Review Intel 10th Generation Comet Lake-S Review Thread

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
Review information on the soon-to-be-released 10th generation desktop lineup, as well as all relevant information will be linked in this thread. OP will be updated as information becomes available in the next few days. Please, post links to reputable sites you want to see in the OP, and I'll add them. Thanks!

Anandtech
Phoronix (Linux Benchmarks)
LTT (YouTube Video)
Gamers Nexus
Euro Gamer
ComputerBase.de
Back2Gaming
HWUB (YouTube Video)
Sweclockers
Nordic Hardware


Reviews Roundup on VideoCardz
 
Last edited:

ondma

Platinum Member
Mar 18, 2018
2,718
1,278
136
Unless we use the same argument that AMD crowd have been using for a while: more cores let you run more things in the background, stream or whatever.
So if we say that games can use 8 cores now (I don't know, honestly), you get 2 extra cores for whatever else there's going on.

Some will choose Intel for the IGP. Some for other specific advantages (some things just work better on Intel).

In parallel loads 3900X is better.
In sequential (single-thread) tasks 10900K wins.

In a real-world, mixed use - it really depends what you're using.

There's no such thing as lower or higher parallel software. If a program uses 8 out of 10 cores, it means there are 8 single-thread tasks running.
An algorithm is called parallel when it's able to divide a problem into many separate elements that can be run independently. "Many" usually means going down to data granularity.
So if I have 2 numerical vectors with 10^6 elements and I add them, it can be split into 10^6 separate processes.
Or the other AMD fan argument for "moar cores", i.e. "future proofing". With a lot more cpu power in the next gen consoles, the 10900k could well be faster than the 9900k/10700k at some time in the future.
 

uzzi38

Platinum Member
Oct 16, 2019
2,565
5,566
146
Torturing the poor chip with a Y-Cruncher AVX-2 power virus? Sure. I'm surprised it wasn't higher. You've got to give it to the chip for taking all those watts and not going belly up though..

Anyway, it's okay to admit you were wrong, you know. 4.9GHz with only 200w and 84c wasn't exactly what you predicted. Stock temps are just ridiculous:

View attachment 21443

Do you honestly believe 84 degrees on a 360mm AIO for a mainstream consumer chip with what will be stock setting in most cases because OEMs laugh at Intel's spec is a good thing?
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
I wonder if he missed that the time scale of the graph is in milliseconds and not seconds?
I did miss that.....

Personally I'd consider 85 degrees on a 360mm AIO with MCE pushing 4.8GHz (not 4.9 because no TVB at this temp) not great at all.
Really? You're not impressed? Try this....

1590189907278.png

For perspective.
 

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,628
136
I did miss that.....


Really? You're not impressed? Try this....

View attachment 21444

For perspective.

Somewhat. This is with MCE off which means the 10900K should be (I say should because motherboards do weird things sometimes) limited to 130W. Dissipating 130W on a 14 nm process is not hard to do for a 360mm AIO. Heck, when overclocked my 2700 consumes right about that same amount and I get sub 70 degree temps running Prime 95 on a 280 mm AIO. Obviously it won't be as fast as probably even the 10700K, but that's comparing GloFo 12 nm to Intel 14 nm and I think we all know Intel's 14 nm is superior.

The heat transfer bottleneck at 7 nm is different though. It becomes much harder to transfer the heat from the internal hotspots to the heat spreader. So a really good to high end air cooler will perform basically the same as a top of the line water cooler because both are able to dissipate the heat at the heat spreader without issue. It's a matter of the density of the process and getting the heat out of the die to be dissipated. In other words, using a 360 mm water cooler on Zen 2 is a huge waste of money as it won't net you much at all over cheaper water or even solid air coolers. However, using a 360mm water cooler will have an effect when trying to deal with over 200 W of power. Intel will most likely have the same heat density issue whenever they are able to get high performance 10nm/7nm chips out.

AMD could probably thin the dies even further like Intel has done, the problem is when you do that it makes the dies more fragile and you essentially decrease your yield. They could also shrink the spacing of the IHS like Intel but again, this decreases your manufacturing tolerances and inevitably leads to less usable products. I don't think AMD will go this route unless they absolutely have to in order to compete, but they probably won't have to for Zen 3 and then we'll have to see what Intel is able to do with their next couple of rounds of high performance products.

*edited for typos and clarity.
 

tamz_msc

Diamond Member
Jan 5, 2017
3,708
3,553
136
I did miss that.....


Really? You're not impressed? Try this....

View attachment 21444

For perspective.
Maybe for the full perspective you should also include the power draw charts from the same source.
image039-770x580.png


One has to wonder why the power consumption of the 9900K and 9700K is so disproportionately high if MCE is indeed off. Unless this fact in this review can be explained properly, I would not trust the thermal results in this review.
 

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,628
136
Maybe for the full perspective you should also include the power draw charts from the same source.
image039-770x580.png


One has to wonder why the power consumption of the 9900K and 9700K is so disproportionately high if MCE is indeed off. Unless this fact in this review can be explained properly, I would not trust the thermal results in this review.

I don't know about the 9900K/9700K because the reviewer doesn't give any detail on those, but if you look at the 10900K results, the chip reaches 93 degrees before dropping down to 4.1 GHz - 4.2 GHz sustained.

10900K-benchmark_21352_image003.png


10900K-benchmark_21352_image001.png


 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,784
136
Torturing the poor chip with a Y-Cruncher AVX-2 power virus?

Spin spin spin, is that all you do? Careful, you're getting dizzy. Guess how much power a 3900x burns running that "power virus"?

142W

A 3950x? 142W.

It's Intel's fault their ridiculous chip has to burn 250W+ running that benchmark.

I wonder if he missed that the time scale of the graph is in milliseconds and not seconds?

Or the fact that Anandtech never mentioned which workload they used to induce boost behavior so they could measure the rampup time? It's probably just an empty for loop.
 

TheELF

Diamond Member
Dec 22, 2012
3,967
720
126
Spin spin spin, is that all you do? Careful, you're getting dizzy. Guess how much power a 3900x burns running that "power virus"?

142W

A 3950x? 142W.

It's Intel's fault their ridiculous chip has to burn 250W+ running that benchmark.
What do you mean with "has to" ,the sole purpose of a stress test is to determine if your overclock is stable,it doesn't produce anything or show you any performance results.So an intel system is stable even with the CPU drawing 250W+ ,so what?

You can also limit the 10900k to 125W in which case it will burn less than the 3900x and will be clocked at 4Ghz pretty much the same as the 3900x,although to be fair you could also reduce TDP on the ryzen as well.
M7LzD2A.jpg


It's Intel's fault their ridiculous chip has to burn 250W+ running that benchmark.
If you want to see something truly ridiculous just look at how much power the 3900x draws for single thread just to not look completely useless.
it's a 105W TDP chip drawing 100W just on a single core,just as much as the 10900k needs to reach 5.3Ghz.
jGc8D8T.jpg
 

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,628
136
What do you mean with "has to" ,the sole purpose of a stress test is to determine if your overclock is stable,it doesn't produce anything or show you any performance results.So an intel system is stable even with the CPU drawing 250W+ ,so what?

You can also limit the 10900k to 125W in which case it will burn less than the 3900x and will be clocked at 4Ghz pretty much the same as the 3900x,although to be fair you could also reduce TDP on the ryzen as well.
M7LzD2A.jpg



If you want to see something truly ridiculous just look at how much power the 3900x draws for single thread just to not look completely useless.
it's a 105W TDP chip drawing 100W just on a single core,just as much as the 10900k needs to reach 5.3Ghz.
jGc8D8T.jpg

Your first graph is CPU power. The second one is full system power, not CPU power for a single thread. Your point isn't valid.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,784
136
@Zucker2k

Nice non-answer. 250W PL2, tau 56s. 250m y-cruncher lasts less than 56s so it can and will sustain PL2 numbers provided that y-cruncher puts that kind of load on the CPU (which it does, albeit sporadically). Nothing the video says contradicts this analysis. For any run lasting longer than 56s, it slows down. Pretty badly actually.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,432
136
@Zucker2k

Nice non-answer. 250W PL2, tau 56s. 250m y-cruncher lasts less than 56s so it can and will sustain PL2 numbers provided that y-cruncher puts that kind of load on the CPU (which it does, albeit sporadically). Nothing the video says contradicts this analysis. For any run lasting longer than 56s, it slows down. Pretty badly actually.
What I would like to see is a benchmark thread with something sustained, twofold. Performance and power draw. Something like Folding@home, ppd for performance and temp would be measured during the run. You would only need like 20 minutes. That would give you sustained load temp. And maybe using 3900x and 10900k and then the exact same 240mm aid. That would be one not to argue with, but "average sustained load"
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,784
136
What I would like to see is a benchmark thread with something sustained, twofold. Performance and power draw. Something like Folding@home, ppd for performance and temp would be measured during the run. You would only need like 20 minutes. That would give you sustained load temp. And maybe using 3900x and 10900k and then the exact same 240mm aid. That would be one not to argue with, but "average sustained load"

Blender 1hr render puts the 10900k @ 4.3 GHz, 125w. Give or take.
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
Blender 1hr render puts the 10900k @ 4.3 GHz, 125w. Give or take.
So what's your issue exactly, Intel's stock turbo boost is too aggressive for your liking? A lot of the reviews are showing that 250w+ only happens under extreme cases involving AVX-2 and stress-testing.
 
  • Like
Reactions: pcp7

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,478
14,432
136
So what's your issue exactly, Intel's stock turbo boost is too aggressive for your liking? A lot of the reviews are showing that 250w+ only happens under extreme cases involving AVX-2 and stress-testing.
MY problem, is that they pick and choose. They use high power to get a good benchmark, and low power to show low power usage. You can't do both. If you get the performance out of this chip, it DOES take a lot of power. Nobody seems to acknowledge that. Nobody that likes Intel. And when it takes a lot of power, it takes a lot to cool it, as it is HOT.

The 125 watt power ? that would provide a very bad benchmark.
 
  • Like
Reactions: Tlh97 and lobz

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
So what's your issue exactly, Intel's stock turbo boost is too aggressive for your liking? A lot of the reviews are showing that 250w+ only happens under extreme cases involving AVX-2 and stress-testing.
No, the problem is the hypocrisy of always taking the one single data that suits your argument, and when challenged, contradicting the challenge with data taken from another workload/setting/review/planet, and when that gets pointed out, going back to the first data, and this whole thread has been a pointless carousel because of that so far.
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
MY problem, is that they pick and choose. They use high power to get a good benchmark, and low power to show low power usage. You can't do both. If you get the performance out of this chip, it DOES take a lot of power. Nobody seems to acknowledge that. Nobody that likes Intel. And when it takes a lot of power, it takes a lot to cool it, as it is HOT.

The 125 watt power ? that would provide a very bad benchmark.
my thoughts exactly
 
  • Like
Reactions: Tlh97 and Makaveli