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

ondma

Golden Member
Mar 18, 2018
1,591
458
106
I said "if", not "when". Learn to read.

You and I both know that Intel has no business launching 400W parts on server sockets in this day and age. Cooper Lake was kicked off the 2P platform for a reason.
Then why not wait for your criticism until you find whether it is true or not?
 

DrMrLordX

Lifer
Apr 27, 2000
16,506
5,482
136
Then why not wait for your criticism
Nobody else does. Stop being so sensitive. It's tedious.

To the best of my knowledge, Intel hasn't teased any 10nm server parts over 270W for the Ice Lake generation. The mere fact that anyone considers a 400W 10nm SKU to be credible speaks volumes about Intel right now.
 

uzzi38

Golden Member
Oct 16, 2019
1,166
1,972
96
The mere fact that anyone considers a 400W 10nm SKU to be credible speaks volumes about Intel right now.
To be fair it's a pretty old rumour now, there's absolutely no guarantee it will happen especially not - like you said - when you consider that 2P Cooper got canned and CL-AP has had minimal traction.
 

DrMrLordX

Lifer
Apr 27, 2000
16,506
5,482
136
To be fair it's a pretty old rumour now, there's absolutely no guarantee it will happen especially not - like you said - when you consider that 2P Cooper got canned and CL-AP has had minimal traction.
My expectation is that 38c 270W Ice Lake-SP will be the top SKU they release, assuming they actually release it. Trying anything bigger that chews up more power would be completely insane unless a big customer specifically wants the SKU for internal use only.
 

FriedMoose

Member
Dec 14, 2019
48
28
51
Per APISAK on twitter the 1165G7's Geekbench 4 score is 23% above the 3900XT in both FP and INT:




Edit: ST FP and INT scores are about 35% higher than the best 4700U:


Edit 2: Comparing multi-core FP and INT

1165G7: 26823 INT 27160 FP

4700U: 29612 INT 29285 FP

4800U: 35377 INT 36339 FP


The 4800U leads the 1165G7 by 32% in INT and 34% in FP, which matches up almost perfectly with the difference in 3DMark CPU scores (34%):

 
Last edited:

Bouowmx

Golden Member
Nov 13, 2016
1,043
419
116

WCCFTech, but the author is reporting the flurry of recent Core i7-1165G7, and i5-1135G7 too, Geekbench benchmarks, and doing a comparison to Ryzen 4000U series.

Core i5-1135G7, a more commonplace processor tier, makes a good showing on Geekbench 4: 6049 single, 21469 multi. Compared to Ryzen 5 4500U and 4600U, ~20% lead in single-core score, and for multi-core, 6% lead over 4500U and ~4% deficit to 4600U. Taking a look at the details of the scores, Tiger Lake's multi-core is propped up by high crypto and memory score. If just looking at integer and floating-point, the multi-core score is tied with 4500U, and ~10% deficit to 4600U.

And just to throw more in, top Core i5-10210U scores are (not achieved in same result) 5217, 16854.
 
  • Like
Reactions: lightmanek

FriedMoose

Member
Dec 14, 2019
48
28
51

WCCFTech, but the author is reporting the flurry of recent Core i7-1165G7, and i5-1135G7 too, Geekbench benchmarks, and doing a comparison to Ryzen 4000U series.

Core i5-1135G7, a more commonplace processor tier, makes a good showing on Geekbench 4: 6049 single, 21469 multi. Compared to Ryzen 5 4500U and 4600U, ~20% lead in single-core score, and for multi-core, 6% lead over 4500U and ~4% deficit to 4600U. Taking a look at the details of the scores, Tiger Lake's multi-core is propped up by high crypto and memory score. If just looking at integer and floating-point, the multi-core score is tied with 4500U, and ~10% deficit to 4600U.

And just to throw more in, top Core i5-10210U scores are (not achieved in same result) 5217, 16854.
If you notice the 4800U is running Linux while the 1165G7 is running Windows. Linux is usually 10% faster. If you take the best 4800U Windows score the difference is >30%.

 
Last edited:

Carfax83

Diamond Member
Nov 1, 2010
5,929
705
126
Although I hate Geekbench, it's all we have right now. That said, its an amazing score for Tiger Lake and Intel has shown remarkable progress on their 10nm+. This bodes well for Alder Lake which will use 10nm++, and could very well hit 5ghz+ speeds!

Just goes to show how advantageous process node is to CPU performance.

Also, this is why I have always said that Zen 3 will be the real game changer for AMD if it's going to happen. Zen 2 brought parity more or less with Intel's Skylake CPUs, but Zen 3 could potentially deliver a critical strike against Intel.
 

mikk

Platinum Member
May 15, 2012
2,916
732
136
These scores are really good and confirms Tigerlake i7-1165G7 can boost up to 4.7 Ghz. The lower end i5-1135G7 can boost up to 4.2 Ghz which is similar to CML-U i5, and also comes with G7 graphics. Not many people expected such a CPU improvement with the same amount of cores.
 

IntelUser2000

Elite Member
Oct 14, 2003
7,175
1,746
136
At least for Geekbench the 1165G7 is 24% faster than the 1065G7 Windows-to-Windows best result comparison using Int and FP scores.

There will be cases where the 4700/4800U will have a greater advantage. In Geekbench ST 7W Amberlake performs nearly as well as the 25W U CPUs(16% difference) while in reality its significantly behind(45-50%). So Geekbench doesn't fully stress the CPUs or the tests are less realistic.

4700U shows 50-60% gain over the 1065G7 while in GB its less at 30-35%.
 

TheGiant

Senior member
Jun 12, 2017
727
328
106
At least for Geekbench the 1165G7 is 24% faster than the 1065G7 Windows-to-Windows best result comparison using Int and FP scores.

There will be cases where the 4700/4800U will have a greater advantage. In Geekbench ST 7W Amberlake performs nearly as well as the 25W U CPUs(16% difference) while in reality its significantly behind(45-50%). So Geekbench doesn't fully stress the CPUs or the tests are less realistic.

4700U shows 50-60% gain over the 1065G7 while in GB its less at 30-35%.
the first tick single thread gain is ok, if not very good
it all comes out to what is the all 4 core sustainable boost, there icelake fails

any other benches than this crap of GB?
 

mikk

Platinum Member
May 15, 2012
2,916
732
136
Core i5-1135G7
Compute Units 80
Maximum Frequency 1.30 GHz

Notebookcheck was correct Core i5 get 80 EUs, even though it comes with G7 branding. The iGPU boost of 1300 Mhz is a high one for Intel, usually Intels ULV models topped out at 1150 Mhz in previous generations. I guess this is a result of the improved 10nm process. They are still behind Renoir/Vega @ 7nm regarding iGPU clock speeds, this is something they could improve further for 10++ with ADL-P.
 
  • Like
Reactions: Dayman1225

IntelUser2000

Elite Member
Oct 14, 2003
7,175
1,746
136
the first tick single thread gain is ok, if not very good
it all comes out to what is the all 4 core sustainable boost, there icelake fails
The MT performance might end up with a further lead over Icelake. For the 1135G7 it ends up with 9% gain in MT and 14% gain in MT over 1065G7. There is also an earlier result indicating the same.

Notebookcheck was correct Core i5 get 80 EUs, even though it comes with G7 branding.
Yup. Though I don't think they got all the SKUs. Core i3 is shown as a 48EU while they have the Core i3 as having a G4 SKU. G4 could end up being a 64EU one.

Clocks are pretty good even against Vega, since it has much less compute units. Remember, its Xe-LP, while discrete versions use Xe-HP.

The Pentium and Celerons are supposed to get a 48EU G1. So even the low end is going to beat Icelake in graphics.
 
Last edited:

jpiniero

Diamond Member
Oct 1, 2010
8,407
1,439
126
You really should be comparing it to the 1068NG7 (and the AMD H parts), although there are no Windows GB4 results of the 1068NG7 on there.

Need more data.
 

IntelUser2000

Elite Member
Oct 14, 2003
7,175
1,746
136
You really should be comparing it to the 1068NG7 (and the AMD H parts), although there are no Windows GB4 results of the 1068NG7 on there.
Not comparing against parts with higher TDP. Doubt manufacturers will set it higher than 25W PL1 anyway.

Update: Don't use Geekbench's OpenCL results for comparing graphics. The Iris Plus 655 which is a Gen 9 GT3e is just as fast as Iris Plus G7 in Icelake and beats Tigerlake's score.
 
Last edited:

uzzi38

Golden Member
Oct 16, 2019
1,166
1,972
96
Not comparing against parts with higher TDP. Doubt manufacturers will set it higher than 25W PL1 anyway.

Update: Don't use Geekbench's OpenCL results for comparing graphics. The Iris Plus 655 which is a Gen 9 GT3e is just as fast as Iris Plus G7 in Icelake and beats Tigerlake's score.
Iris Plus 655 scores are weird, most Windows scores are in the ~40k range, but then you have the odd few that score high 50s and occasionally 60k. Not sure what's up with that.
 

mikk

Platinum Member
May 15, 2012
2,916
732
136
OpenCL drivers can have a big effect on the scores, I know this from Gen9. Also some architectures may be less optimized for computing workloads while better for gaming. These OpenCL scores are not that interesting at the moment.
 

jpiniero

Diamond Member
Oct 1, 2010
8,407
1,439
126
Not comparing against parts with higher TDP. Doubt manufacturers will set it higher than 25W PL1 anyway.
Isn't the default PL1 on Tiger Lake U 28W? Could be a Geekbench reporting error as to what the actual base is though.

If you look, the person testing ran it a bunch of times where the performance is much lower. You can't tell if it's due to throttling because of power/heat or the user though.
 

IntelUser2000

Elite Member
Oct 14, 2003
7,175
1,746
136
Isn't the default PL1 on Tiger Lake U 28W? Could be a Geekbench reporting error as to what the actual base is though.
Default doesn't have much meaning nowadays.

The Amberlake chip has a TDP of 5W, with cTDPup of 7W. Guess what Dell and HP uses? 9W.

Icelake has a 15W TDP with cTDPup of 25W. I've seen 15W, 18W, 25W, 28W, and 46W(temporary). Yes that's all PL1.

TDP isn't fixed either. The Icelake Dell convertible is set at 46W when thermal headroom exists and on AC. But is 25W on battery power. Good chance the Tablet mode is even lower, perhaps 15W.

By the way the Ryzen H goes way higher. It consumes 50-60W in some systems. 35W my ass.
 
Last edited:

Hitman928

Platinum Member
Apr 15, 2012
2,968
2,568
136
I thought Renoir HS was 35 watts and the regular H parts were 45 watts?
Yes. Of course it will boost higher (up to 65 W I believe) but will settle at 45W sustained power. It does have a cTDP of up to 54W though so it can use more if the OEM configures it as such. By contrast the 10980HK has a 45 W TDP and can use up to 135 W during boost and OEMs can configure it with a cTDP up to 65W.
 
  • Like
Reactions: spursindonesia

ASK THE COMMUNITY

TRENDING THREADS