• 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."
  • Community Question: What makes a good motherboard?

Discussion Comet Lake Intel's new Core i9-10900K runs at over 90C, even with liquid cooling TweakTown

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

DrMrLordX

Lifer
Apr 27, 2000
16,632
5,639
136
But I think specially with this guy it will be important for the reviewers to figure out what "adequate" cooling will be like comparitively.
Someone needs to review it with both a good 280mm AiO (360mm aren't much better) and an NH-D15 or Dark Rock Pro 4 and compare results.
 
  • Like
Reactions: guachi and Elfear

Magic Carpet

Diamond Member
Oct 2, 2011
3,263
74
91
No new multi-threaded games come anywhere close to fully utilising 10 cores. We're at the point where some new games actually use 5-6 cores, no more.
We're so GPU limited most of the time, especially at higher resolutions, buying 10 cores for gaming only is silly, I quite agree. I'd rather get 8 w/ higher ST performance.
 

mopardude87

Diamond Member
Oct 22, 2018
3,348
1,563
96
Oh man in AND OUT before all the other AMD owners with 3900x+ processors go about ranting about Intel putting holes in the ozone layer. The title of the thread sure got my interest.
 
Last edited:

aigomorla

Cases and Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
18,291
986
126
i don't think you guys can cross compare thermals on both processors.
Intel has a direct core temp, Ryzen does not, unless someone can point me to a program that does.

Also Intel uses Tjunct, values( which is also why you can get some really funny values when you go down really low), while i believe ryzen uses direct thermals.

Its like comparing apples to oranges, because if one reads 90C does not correlate to 90C on the other platform.
 

IntelUser2000

Elite Member
Oct 14, 2003
7,249
1,839
136
Of course it runs that hot.

Remember Pat Gelsinger's presentation in the early 2000s? He said pursuing for ever higher clocks will result in CPUs with the heat density of the surface of the sun. Back in those days we could push our CPUs to 5GHz too - using exotic cooling.

Now we essentially do the same, but absolutely huge air coolers and water coolers have reached mainstream.

Nothing has changed. This is why we won't ever see 6GHz CPUs without a probable 500-700W power rating. This is Netburst generation 2.

Hopefully Intel backs down the clock to 4.5GHz range once they get their steam back. The rest has to be made with uarch improvements.
 

maddie

Diamond Member
Jul 18, 2010
3,363
2,277
136
These power levels must have very high current flows. Significantly higher amperage than a couple years ago. I keep thinking electromigration.

Am I going full paranoid if I begin to suspect that Intel might be sacrificing CPU lifetime to hold on for year or two while they recover? Sort of borrowing against the future, so as to survive today.
 

Falkentyne

Junior Member
May 3, 2010
5
2
81
These power levels must have very high current flows. Significantly higher amperage than a couple years ago. I keep thinking electromigration.

Am I going full paranoid if I begin to suspect that Intel might be sacrificing CPU lifetime to hold on for year or two while they recover? Sort of borrowing against the future, so as to survive today.
That high amperage is spread across multiple cores. Not one core.
4 core 7700k @ 100 amps is the same as 8 core 9900k @ 200 amps.
 

piokos

Senior member
Nov 2, 2018
554
202
86
i don't think you guys can cross compare thermals on both processors.
Intel has a direct core temp, Ryzen does not, unless someone can point me to a program that does.

Also Intel uses Tjunct, values( which is also why you can get some really funny values when you go down really low), while i believe ryzen uses direct thermals.

Its like comparing apples to oranges, because if one reads 90C does not correlate to 90C on the other platform.
Exactly.
The only thing worth looking at is distance to thresholds (for boost or the Tmax limit that causes throttling), because that's what affects performance.

People spend way too much time worrying about temperatures ("my CPU hits 90*C - will it melt the butter wires?"), not to mention comparing them between platforms. It literally makes as much sense as comparing cooler noise figures between reviews that measure using different microphones and different distance to the fan.

As for the 10900K - the fact that it only reaches 90-95*C at PL2 (250W) with a modest AiO is pretty much a miracle.
It means one can push this CPU to the limit with cooling that costs 20% of CPU price.
Nothing here will improve until Intel is able to shift down to another node, 14nm has reached its utter limits. All while at the same time this is actually astounding that Intel were able to do this (up to 5.3GHz with 10C/20T) all on a very aged, very run down 14nm node.
Assuming identical performance, a more dense node (be it 10nm or 7nm) will result in lower power consumption, but very likely higher temperatures. Just saying. ;)
 

FaaR

Golden Member
Dec 28, 2007
1,057
408
136
At some point the Lakes really have to have dried up.
Especially as Intel drank coffee lake in desperation already trying to fix their 10nm process, then drowned their sorrows in whiskey lake... :p

Anyway, as a PC enthusiast I don't really worry about power consumption in a high-end chip; we know it's going to draw a lot of power and it's going to run hot. If it doesn't at this stage in computing history, then something's basically wrong. It's like complaining that a newly-bought super sports car is loud and thirsty; you knew that would be the case before purchase, so why even bother over it?

It was different back in the 1990s when CPUs were measured in tens of millions of trannies only at most, not in the billions. :p Now there's not even any way to reach max performance without hitting the thermal cap; the thermal cap IS the limit for max performance, without resorting to phase-change refridgeration cooling or something even more exotic.

But that's alright. Don't worry about it! :)
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,885
9,080
136
Especially as Intel drank coffee lake in desperation already trying to fix their 10nm process, then drowned their sorrows in whiskey lake... :p

Anyway, as a PC enthusiast I don't really worry about power consumption in a high-end chip; we know it's going to draw a lot of power and it's going to run hot. If it doesn't at this stage in computing history, then something's basically wrong. It's like complaining that a newly-bought super sports car is loud and thirsty; you knew that would be the case before purchase, so why even bother over it?

It was different back in the 1990s when CPUs were measured in tens of millions of trannies only at most, not in the billions. :p Now there's not even any way to reach max performance without hitting the thermal cap; the thermal cap IS the limit for max performance, without resorting to phase-change refridgeration cooling or something even more exotic.

But that's alright. Don't worry about it! :)
Ryzen 3000 series does not run hot or take a lot of power, this is what many of us are now used to, so yes, its a big deal that its hot and power hungry,.
 
  • Like
Reactions: DarthKyrie

aigomorla

Cases and Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
18,291
986
126
Ryzen 3000 series does not run hot or take a lot of power,
FALSE

These guys run rediculously hot due to the 7nm arch.
Less footprint, same heat output makes it a spot welder for the heatsink.

As for power, i was seeing modest power draw.
It was not WOW, and at the same time not OMG.... keep note this is not the X version which runs hotter, and draws more power.

If you want to compare power draw which one can, the "T" line of intel chips (i5-9600T) will probably run circles around Ryzen 5 in power draw and heat output, but lose out in performance.

But hey were looking at at TDP of 35W(intel) vs 65W(amd)
And there will probably be a T line series for this gen cpu as well.
 
  • Like
Reactions: killster1

coercitiv

Diamond Member
Jan 24, 2014
4,041
4,649
136
FALSE

If we evaluate "hot" CPUs based on their stock cooler performance then Intel CPUs have been running HOT ever since Coffee Lake. Stock coolers are made to dissipate 65W closer to max CPU temps and the only chips running cold beneath that small mass of metal will be those that actually use less than 60W under load.

Techspot even has a somewhat apple to apples comparison between Intel and AMD stock coolers (with the caveat that testing conditions make the comparison valid only between their results since both fan speed and airflow were limited).
Temps.png

See what happens when metal mass increases? Max temps drop fast: from 90-95C with puny stock coolers to 77C with a chunkier piece of aluminum.

Is it the hot 7nm arch or simply a severe lack of metal in the cheapest stock coolers?
 

brianfcoleman

Junior Member
Apr 19, 2020
2
6
36
So I'm a software engineer and one of the things I do is compile Chromium on Windows, a very large C++ project. My workstation has an i7 8700 and I did an experiment about a year ago to measure the impact of various PL1 and PL2 power settings on compilation times. The i7 8700 has a base clock of 3.2GHz and an all core turbo of 4.3GHz according to Wikipedia Coffee Lake article. I compiled a debug build of Chromium in each case.

I found that with my motherboard, an AsRock Z370m itx/ac, the default PL1 and PL2 setting was 120W and the compilation time was about 30 minutes. I use a low profile cooler, a Noctua NH-L9i. 120W is pushing that cooler to its limits.

When I reduced PL1 and PL2 to 65W, the compliation took 33 minutes, about 10% longer. The CPU was still able to clock to 3.8GHz on all cores within this power envelope, which is 88% of the theoretical maximum stated by Intel of 4.3Ghz without overclocking.

I was very happy with this result and I believe that it shows that in the case of my chip, increasing the power consumption by a factor of 2 delivered very much less than a linear gain in performance.
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
16,632
5,639
136
@coercitiv

I think what he means is that Matisse has some serious hotspots. It's actually very hard to reduce hotspot temps without a massive shift in delta T (active cooling).
 

coercitiv

Diamond Member
Jan 24, 2014
4,041
4,649
136
I think what he means is that Matisse has some serious hotspots. It's actually very hard to reduce hotspot temps without a massive shift in delta T (active cooling).
If that is the case, then not all Ryzen 3000 chips are created equal - there's simply too much of a difference from a 58C idle reported in that thread (assuming fans were spinning at that temp ) and the 36-40C seen by the Techspot reviewers. I believe @VirtualLarry also had issues with idle temps during his R5 3600 troubleshooting, although I did not follow that thread to it's conclusion (if any).
 

aigomorla

Cases and Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
18,291
986
126
I think honestly as things get smaller, we need a better X axis heat transfer so more of the IHS can spread the heat out on more heat pipes.
As it is, the heat zone on smaller nodes is so tiny, your going to have massive deltas from center to edge.
By having better X - Axis heat transfer like graphene for example, you could move the delta out smaller from edge to center, and then have those monster heat pipes we have on sinks to work more efficiently.

Otherwise y-axis thermal transfer will need to get more and more efficient, which again, unless we shift to a different medium, were probably already at the edge of diminished return catigory.
 

moinmoin

Platinum Member
Jun 1, 2017
2,074
2,479
106
it shows that in the case of my chip, increasing the power consumption by a factor of 2 delivered very much less than a linear gain in performance.
That's indeed the case, the most power efficient frequency for both Intel's and AMD's desktop chips are far below what they set as stock frequency nowadays.
 

shiznit

Senior member
Nov 16, 2004
422
13
81
I'm happy AMD is doing well but I don't understand all the hate about Intel's temperatures. If a 5.3 GHz CPU configured for top single threaded performance and memory latency crosses 90C on some other (likely synthetic workload) you have, it probably wasn't made for you. It's still a great value for those using 144-240hz monitors who want the gaming performance.
 
  • Like
Reactions: Arkaign

FaaR

Golden Member
Dec 28, 2007
1,057
408
136
Ryzen 3000 series does not run hot or take a lot of power
You need to define those weasel words within some sort of bounds of comparison if you want me to buy what you're selling... My recollection is the high-end 3000 series does drink quite a lot of juice when running all cores pegged at top clock - as you would expect. Basically if you're not pulling down the amps you're leaving performance on the table.

It's totally anecdotal of course, but one review from launch last year I checked just now to refresh my memory ran a Ryzen 3950X 16-core first with unlocked power budget using precision boost at which point the processor pulled ~185W in a stress test, then at 4.35GHz all cores @ 1.35Vcore with an AIO watercooler, CPU hit 90C with radiator fans on full and burned ~220W. That's not exactly peanuts I'd say, but again, expected, because it's a high-end processor being pushed. Let's not pretend AMD chips run on angel farts and don't produce any significant heat, that's not doing anyone any favors. It's just silly and contrary to reality.

so yes, its a big deal that its hot and power hungry,.
No it isn't. High-end chips drawing a lot of power is totally not a big deal, or unexpected. It's more like it's been for like...the past decade or so now. And future chips regardless of manufacturer isn't going to get significantly better either since process shrinks aren't giving us the kind of power savings they used to. Besides, we'll eat up those savings by throwing in more transistors there instead for more performance and more cores, along with faster I/O buses that draw more power than the old ones did.

PCIe gen 4 increases power draw significantly versus gen 3, and clocking up your DIMMs does the same thing. On my own 7900X, idle CPU power draw goes up at least 10W just from raising memory bus clock from whatever is standard (2166MT, maybe) to the 3600MT spec of the DIMMs.

It's the price we pay for More Fun. Just accept it, and move on.
 

ASK THE COMMUNITY