Intel Comet Lake Thread

Page 20 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

moinmoin

Diamond Member
Jun 1, 2017
4,944
7,656
136
Time for me to eat some crow, I never believed they would end up doing this crap.

These are not 125W TDP CPUs, and from now on Intel's TDP numbers are just marketing names. What a sad legacy for a CPU architecture that was once the best at everything.
Maybe it is just me but that was quite naive from you.
To me writing was on the wall with 8 and especially 9 series.
And I have to agree with you, skylake is way past its time.
It's only 17 months since Anandtech felt it's necessary to explain Intel's TDP and Turbo. It's only 9 months since they had an interview with Intel's "Chief Architect of Client Performance Segmentation". Logic and reason clearly flew out of the window while some didn't (want to) look.
 
  • Like
Reactions: RetroZombie

Gideon

Golden Member
Nov 27, 2007
1,619
3,643
136
Here is the reminder of Intel power draw from an older, more civilized age (with a board pumping a little too much voltage into to the CPU)

76803.png
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
What has AMD done in that time span? Maxing out their chips? Intel doesn't have a lot of options, but push their chips as well. It's all about "boost." Nowadays, the mobo manufacturers also have a say how long those boosts can last. Well, competition.
 

blckgrffn

Diamond Member
May 1, 2003
9,120
3,048
136
www.teamjuchems.com
What has AMD done in that time span? Maxing out their chips? Intel doesn't have a lot of options, but push their chips as well. It's all about "boost." Nowadays, the mobo manufacturers also have a say how long those boosts can last. Well, competition.

:oops:

I was joking before about these being these being the Intel 9xxx FX processors, AMD is almost completely flipped the tables now. Their chips are largely running much closer to TDP numbers and can mostly be effectively cooled with air while giving same/better performance across a wide range of applications.

Combined with motherboard & cooler prices to even use the promised performance these CML processors promise you are looking at a very expensive solution that offers little more than we already have.

I guess I am not getting the same take away from this situation that you are.
 
  • Love
Reactions: spursindonesia

SPBHM

Diamond Member
Sep 12, 2012
5,056
409
126
10600Kf with a little OC is probably hard to beat for gaming value, 10400F looks very interesting also..

and the 10700Kf, will be interesting to see against the 3700x.

i3 finally with 4c/8t is also pretty good tbh.
 
  • Like
Reactions: lightmanek

AtenRa

Lifer
Feb 2, 2009
14,001
3,357
136
Ahh nice, competition at last.

For gaming the Core i3s are excellent in price/performance and very competitive against everything AMD has in the same segment.
The new Core i5 10400F is very competitive against the Ryzen 3600 both in games, MT loads and price.

AMD needs to react with price cuts or to launch ZEN 3 soon.
 

Ajay

Lifer
Jan 8, 2001
15,406
7,833
136
But the review is out:

And yep, the PL2 is 250W, but mobo builders built for 320-350W!

In comparison:

32 core Threadripper 3970X never exceeds 280W without overclocking
64 core Threadripper 3990X never exceeds 280W without overclocking

16 core Ryzen 3950X never exceeds 145W without overclocking
8 core Ryzen 3770X never exceeds 88W without overclocking


The $800 Gigabyte Z490 AORUS XTREME is rated for 1440 AMPS!!! I can't stop laughing. [That would be 1.7KW at a Vcore of 1.2v].
 

AtenRa

Lifer
Feb 2, 2009
14,001
3,357
136
10600Kf with a little OC is probably hard to beat for gaming value, 10400F looks very interesting also..

and the 10700Kf, will be interesting to see against the 3700x.

i3 finally with 4c/8t is also pretty good tbh.

10700F should be exactly what the 9900K is.
 

Ajay

Lifer
Jan 8, 2001
15,406
7,833
136
10700F should be exactl what the 9900K is.
At least the price is a little better. If the 9900K was priced the same, I probably would have bought it over my 3900X (which, honestly, is a bit of overkill for my current needs).
 
Apr 30, 2020
68
170
76
At some point you just start straight lying about your TDP. Like, I get how CPUs can exceed TDP, but this is just absurd. This is not a 125w CPU. It's a 250w CPU (that apparently can hit 300w according to the mobo manufacturers!)

At least Ryzen 3000 CPUs stick to the PPT limit, which is typically TDP + 30%. The cores themselves usually stay within the TDP limit and the total package power including uncore will be within the PPT limit. AMD could probably squeeze a few more MHz too if they let a 3800X scream away at 1.5v, 300w all-core boost.
 

OriAr

Member
Feb 1, 2019
63
35
91
At some point you just start straight lying about your TDP. Like, I get how CPUs can exceed TDP, but this is just absurd. This is not a 125w CPU. It's a 250w CPU (that apparently can hit 300w according to the mobo manufacturers!)

At least Ryzen 3000 CPUs stick to the PPT limit, which is typically TDP + 30%. The cores themselves usually stay within the TDP limit and the total package power including uncore will be within the PPT limit. AMD could probably squeeze a few more MHz too if they let a 3800X scream away at 1.5v, 300w all-core boost.
It's 125W CPU at base clocks, which is how TDP is calculated.
BTW, I don't expect it to draw 250W at stock outside torture tests really, power consumption will probably be around 200W while boosting on all cores. (Just like the 9900K very rarely gets above 150W).
 
Apr 30, 2020
68
170
76
It's 125W CPU at base clocks, which is how TDP is calculated.
BTW, I don't expect it to draw 250W at stock outside torture tests really, power consumption will probably be around 200W while boosting on all cores. (Just like the 9900K very rarely gets above 150W).
I understand how TDP is calculated. However in this case, nothing appears to be limiting the power during boost. The motherboard manufacturers all seem to be on board ready to cram 300w+ into the CPU during max boost. It basically just seems that as long as the CPU stays cool enough, it will plow as much power as the board will feed it.

At least AMD CPUs stick to a hard PPT limit even while boosting.
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
14,328
4,913
136
Hard pass from me. Moving to a warmer climate so want as much efficiency as possible.

14nm+++(+) clearly doesn't deliver.
 
  • Like
Reactions: lightmanek

RetroZombie

Senior member
Nov 5, 2019
464
386
96
Intel's TDP has always been rated at base clocks, nothing shady there.
It's 125W CPU at base clocks, which is how TDP is calculated.
But the cpu doesn't run at base clocks, that's where the shady relies.

These are not 125W TDP CPUs, and from now on Intel's TDP numbers are just marketing names. What a sad legacy for a CPU architecture that was once the best at everything.
And unfortunately it seam it will end up remembered as one of the worst ever.
 
  • Like
Reactions: Zepp

DrMrLordX

Lifer
Apr 27, 2000
21,609
10,803
136
What has AMD done in that time span?

First off, this is a Comet Lake thread. So I don't know why you're bringing them up. Trying to throw somebody else under the bus doesn't make anything Intel is doing better in comparison. Some people call this 'whataboutism' though in truth, you're committing classic Tu Quoque:


Secondly, what has the competition done? Introduce new architectures on new nodes and sold them throughout their entire product lineup. It's completely irrelevant as to whether or not some competitor's uarch is "maxed out". They are gaining in performance year by year while Intel is not, at least not in their halo desktop CPUs.

AMD needs to react with price cuts or to launch ZEN 3 soon.

They already cut prices. You can get an R5 3600 for the same price, and the boards are cheaper. We've gone over this in other threads. Intel is releasing a product that is barely competitive with something that's been out since last July.

At some point you just start straight lying about your TDP.

Intel was also lying about the 9900k. It was really a ~160W TDP CPU in default configurations.

It's 125W CPU at base clocks, which is how TDP is calculated.

It will never run at base clocks unless you change settings in the UEFI.

No, they couldn't. Zen 2 is maxed out.

All they had to do was use the high-performance libraries. One little tweak and it's on. Not that they're going to bother, because instead of wasting time retooling an old design over and over again, they're just gonna release something new this year. Comet Lake is in serious trouble.
 

RetroZombie

Senior member
Nov 5, 2019
464
386
96
Hard pass from me. Moving to a warmer climate so want as much efficiency as possible.
That's why such products should be forbidden to be released.

With global warming on the agenda how some company can get away with fake specs and no organization in the world have the power to cancel it or force it to release the correct technical data.

We need the European Union energy label system on cpus as soon has possible, where Intel would get a D, the worst rating.

Intel cpus are like chinese laptop replacement power adapters, they say it can handle 120Watts but burn out after reaching just 65Watts.
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
They are gaining in performance year by year while Intel is not, at least not in their halo desktop CPUs.
Fact check: The 10900K brings 25% more in performance in core count alone, not counting TVB, and All-Core clocks, and no price increase. There are people who prefer Intel over AMD, and for those people the 25%+ performance will be well worth their investment.
 

gdansk

Platinum Member
Feb 8, 2011
2,078
2,559
136
We knew it was a new socket but for me that's the most disappointing part. That and the Z490 motherboards with PCIe 4.0 support without any CPU that will support it. Might still buy one of the 35W parts. Is the PL2 listed for those?
 

DrMrLordX

Lifer
Apr 27, 2000
21,609
10,803
136
Fact check: The 10900K brings 25% more in performance in core count alone

Fact check: IT'S THE SAME DAMN UARCH. This is the third time all they've done is strap extra cores onto an old design. They are not offering any more per-core performance, and they can't increase clockspeeds anymore either. Here you are crowing about increaesed MT performance? Let me put this in perspective for you:

7700k->8700k: 50% increased MT (theoretical)
8700k->9900k: 33% increased MT
9900k->10900k: 25% increased MT

Thanks to Amdahl's law you are only going to get the full increase in performance in rare cases. The more cores you add, the worse it gets. At least with the 8700k and the 9900k they were able to squeeze out some more clockspeed (at a price). Now they're stuck.

not counting TVB, and All-Core clocks

TVB is garbage. We've already seen it on the 9980HK, and it is, in a word, pathetic. You are not going to see 5.3 GHz in real-world applications. As for '"all core clocks", they're just overclocking the CPU and expanding its already-ridiculous power budget. For someone that was running a 5 GHz 9900k, they are not getting anything new or better. For anyone that hesitated to run a 9900k @ 5 GHz (or higher) due to power output and cooling requirements, the 10900k brings them closer to the point that they can no longer cool the chip running at bone stock (250W or more!!!). Throttle throttle throttle. This effect will be seen up and down the entire product stack. Intel can't improve the voltage/clockspeed curve any further, nor can they realize any better IPC. So they have to expand power budget to raise all-core clocks.

@gdansk

I would look at some of the existing 35W Coffee Lake parts and see if you can find the PL2 values for those. They should be about the same. Intel has expanded TDP for much of the lineup with Comet Lake-S which is why the PL1 and PL2 values are going up.
 
Last edited: