• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Discussion Intel current and future Lakes & Rapids thread

Page 411 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

uzzi38

Golden Member
Oct 16, 2019
1,496
2,834
106
Definitely looks like they are doing a respin to help frequencies and/or yield. 270 W is really pushing it in a workstation processor.
Uh, you know Threadripper is 280W, right? OEMs are already building systems around these kinds of TDPs.
 
  • Like
Reactions: lightmanek

jpiniero

Diamond Member
Oct 1, 2010
9,168
1,798
126
Uh, you know Threadripper is 280W, right? OEMs are already building systems around these kinds of TDPs.
Yeah, but Threadripper Pro has the benefit of having the chiplets being spaced out a bit. Plus they are surely extremely well binned chiplets.
 

JoeRambo

Golden Member
Jun 13, 2013
1,080
897
136
I assume the recent interview with mr. Norrod from AMD is the 1% you talk about? Because that interview was EXCELLENT.
Agreed, but it was less of formal "launch" or "event" interview format, and more podcast one. Looking forward to further interviews like that.
 

Mopetar

Diamond Member
Jan 31, 2011
5,749
2,509
136
I think that just comes down to where someone has a great product that they don't even need to sell you because it speaks for itself or if they've been handed a turd sandwich that they need to pass off as palatable to people who know better.
 

IntelUser2000

Elite Member
Oct 14, 2003
7,391
2,087
136
Also, an interesting note of Ice Lake for workstations.
Eh, it's just a successor to the 28 core Purley "workstation". A proper HEDT would be a successor to the Skylake-X 18 core parts.

Unfortunately we can't see the anticipated worst case production week window, I guess it's 4-8 weeks down on the best case week.
Great. Tigerlake hype machine Part 2. Hype until June, announce in July, "launch" in August overview of it in so-called tech conferences by September, and on the shelves by October/November.
 
  • Like
Reactions: Tlh97 and uzzi38

eek2121

Senior member
Aug 2, 2005
862
955
136
Eh, it's just a successor to the 28 core Purley "workstation". A proper HEDT would be a successor to the Skylake-X 18 core parts.



Great. Tigerlake hype machine Part 2. Hype until June, announce in July, "launch" in August overview of it in so-called tech conferences by September, and on the shelves by October/November.
Tiger Lake is a solid chip thus far. Why the hate? 8-core H variants are shipping next month. 15W variants are rumored to be getting a respin.

I suspect Tiger Lake will exist alongside Alder Lake for quite a while. IMO it is the first decent chip Intel has pushed out in years.

EDIT: The 5800H edges out The AMD Ryzen 9 5900HX in Geekbench, for example.
 
Last edited:

mikk

Diamond Member
May 15, 2012
3,117
937
136
Great. Tigerlake hype machine Part 2. Hype until June, announce in July, "launch" in August overview of it in so-called tech conferences by September, and on the shelves by October/November.

There was a rumour about a September launch which probably is a best case window, I doubt they will launch it earlier than that, more likely October. Tigerlake-U is a mobile chip, the initial mobile ramp up is always slower than for desktop. Look to AMD and Cezanne 3 months after launch, there is almost nothing. For this reason we will see ADL-S desktop first followed by mobile, which you didn't believe last year by the way!

Speaking about Tigerlake-U....volume looks great. Combined with Icelake 10nm ULV volume seems a lot higher than 14nm ULV by now, many of these 14nm devices are old. In reference to geizhals.de available devices/configurations:

Tigerlake-UP3 423
Cometlake-U 395
Icelake-U 327

Renoir U+H 267
Cezanne 8
Lucienne 14
 

eek2121

Senior member
Aug 2, 2005
862
955
136
There was a rumour about a September launch which probably is a best case window, I doubt they will launch it earlier than that, more likely October. Tigerlake-U is a mobile chip, the initial mobile ramp up is always slower than for desktop. Look to AMD and Cezanne 3 months after launch, there is almost nothing. For this reason we will see ADL-S desktop first followed by mobile, which you didn't believe last year by the way!

