Discussion Intel current and future Lakes & Rapids thread

Page 95 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
I'm skeptical about the model names, seems too busy, plus the 6 core i7 having a 300 mhz less turbo than the quad i7?

In theory adding 2 extra cores adds nothing else, just the two cores. In practice, higher core parts have higher idle power. So during the transition process it makes sense to sell both, with the lower core version for higher clocks.

IOn top of that 10nm is (up to half the power?) and Gen11 cores are reportedly smaller than Gen9.5 (indicating a 1:1 power ratio per core isn't likely), so overall we should expect the IGP power budget to remain mostly stationary relative to previous gen.

Half the power with new process alone is doubtful. In the 65/45nm days they claimed 30% reduction in power with a new process. I can't believe that its any better now. If anything, it'll be less. While the EUs are 25% smaller ISO-process, there's also 2.7x as much. Even reaching an identical level will require a big leap. Perhaps that's what Gen 11 is, but let's see it first.

The base clocks in that table look like Y class to me, while the boost clocks are U class, although I was expecting less than 20% drop in fmax. This contradiction alone make me question the veracity of the leak.

The boost clocks are also Y class. The U chips now clock close to 5GHz.

I don't know, I'm not making any conclusions, but it seems to fit with the earlier leaked roadmaps. Of course, much earlier than that though, we had leaks for 4-core Y ICL.

If we assume Y, then why are they segmenting the GPU on a 5-7W chip? It probably makes no difference anyway.

It probably doesn't matter that much for CPU performance, as it can clock to whatever it wants. Why is the base so low if its a U chip though?

Raises more questions than answers. Though there's another possibility. Think of Broadwell-C. A GPU-focused part with lower CPU performance. Doesn't cannibalize 4790K or Skylake.

I said the customer preference will be very skewed towards ICL parts if the chip is very advantageous over Comet Lake parts. Perhaps Comet Lake is for dGPU parts and Icelake is for iGPU.
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,682
136
Half the power with new process alone is doubtful. In the 65/45nm days they claimed 30% reduction in power with a new process. I can't believe that its any better now. If anything, it'll be less.
I'm betting on less too, but combined with a 10% clock speed drop and other arch specific efficiency gains it should be enough to offset the extra transistor budget for Gen11. Even if it isn't, base clocks shouldn't drop this much - they're essentially dropping base clocks by ~30%, which combined with the 10nm power savings means the base power budget for the CPU drops by more than 50% (assuming it's weighed against the GPU power budget for these base clocks, which I still don't think it is)

While the EUs are 25% smaller ISO-process, there's also 2.7x as much. Even reaching an identical level will require a big leap. Perhaps that's what Gen 11 is, but let's see it first.
It's not a big leap if you combine leaner EUs with 10nm power savings and also adjust clocks to meet your target. Remember fast DDR4 can only carry these chips so far.

The boost clocks are also Y class. The U chips now clock close to 5GHz.
fmax may look like Y class, but look at multi-threaded boost clocks, why let the chip boost to 3.5Ghz on all cores when all you have is 7W? You said it yourself, if 10nm doesn't bring half the power at 2GHz base, it sure as hell won't bring half the power at 3.5Ghz.

Why is the base so low if its a U chip though?
My question exactly - to me the base clocks simply don't fit the boost clocks. (even with unexpectedly low boost clocks for U class considering what they achieved on 14nm).
 

Gideon

Golden Member
Nov 27, 2007
1,608
3,570
136
Yes. In fact at this point I think they csan't even match a 9900k in pure performance. 10nm was designed to be dense and power saving and not high clocking. Same as 14nm initially. 14nm-pluses are essentially less dense versions of intial 14nm to allow for higher clocks. I assume we will see the same thing with 10nm if it ever makes it to the desktop at all.
Intel would not spend billions on 14nm capacity if 10nm was up to speed. 10nm probably only good for laptop CPUs due to small size of the chips (yields) and being about low power not performance. Server chips are too large and desktop chips on 14nm simply clock too high to make anything on 10nm that can compete on pure performance. intels half-a decade 14nm optimizing is now biting them in the ass as they will struggle to get out something faster.

