Discussion Intel current and future Lakes & Rapids thread

Page 129 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
H

Hitman928

Diamond Member
Apr 15, 2012
5,243
7,792
136
TheGiant said:
well I wanna see actual dell/hp/lenovo builds tested
my surface pro 4 6300u 2c4t is aging
the fan bothers me as it is annoying to have in 2019
I think it is the same like with ryzen 3k, heat flow density is too high
Click to expand...

This will be a continuing problem going forward, unfortunately.
 
  • Like
Reactions: majord
D

dahorns

Senior member
Sep 13, 2013
550
83
91
CSMR said:
OK faster GPU but that's not relevant to the argument. You picked up on a red herring to extend this discussion but the appropriate comparison is obvious and fully explained in each of my posts. 25W to 28W is more similar than 15W to 28W, in power and in CPU performance and that's why the comparison is apt. I don't know why you are pushing this but if you want to reply on this point please see my previous posts for answers.
Click to expand...

This whole chain started based on your initial post comparing the 1065G7 results to the i7-8559u and concluding: "I'd estimate CPU performance the same, GPU performance 5% better based on the notebookcheck benchmarks, at a 10% lower TDP. Disappointing. They appear to have made very little progress and are covering it up - successfully as news outlets including AnandTech have taken the bait - by comparing the best new IceLakes to mediocre previous-gen processors, not to the best previous Intel processors."

My criticisms: (1) your comparison to a product with edram is simply not representative of the progress made, particularly on the GPU front, (2) the GPU on the i7-8559u has a much higher power budget than the 1065G7 in either 15W or 25W configurations, and (3) you've severely understated the GPU performance differences. As a result, your conclusion is suspect, or, as I said initially "an odd-take".

I do agree that CPU performance between the two when comparing the 25W 1065G7 is pretty close, with each alternating victories in difference workloads. Prime testing of each appears to confirm a ~12% difference in available power as suggested by the TDPs. The 8559u stabilizes at 24.2W (at 2.9 ghz) to processing cores during a prime run while the 1065G7 stabilizes at 21.4W (at 2.5 ghz).

However, as I previously outlined, GPU performance is substantially superior on the 1065G7 and at a much lower power usage. The only GPU tests that show an advantage for the 8559u are ones at very low quality that are likely limited by CPU or benefiting greatly from the edram and Overwatch (seems unusually low and likely represents a driver issue of some sort). Rocket league at 1080p gives between a 6% and 11% advantage to the 1065G7 , SOTR its + 27% to 32%, Witcher 3 is +84% to 94% at the only setting run on both (apparently because the 8559u couldn't even run Witcher 3 at higher settings). Again, this is all where the GPU on 1065G7 appears to be capped at 15W (1.1. ghz, its max rated speed) while the 8559u draws 47% more power (22W) at roughly the same frequency and can burst to ~40W (1.2 ghz).

The difference is basically this: for the 8559u, either the computational cores or the graphics cores can use the full power budget. The 1065G7 is balanced differently. Its computational cores can utilize 25W, but its graphics cores are capped at 15W. Thus, if you're running an entirely computational benchmark on the 1065G7, you're benching a 25W CPU. If you're running an entirely graphics benchmark, you're benching a 15W GPU. For the 8559u its 28W either way. Comparing them solely off of the TDP for the entire chip is simply not providing the whole story.
 
Last edited:
  • Like
Reactions: majord and Zucker2k
M

mikk

Diamond Member
May 15, 2012
4,133
2,136
136
On legitreviews there is quite a big gap in Rainbow Six Siege, Metrox Exodus and Three Kingdoms. All the other tests didn't show big differences between 15W and 25W for gaming but this is preliminary, we need more tests on production units and also new drivers. It might look different in 2-3 months. The Icelake GPU is a different tier for sure, a thin 15W device with basic cooling beats 30W Iris Pro 655 on a NUC with better cooling. i7-1068G7 should crush it.
 
  • Like
Reactions: Ottonomous
D

dahorns

Senior member
Sep 13, 2013
550
83
91
Hitman928 said:
Doubtful. If you look at the Anandtech benches, when running the 3DPM benchmark it dropped to 1.0 - 1.1 GHz to stay at 15 W. This is an AVX512 load so it will use a significant more amount of power than a more standard load, but even when increasing the cTDP by 67% to 25 W, it only was able to get ~40% higher frequency (1.4 - 1.5 GHz). Obviously non AVX512 loads will use a lot less power / Hz but it seems doubtful that you will be able to get more than ~3x the all core boost rate and still stay under the actual 15 W TDP limit.

Their power test also reveals some tweaks intel made that (IMO) make Ice Lake look better than it probably should compared to current intel mobile CPUs. First, the test system was configured to be able to use up to 50 W of power (even in 15 W mode) for an extended PL2 window. As they mention, most laptops only allow up to 35 W or less. This lets the test system boost higher and longer than it will in most OEM configured laptops, especially any thin and light ones like you would see with a 15W CPU. Anandtech also noted that the test system's fan never stopped running which means it will take longer for it to thermally throttle compared to an OEM configured laptop.

These tweaks will have the biggest effect on single / lightly threaded workloads, especially ones that last only a short amount of time. Longer, well multi-threaded workloads will see the CPU spend most of its time thermally/power limited so you get a better comparison against the OEM laptops which will be in the same situation. In those tests (non AVX512 ones at least), the CPU perf/w comparison shows pretty minimal gains. GPU does show some nice gains in perf but I'll wait to see actual shipping systems with comparable configurations before trying to conclude how much of an increase it is.
Click to expand...