Speaking about Tigerlake-U....volume looks great. Combined with Icelake 10nm ULV volume seems a lot higher than 14nm ULV by now, many of these 14nm devices are old. In reference to geizhals.de available devices/configurations:

Tigerlake-UP3 423
Cometlake-U 395
Icelake-U 327

Renoir U+H 267
Cezanne 8
Lucienne 14
The ramp will be slower than Rocket Lake. Sapphire Rapids and Alder Lake both are launching around the same time. Intel appears to be going all in on 10nm and Golden Cove.

EDIT: I see a lot of people hating on ADL-S, kind of odd to me, as an 8+8 ADL-S will be significantly faster than a Ryzen 5950X.
 

TESKATLIPOKA

Senior member
May 1, 2020
378
383
96
.....
EDIT: I see a lot of people hating on ADL-S, kind of odd to me, as an 8+8 ADL-S will be significantly faster than a Ryzen 5950X.
I seriously wonder how you got to this conclusion about 8+8 Alder Lake being faster than a 16 core Ryzen 5950X.
If we talk only about 8 threads or less, then It should be faster, but after that the difference will become increasingly smaller until Zen3 turns out faster. Let's be honest, those small cores will have a lot less IPC and clockspeed than the big ones. At best It could be faster than Ryzen 5900X in multi thread but slower than Ryzen 5950X.

edit:
I will do a napkin math where in the first case Alder has 10% better performance than Rocket Lake in Cinebench and the small core provides 50% of big cores performance. In the second case It's 20% and the small core provides 60%.
Cinebenech R23 Single(TPU):
Ryzen 9 5900X: 1622 (100%)
Ryzen 9 5950X: 1658 (102%)
Core i9-11900K: 1676 (103%)
Alder Lake 8+8: 1844 (114%)
Alder Lake 8+8: 2011 (124%)

Cinebenech R23 Multi(TPU):
Core i9-11900K: 13531 (100%)
Ryzen 9 5900X: 21707 (160%)
Alder Lake 8+8: 22326 (165%)
Ryzen 9 5950X: 25916 (192%)
Alder Lake 8+8: 25980 (192%)

As shown Alder Lake needs to be the second case If It wants to be a tiny bit faster than Ryzen 9 5950X in multi thread and this was just a napkin math, yet I can already conclude that It won't be significantly faster than Zen3 in multi, that's just impossible for 8+8 configuration from the info we got so far.
On the other hand, in mobile segment Alder Lake could be the fastest CPU you can get simply because AMD won't release anything higher than 8 core as a monolith chip and chiplet design is not very suitable for the mobile segment.
 
Last edited:
  • Like
Reactions: Tlh97 and Mopetar

eek2121

Senior member
Aug 2, 2005
862
955
136
I seriously wonder how you got to this conclusion about 8+8 Alder Lake being faster than a 16 core Ryzen 5950X.
If we talk only about 8 threads or less, then It should be faster, but after that the difference will become increasingly smaller until Zen3 turns out faster. Let's be honest, those small cores will have a lot less IPC and clockspeed than the big ones. At best It could be faster than Ryzen 5900X in multi thread but slower than Ryzen 5950X.

edit:
I will do a napkin math where in the first case Alder has 10% better performance than Rocket Lake in Cinebench and the small core provides 50% of big cores performance. In the second case It's 20% and the small core provides 60%.
Cinebenech R23 Single(TPU):
Ryzen 9 5900X: 1622 (100%)
Ryzen 9 5950X: 1658 (102%)
Core i9-11900K: 1676 (103%)
Alder Lake 8+8: 1844 (114%)
Alder Lake 8+8: 2011 (124%)

Cinebenech R23 Multi(TPU):
Core i9-11900K: 13531 (100%)
Ryzen 9 5900X: 21707 (160%)
Alder Lake 8+8: 22326 (165%)
Ryzen 9 5950X: 25916 (192%)
Alder Lake 8+8: 25980 (192%)

As shown Alder Lake needs to be the second case If It wants to be a tiny bit faster than Ryzen 9 5950X in multi thread and this was just a napkin math, yet I can already conclude that It won't be significantly faster than Zen3 in multi, that's just impossible for 8+8 configuration from the info we got so far.
On the other hand, in mobile segment Alder Lake could be the fastest CPU you can get simply because AMD won't release anything higher than 8 core as a monolith chip and chiplet design is not very suitable for the mobile segment.
The big cores will be around 20% faster. The small cores will be around 10% slower.

Simple math. Alder Lake will beat top Zen 3 chips by around 10%.

EDIT: The leaks I have been able to find appear to indicate that the little cores are Skylake with AVX-512 stripped out along with changes in a few other areas. Early engineering samples had all cores clocking up close to 4.8 ghz.

Intel’s golden cove cores are HUGE. Larger than Zen 3. They are also fast.
 
Last edited:
  • Like
Reactions: lightmanek

coercitiv

Diamond Member
Jan 24, 2014
4,293
5,466
136
The big cores will be around 20% faster. The small cores will be around 10% slower.

Simple math. Alder Lake will beat top Zen 3 chips by around 10%.
How exactly are the small core going to be just 10% slower?

The estimate for small cores used on the forums is around Skylake IPC. That's at least 15% lower ST performance at ISO clocks. For MT performance you'd have to account for the lack of SMT, which again lowers performance by 15% or so. That puts the small core in ADL-S at around 30% performance deficit versus a Zen 3 core in throughput oriented workloads.

I can see how ADL-S will be significantly faster than 5950X chips in workloads that use up to 8 threads, maybe even up to 16 threads if Gracemont cores clock past 4Ghz, but going higher than that in thread count will see ADL-S slowly fall behind. Aren't Intel themselves claiming up to 2X MT performance over Rocket Lake?
 

TESKATLIPOKA

Senior member
May 1, 2020
378
383
96
The big cores will be around 20% faster. The small cores will be around 10% slower.

Simple math. Alder Lake will beat top Zen 3 chips by around 10%.
Big cores could be 20% faster, but small cores won't be only 10% slower than Rocket Lake or Tiger Lake! They have supposedly Skylake IPC which is more than 10% slower and then the clocks will be also noticeably lower + supposedly no HT for them. So your simple math is wrong.
 
  • Like
Reactions: Tlh97 and Markfw

IntelUser2000

Elite Member
Oct 14, 2003
7,391
2,087
136
EDIT: The leaks I have been able to find appear to indicate that the little cores are Skylake with AVX-512 stripped out along with changes in a few other areas.
The Gracemont cores are NOT Skylake. It's just expected to have around Skylake performance. It's a extension of the dual decode cluster design that was introduced with Tremont.

Aren't Intel themselves claiming up to 2X MT performance over Rocket Lake?
Yea, and likely in mobile as with the MLID leak. Actually I expect even the 2+8 part to beat 4 core Tigerlake-U by a decent amount(20-30%).

I'm expecting it'll beat the 5900X, but 5950X will need the Golden Cove cores to be a lot faster. They are only saying 20% in single thread boost, which is a firm no on matching 5950X nevermind beating it.

Tiger Lake is a solid chip thus far. Why the hate? 8-core H variants are shipping next month. 15W variants are rumored to be getting a respin.
Tigerlake is an OK chip, nothing more. Still gets it's ass kicked by Renoir in MT because the core is too big and power hungry to fit more than 4 of them. Xe will be shadowed by the graphics in Remembrandt, maybe even by Van Gogh!
 
Last edited:

andermans

Member
Sep 11, 2020
57
52
51
I'd be careful with looking at per core performance improvements to determine MT performance. What often gets quoted are ST performance or IPC numbers and those may not always be an indication of MT performance. A great example here is Zen3, which got like +19% IPC and >+25% ST performance but only <+10% MT performance from the 3950x to the 5950x.

Aside from that the top Tremont cores get about ~750 on Geekbench for ST: https://browser.geekbench.com/v5/cpu/search?utf8=✓&q=n6005 . An expected +30% on top of that is ~1000 which is pretty close to the original Skylake (core 6000 series), but current rocket lake is closer to 1800. So Skylake level performance is likely more like 50% of big core than just 10% slower. That said see my advice above about not looking at ST perf for MT predictions.
 

andermans

Member
Sep 11, 2020
57
52
51
Depending on power I think Tigerlake being outdone by Rembrandt/Van Gogh is actually quite likely. Looking at
, at 15W it is a coin toss between Tigerlake and Cezanne and at 25/28W Tigerlake is 0-30% faster.

Van Gogh likely will not play in the 25/28W range but at the 15W it would have to do very little to turn that into a win. Rembrandt will have +50% GPU cores (besides the upgrade to RDNA2), which while it probably won't provide +50% perf is likely enough to beat Tigerlake even at the 25/28W power level.

Of course the real question is the graphics included with Alderlake as that is a more timely competitor for these. AFAIU there are no big architectural changes, but a +10% or so due to clock tuning or tweaks could make a noticeable difference in the performance race.
 

Bouowmx

Golden Member
Nov 13, 2016
1,103
493
146
When is Tiger Lake-H coming? Feels like a perpetual wait.
Also, is it going to be replaced by Alder Lake-S BGA or -P?
 

IntelUser2000

Elite Member
Oct 14, 2003
7,391
2,087
136
Aside from that the top Tremont cores get about ~750 on Geekbench for ST: https://browser.geekbench.com/v5/cpu/search?utf8=✓&q=n6005 . An expected +30% on top of that is ~1000 which is pretty close to the original Skylake (core 6000 series), but current rocket lake is closer to 1800.
Here are the comparisons of some CPUs. Tremont, Goldmont Plus refresh, and Goldmont Plus.

GB5 Integer
N6000: 702
N5030: 520
N5000: 488
6700K: 1171 @ 4.6GHz
11900K: 1673

I expect the 10W version of Tremont to be few % faster and get 750. Such 10W chips are more representative of the Gracemont that will be in Alderlake as it won't be power bound as the 6W mobile chips are.

Still, overall you are right. I believe it can beat 5900X, but not 5950X.

Golden Cove = 1.5x Skylake/Gracemont
Hyperthreading = 1.2x
Clock speed = 1.3x

You are looking at each Gracemont core being equal to about 40% of the Golden Cove core in MT performance.

I wonder if this prediction is as accurate as your "Tigerlake graphics performing about 30-40% better than Icelake" last year, it was a bit off.
Actually in that HW Unboxed review it's not far off. Yes I know they do better, but when you parade that it'll be so good for so long, you expect much more. Not to mention they still suffer from driver issues. It's not just under Batman! It's not a make or break issue but a noticeable downgrade from AMD/Nvidia.

Also I don't believe Tigerlake is bandwidth bound as you think. I think the CPU cores are too power hungry which is why Renoir gets the advantage under 15W.

LPDDR5-5400 is a mere 25% bandwidth improvement. If it was 100% bound by bandwidth, then it's a possible 25% gain. If that's a 60% gain(which is a LOT) then we'll get 15%.

Remembrandt I think will be a big deal on the iGPU side and I expect 60-80% improvement as well.
 

Exist50

Senior member
Aug 18, 2016
249
290
136
In terms of Alder Lake multithreaded performance, and Gracemont vs Golden Cove, need to consider sustained vs burst performance. Peak clock speeds are only relevant for short loads. For sustained, you need to also factor in how much power the core consumes. That's probably going to significantly hurt Golden Cove and relative help Gracemont.
 

ASK THE COMMUNITY