10nm has some really aggressive UHP (Ultra High Power i presume) cell libraries, that sacrifice way more densitiy than 14nm++ (vs 14nm HD):

intel-10nm-cells-density.png


It remains to be seen when Intel finally manages to ship silicon using that process. If they do however, it should clock quite decently. The main question is of course when will that be?
If I had to guess, I'd say, probably a year after the mobile Ice Lake releases (if no additional issues crop up)- Considering that Intel actually seems to ramp up 10nm now. $530M dollar loss for ramp-up at least means they are somewhat more serious of actually delivering something by the year-end.
 

TheGiant

Senior member
Jun 12, 2017
748
353
106
Yes. In fact at this point I think they csan't even match a 9900k in pure performance. 10nm was designed to be dense and power saving and not high clocking. Same as 14nm initially. 14nm-pluses are essentially less dense versions of intial 14nm to allow for higher clocks. I assume we will see the same thing with 10nm if it ever makes it to the desktop at all.
Intel would not spend billions on 14nm capacity if 10nm was up to speed. 10nm probably only good for laptop CPUs due to small size of the chips (yields) and being about low power not performance. Server chips are too large and desktop chips on 14nm simply clock too high to make anything on 10nm that can compete on pure performance. intels half-a decade 14nm optimizing is now biting them in the ass as they will struggle to get out something faster.
oki
I have a surface pro 4 (i5 6300U 2C/4T turbo 2.9GHz). the 15W TDP but the fan is often on when longer playing youtube or MS Teams calls
it is time to get something new..
Is reasonable to expect a new ICL-U 4C/8T with new gen iGPU (very nice according to leaks) with turbo of around 4.0GHz all cores and real 15W TDP ? this year?
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,682
136
and real 15W TDP
TDP has been "real" all along. I have a i7 4500U that I measured the hell out of when I purchased the ultrabook, it's behavior is a strict interpretation of Intel's documentation on power management.

If you want next to zero fan noise for low intensity tasks on ultra-light devices then aim for 7W TDP, or alternatively for devices with bigger heatsinks. The Haswell era 13" Macbook my wife uses has lower fan noise and spends more time being passively cooled than my 13" Haswell ultrabook, and it does so by passively dissipating more heat than the thinner and considerably lighter Windows PC.

Unfortunately for you Skylake only has hybrid support for VP9 decoding, so power usage under Youtube is likely more than you would get under KBL or anything newer. That's one of the reasons I'm looking for an upgrade myslef, that and the need for short burst ST performance, as Photoshop is a curse when it comes to performance and fragments of the software are still entirely dependent on ST performance.
 
Last edited:

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
I said the customer preference will be very skewed towards ICL parts if the chip is very advantageous over Comet Lake parts. Perhaps Comet Lake is for dGPU parts and Icelake is for iGPU.

If that SKU list was actually legit, it's pretty questionable that the Icelake models would even be faster in sustained MT than Comet. Not to mention the OEM actually has to have Icelake product to sell.
 

Dayman1225

Golden Member
Aug 14, 2017
1,152
973
146
aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9LL00vODM1MzY2L29yaWdpbmFsLzIwMDI2ODc2NjguanBn

According to an supposed leaked XPS roadmap we should see an XPS 2-1 powered by ICL U in August and an XPS 13 7200 in Feb’20.
 

TheGiant

Senior member
Jun 12, 2017
748
353
106
4.0GHz clock for quad-core CPU at 15W is very unrealistic.
why is that

the i7-8650U at 31W in MS SUrface book 2 can sustain 3,5GHz all core, 4C/8T

so the low power 10nm isn't good enough for ok not 4 but 3,5GHz all core with new uarch at 15W ?

I don't like that. Apple A12X is already as fast as regular i7 8700 in geekbench (grain of salt)...
 

Thala

Golden Member
Nov 12, 2014
1,355
653
136
why is that

the i7-8650U at 31W in MS SUrface book 2 can sustain 3,5GHz all core, 4C/8T

so the low power 10nm isn't good enough for ok not 4 but 3,5GHz all core with new uarch at 15W ?

I don't like that. Apple A12X is already as fast as regular i7 8700 in geekbench (grain of salt)...

It is practically given that the new uarch is larger and consumes more power per clock than Skylake. So parts of the power gain given by technology is eaten up by uarch. The question of course is, how large is that part.
Regarding A12X, you should not compare Aarch64 to x64 with respect to efficiency.
 
Last edited:

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
According to an supposed leaked XPS roadmap we should see an XPS 2-1 powered by ICL U in August and an XPS 13 7200 in Feb’20.

Why is the XPS 13 called the 7390 series? There's Latitude 7390, but XPS?

Update: There's a minor issue with the Dell roadmap. XPS 13 9380 is shown as in development, and the whole slide is from Feb 2018.

coercitiv said:
TDP has been "real" all along.
@TheGiant Yea, it has to be especially in thermally constrained form factors. They can pretend its different on desktops, and thick DTRs, but on ultrabooks that won't fly.

Gideon said:
10nm has some really aggressive UHP (Ultra High Power i presume) cell libraries, that sacrifice way more densitiy than 14nm++ (vs 14nm HD):

You could look at it the other way around. The density jump is impressive. They are deciding to further branch out 10nm to allow specialization. Even the 10nm UHP is double the density of HD High Density version of the 14nm process.

By the way, BK's legacy continues.
 
Last edited:

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
By the way, BK's legacy continues.
So does Otellini's. Maybe Swan can edge this 'titanic' in the right direction (though missing out on mobile chip production will continue to place Intel at a disadvantage for years to come).
 

Dayman1225

Golden Member
Aug 14, 2017
1,152
973
146
1557350933476.png
Icelake Server is sampling now to customers, coming in H1'20 - Moving to 4-5 Quarter server Cadence, Sapphire Rapids in 2021.
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
So wait.

Icelake-U/Y in June is supposed to be limited quantity. Higher volume shipments come in Q4, right? And if IceLake server is sampling now, why are they even bothering with Cooper Lake?

What's really going on here?
 

Dayman1225

Golden Member
Aug 14, 2017
1,152
973
146
1557353743868.png
  • Tigerlake coming in 2020.
  • Xe Graphics Engine,
  • New CPU Core,
  • Next gen I/O and lastest display tech.
  • Tigerlake Silicon already back and booting windows and chrome.


1557353654116.png

  • Tigerlake is 4x Graphics perf vs WHL. (15w vs 25w).
  • ICL 2.5-3x perf boost over WHL. (15w vs 15w)
  • Tigerlake 9w 4+2 is 2x productivity over Amber Lake 2+2 5w
  • Tigerlake 4x encode perf over WHL. 4K60 -> 8K 60
Gregory Bryant also confirmed Client dGPU in 2020
 
  • Like
Reactions: coercitiv

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,434
136
So wait.

Icelake-U/Y in June is supposed to be limited quantity. Higher volume shipments come in Q4, right? And if IceLake server is sampling now, why are they even bothering with Cooper Lake?

What's really going on here?
I would say a lot of PR to try and salvage face, and maybe not loose a lot of business to Ryzen 3000/Rome.
 

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
And if IceLake server is sampling now, why are they even bothering with Cooper Lake?

Because they can't actually yield the 10 nm products. Cooper Lake for instance is what will be actually bought versus Icelake Server; and for client Comet Lake and Rocket Lake versus Icelake Client and Tigerlake.
 

Dayman1225

Golden Member
Aug 14, 2017
1,152
973
146
Because they can't actually yield the 10 nm products. Cooper Lake for instance is what will be actually bought versus Icelake Server; and for client Comet Lake and Rocket Lake versus Icelake Client and Tigerlake.
You really love making big assumptions without any information to back it up
 

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
You really love making big assumptions without any information to back it up

Let me put it this way, you kind to have to read the tea leaves. That they are fully refreshing products with 14 nm in 2020 is pretty solid evidence of what the plan is. And the plan is to sell 14 nm until they can get 7 nm out; and use 10 nm as a distraction to Wall Street.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
The GPU portion of Icelake quad core has increased in comparison to CPU, but in relation to the SoC, it did not change when compared to Skylake derivatives.

This means the GPU is surprisingly compact despite the significant performance and featureset jump. It's a combination of improved architectural efficiency and greater process scaling compared to others like CPU cores and I/O.

Also on a sidenote. Core M was released 2015 as a new product category with supposedly great enhancements to allow low TDP. However the final product was a disappointment.

Lakefield is in the roadmap as M-series. Lakefield could be thought as a proper Core M. Since a new CPU architecture from inception to being on shelves take 4-5 years, it makes sense.