchmark.
 
I guess we can pretty much nullify all of this then. GB v4? Might as well be posting 3dmark 2000. It really puts the whole "leak" under scrutiny.
 
I guess we can pretty much nullify all of this then. GB v4? Might as well be posting 3dmark 2000. It really puts the whole "leak" under scrutiny.

That was my line of thinking. This benchmark result is a seemingly huge grain of salt ripe for the taking.
 
That was my line of thinking. This benchmark result is a seemingly huge grain of salt ripe for the taking.

Double is dubious, however, if Intel was able to get clockspeeds close to their current desktop crop and the Epyc was running at their traditionally slower all-core clocks (like most large core-count Xeons), coupled with the IPC gains of Ice Lake, significantly faster at the 12-core mark makes a bit of sense in certain workloads.
 
same tech, twice the performance. i don't get it :(
 
I can't believe anyone actually uses this benchmark for ANYTHING other than comparing the same exact architecture.
 
So basically Intel is coming out with nothing of importance?
I hope they catch up so it will drive cpu prices down. It was funny to see Intel reduce prices.
 
Aren't the Ice Lake chips destined for laptops/tablets in the 7-28 watt range.... I mean if that is still true than these scores are nothing to scoff at and are actually pretty impressive.
 
Last edited:
Is there any Epyc chip that sports less than 16 cores? I guess Intel's new processor might compete with the 16- and 24- core Epycs but the rest of the AMD lineup is going to ROFL-stomp the Intels.

RE: News editorial tone... News posts are required to be accompanied by commentary; objectivity isn't needed.
It’s kind of click-baity of an article but Intel has made it pretty clear that Ice Lake and Commet Lake are laptop and tablet chips. If these went toe to toe with EPYC’s than AMD would be in serious trouble.
 
It’s a laptop part, if it weren’t than AMD might as well pack up and go home.

The difference between the 6700K I had for a bit and the 8550U in the laptop I have now is... the 8550U has a lower power target and a higher-spec GPU. A Skylake-based Xeon also the same, some of those even came with GPUs enabled. All are 4C8T Skylake parts..

Clock for clock, if that were possible to test, should be about the same. It should be similar when AMD gets around to releasing competitive parts in that 15w range; you'd be able to compare their mobile parts with their server parts, clock for clock.

[also, with the latest Intel stuff, I've mostly lost track of what the architecture is named versus what the specific parts are named... in this case, Ice Lake could simply refer to the 10nm architecture that follows 14nm Skylake and its derivatives]
 
The difference between the 6700K I had for a bit and the 8550U in the laptop I have now is... the 8550U has a lower power target and a higher-spec GPU. A Skylake-based Xeon also the same, some of those even came with GPUs enabled. All are 4C8T Skylake parts..

Clock for clock, if that were possible to test, should be about the same. It should be similar when AMD gets around to releasing competitive parts in that 15w range; you'd be able to compare their mobile parts with their server parts, clock for clock.

[also, with the latest Intel stuff, I've mostly lost track of what the architecture is named versus what the specific parts are named... in this case, Ice Lake could simply refer to the 10nm architecture that follows 14nm Skylake and its derivatives]
But AMD doesn’t have any 3000 series laptop parts the ones they do have is actually their 2000 series put out on a 12nm process and they are price competitive at best.... they don’t plan on any laptop parts till the 4000 series and they have been pretty tight lipped there.
 
But AMD doesn’t have any 3000 series laptop parts the ones they do have is actually their 2000 series put out on a 12nm process and they are price competitive at best.... they don’t plan on any laptop parts till the 4000 series and they have been pretty tight lipped there.

I'm taking it on faith that they'll get there, really. They have a pretty good shot at the market if they act. More to the point, the comparison is hypothetical, as it would apply once AMD pushes their most current architecture down the line.
 
I'm taking it on faith that they'll get there, really. They have a pretty good shot at the market if they act. More to the point, the comparison is hypothetical, as it would apply once AMD pushes their most current architecture down the line.
Yeah let’s hope I haven’t seen any reviews of anybody undervolting the 3000 lineup to that point I’m not sure if they even function that low at this stage.
 
Back
Top