The notebookcheck prime tests show all cores at ~1900 mhz using 14.9W CPU package power. Don't know if that will translate to actual workloads, but at least it is a number for reference.
 
Last edited:
M

majord

Senior member
Jul 26, 2015
433
523
136
dahorns said:
The notebookcheck prime tests show all cores at ~1900 mhz using 14.9W CPU package power. Don't know if that will translate to actual workloads, but at least it is a number for reference.
Click to expand...


as per my last post,


https://www.tomshardware.com/reviews/intel-10nm-ice-lake-test-benchmarks,6257.html

At 15W, the i7-1065G7 started at a peak score of 496.9 dropped into the high 460’s and low 470’s before increasing several points in run 7 and 8 and then settling in the mid to low 470’s. During the test, the processor ran at an average of 2.1GHz.



At 25W, it started at 704.3 points, immediately dropped to 641.2, and then settled around 670 points for the rest of the runs. The CPU ran at an average of 2.8GHz.


this should give some idea of clocks under more real-world all-core workloads
 
  • Like
Reactions: dahorns
B

Bouowmx

Golden Member
Nov 13, 2016
1,138
550
146
The "real" base price IMO can be considered 1400 USD for the model with 4 cores (i5-1035G1). G7 graphics (i7-1065G7) price is 1650 USD, which makes chasing integrated GPU performance incredibly foolish.
 
  • Like
Reactions: Lodix
DrMrLordX

DrMrLordX

Lifer
Apr 27, 2000
21,620
10,829
136
Yup definitely looks like Rocket Lake is getting a 10nm iGPU. Think they're putting SoC functions on that die as well? It'll be interesting to see what topology they use.
 
A

Abwx

Lifer
Apr 2, 2011
10,939
3,440
136
So it does hardly 500pts@15W, i thought that it was 580 or so, but obviously the Intel for controled press in house laptops had turbo power enabled at PL2, such that at 15W it would boost at 22-25W for short term if temp is low enough.

This put things in perspective when we know that at 15W AMD s Picasso score almost 10% better, and KBL is hardly beaten since it s stil in the 460-480pts@15W along with Raven Ridge, dunno why efficency is so low for a new node, since CB is FP this point to FP efficency being downgraded by the AVX 512bit units, just like in the server department, in Integer it shouldnt suffer from such a setback.
 
  • Like
Reactions: Ottonomous
J

jpiniero

Lifer
Oct 1, 2010
14,584
5,206
136
Bouowmx said:
Seeing those Rocket Lake specs (PCIE 4 and DMI x8 suggesting new motherboard), Comet Lake being on its own platform just looks foolish.
Click to expand...

It could just be backwards compatible but not forwards (ie: Rocket Lake boards will work with Comet but not vice-versa).
 
J

jpiniero

Lifer
Oct 1, 2010
14,584
5,206
136
ondma said:
The current XPS 13 starts at 1000.00 already. It is a premium notebook.
Click to expand...

You do get the i5 with the base model on the regular XPS 13 though. That's kind of what I've been saying all along, that 10 nm yields are so bad they don't actually intend to sell that many. Plus, the i5 is the G1 model and not the G4.
 
C

cortexa99

Senior member
Jul 2, 2018
319
505
136
I don't quite know how DMI/PCIE count decide compatibility. Are CML/RKL share same socket with CFL? I find the author who leak this at that Taiwanese website just mentioned that if you use CMP-H PCH board you cannot upgrade to RKL-S CPU...... So current Z370/390 cannot upgrade to RKL-S for sure?

He also mention at last sentence that AMD is a better choice in coming years which is sarcastic......
<這幾年閉著眼睛買AMD應該都沒錯>
Click to expand...
 
N

NostaSeronx

Diamond Member
Sep 18, 2011
3,686
1,221
136
This is a rumor until proven otherwise...
1. The old team was laid off, or was given a heavy reprimand and to work with the new team.
2. The new team was in stealth working on porting 10nm to 7nm.

With that we have 7E, which is like 7-/7 minus, rather than 7 or 7+, etc.
The chiplets have an internal data fabric and an external data fabric in the shapes of rings. Every two-cores in the chiplet have a NoC. The first chiplet connects to the platform chiplet with >5 UPI? connections and connects to second chiplet with >5 UPI? connections. The second connects to the first and the third chiplet, so on.

It will re-use the same platform that uses the 2-die server package, however it doesn't need two dies now... so bunch of logic chips and up to 12x 24 GB HBM2e.

The rumor is a couple months old, so expect a surprise "we aren't launching Icelake server now. Here have 7nm now rather than later."
 
J

jpiniero

Lifer
Oct 1, 2010
14,584
5,206
136
EUV designs appear to be much different. Doing a 7- type design would make a ton of sense as a backup plan, since they really can't afford to be that long behind on moving to EUV compared to the foundries. Would be a lot of work.

It wouldn't surprise me if Icelake Server ends up getting cancelled, especially after the delay.
 
DrMrLordX

DrMrLordX

Lifer
Apr 27, 2000
21,620
10,829
136
jpiniero said:
It wouldn't surprise me if Icelake Server ends up getting cancelled, especially after the delay.
Click to expand...

Icelake-SP? Isn't that coming out soon? Also Intel has two 10nm server chips on their roadmaps - Icelake-SP and Sapphire Rapids. It might make more sense to cancel Sapphire Rapids and push Granite Rapids forward in the product pipeline.
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom