Intel Enthusiast-Grade K Processors in the Comet Lake-S Family Rumored to Feature 125 W TDP

[DeathtoGnomes]

Now that this thread is mostly dead, I'd like to thank everyone for the TDP "discussion", it was a very "pass the popcorn" read. I do think that building around heat is a very niche market, which means 99% of everyone else buys a cpu and then buys an appropriate cooler, not the other way around. I for one dont care about the TDP as long as the chip does its job and wears the right hat.

 

[Darmok N Jalad]

I meant what I said, and I originally meant that I thought the 10 core cpus wouldn’t have iGPUs. I may have been wrong if those leaks are correct. And the reason many people don’t know how QuickSync works is because it’s useless to them, as they usually are gamers. For normal people, iGPU is a rather handy thing to have, it saves a bunch of electricity.

When I had a 27” iMac, I noticed Apple still used the IGP when streaming video, even though there was a dedicated GPU. Probably did so for the same reason.

 

[AnarchoPrimitiv]

I think the most astonishing phenomenon here is how common it is for individuals to couple their sense of social identity to a brand to such an extant that they actually appoint themselves as ready defenders of that respective brand's "honor". Intel and AMD must be happy to know that the unrelenting hivemind of their fanboy armies are out their on a daily basis doing PR work for them for free.

Perhaps some individuals *cough, cough* should take a few minutes to research the intersection between Social Identity Theory and brand loyalty so as to get a grasp upon their own behavior.

 

[Frick]

I think the most astonishing phenomenon here is how common it is for individuals to couple their sense of social identity to a brand to such an extant that they actually appoint themselves as ready defenders of that respective brand's "honor". Intel and AMD must be happy to know that the unrelenting hivemind of their fanboy armies are out their on a daily basis doing PR work for them for free.

Perhaps some individuals *cough, cough* should take a few minutes to research the intersection between Social Identity Theory and brand loyalty so as to get a grasp upon their own behavior.

The way of the new world dude, where "I disagree" is the most awful personal attack imaginable.

 

[Ravenmaster]

I won’t buy another intel processor until they move away from 14nm tech

 

[ppn]

I won’t buy another intel processor until they move away from 14nm tech

Add 10 and 7 to the moving away from list because 5nm is the real thing.

 

[DeathtoGnomes]

The way of the new world dude, where "I disagree" is the most awful personal attack imaginable.

I Disagree! Thats downright flaming!

 

[Master Tom]

I have set my i9-9900K to 165W.

 

[cdawall]

riiiiiiight,temperatures don't matter,but tdp is so important people are arguing over it constantly.

Neither is actually important. CPU stability is important, past that its a flame game of fanboys.

My CPU runs quite warm under load. It also pulls more wattage than just about anyone else's on the page. Actual enthusiasts build around that instead of complaining about it. If I can dissipate 800w of CPU then you can all figure out these little 200w loads in full tower cases.

 

[rtwjunkie]

Can we just stop talking about TDP, nobody cares.

It’s actually important if you are determining how efficient a cooler to buy. If you believe Intel for instance, that your CPU has 95w TDP and the capacity of your cooler you bought is only 150, you are probably screwed. This is because in reality that chip is hitting 200, and a better cooler would have been in order if this info was made public.

 

[Crackong]

It’s actually important if you are determining how efficient a cooler to buy. If you believe Intel for instance, that your CPU has 95w TDP and the capacity of your cooler you bought is only 150, you are probably screwed. This is because in reality that chip is hitting 200, and a better cooler would have been in order if this info was made public.

Please also put thermal density into consideration.

Die size of the new cometlake 10 core is expected to be 200mm².
Existing 10 core 7900x is 322 mm².

The 7900x is a 140W TDP Chip , 140 / 322 = 0.435W / mm²
10 core comet lake is 125W , 125 / 200 = 0.625W / mm²

This 10 core comet lake is 44% more than an 7900x.

Then , thermal conduction is directly proportional to surface area.
So the 10 core comet lake has 38% less surface area than the 7900x
Means the rate of conduction of 10 core comet lake is at least 38% lower than the 7900x

So this thing has 44% more heat density and 38% less thermal conductivity than a 7900x .

Means the "125W TDP" new chip actually requires a BIGGER cooler than 7900x.
Despite the 7900x technically has a higher TDP.

 

[DeathtoGnomes]

o this thing has 44% more heat density and 38% less thermal conductivity than a 7900x .

I followed everything. til you said this... How does the IHS factor?

 

[londiste]

Means the "125W TDP" new chip actually requires a BIGGER cooler than 7900x.
Despite the 7900x technically has a higher TDP.

Why would bigger cooler help with thermal density issues? Contact area does not change and if cooler is capable of cooling down the heat in the first place, the effect from increased capability is minor at best.

 

[Crackong]

I followed everything. til you said this... How does the IHS factor?

I think the IHS affects 2 things : Thickness and surface area.

A thicker IHS , or should say the thickness of anything between the heat source and the cooler hurts the heat transfer, and this is a known problem for 9900k from Der8auer 's results.

The surface area provided by the IHS is also a factor, bigger is better of cause.

Why would bigger cooler help with thermal density issues? Contact area does not change and if cooler is capable of cooling down the heat in the first place, the effect from increased capability is minor at best.

Heat transfer rely on temperature differences.

Let's say the Temperature of the top of the CPU is T1, and the Temperature of the bottom of the Cooler is T2 .

The bigger the difference between T1 and T2 , the better the heat transfer.

A bigger cooler usually faster in dissipating heat, so the bottom of the cooler remains cooler, thus helping the heat transfer.

Of cause there are diminishing returns upon certain level.

It is like an equilibrium.

 

[DeathtoGnomes]

I think the IHS affects 2 things : Thickness and surface area.

A thicker IHS , or should say the thickness of anything between the heat source and the cooler hurts the heat transfer, and this is a known problem for 9900k from Der8auer 's results.

The surface area provided by the IHS is also a factor, bigger is better of cause.

wont that mess up that math you did?

A bigger cooler usually faster in dissipating heat, so the bottom of the cooler remains cooler, thus helping the heat transfer.

the heat transfer rate I think is measured in BTUs, i forget exactly.

 

[notb]

Why would bigger cooler help with thermal density issues? Contact area does not change and if cooler is capable of cooling down the heat in the first place, the effect from increased capability is minor at best.

Because it can transfer heat faster.

The fact that a cooler is rated for 150W doesn't mean it'll be fine for all 150W CPUs. It means it was made for (and tested with) some particular CPUs that were seen as representative.

All currently sold coolers have been designed for 14nm CPUs.
"TDP 150W" means a cooler should be fine for a CPU that (constantly) consumes 150W of power and heats up similarly to pre-7nm CPUs.
For a 7nm CPU you need a much more powerful solution. Check what AMD recommends for 3950X.

 

[Crackong]

wont that mess up that math you did?

Indeed quite a bit.

The size of a 9900k is 1406.25 mm², consider 65% as IHS size ~900 mm²
The size of a 7900x is 2362.5 mm², consider 65% as IHS size ~1500 mm²

So the 10 core CometLake now has

44% more heat density
38% less thermal conductivity (die)
40% less thermal conductivity (IHS)

than a 7900x .

What a furnace.

 

[notb]

It’s actually important if you are determining how efficient a cooler to buy. If you believe Intel for instance, that your CPU has 95w TDP and the capacity of your cooler you bought is only 150, you are probably screwed. This is because in reality that chip is hitting 200, and a better cooler would have been in order if this info was made public.

I don't see how this could happen.
TDP in consumer CPUs is given for base clocks. Everything above impacts how long boosts will be possible. You can pair a 150W cooler with a 9900K. It'll be fine.

Of course you'll need something bigger for sustained boost under load - if that's how you're going to use that CPU.

The problem here is that we test consumer CPUs by running hour-long benchmarks under very strong coolers, so the CPU isn't limited and pulls as much as it can. This isn't realistic.
Also, "thermal throttling" is not something to be afraid of. It's a safety mechanism.

Intel CPUs are very efficient when you don't mess with them (OC, voltage manipulation etc) - on par with Zen2, miles ahead of early Ryzen CPUs.

 

[londiste]

Intel CPUs are very efficient when you don't mess with them (OC, voltage manipulation etc) - on par with Zen2, miles ahead of early Ryzen CPUs.

Current Intel cannot compete with Zen2 when it comes to core power efficiency. 7N claim is something around 30% in power efficiency over 14nm and it shows. Intel can compete at lower core counts, partially due to IO Die and additional base consumption but at around 8 cores or more Zen2 simply gets ahead.

CPUs are by and large power-limited. It does not quite apply to single-core loads - where Zen2 goes and happily consumes 20W or a bit more on a single core - or gaming where many threads can be in use but core utilization is fairly low. It does apply fully to productivity use cases - rendering, encoding etc. Thanks to better power efficiency Zen2 can retain more of its clock speed than Coffee Lake at the same power level.

 

[notb]

Indeed quite a bit.
[cut]

Indeed none of that post makes sense.
9900K die size is 174mm2.
I'm not sure why you're looking at IHS size, let alone multiply it by 0.65.

Current Intel cannot compete with Zen2 when it comes to core power efficiency. 7N claim is something around 30% in power efficiency over 14nm and it shows. Intel can compete at lower core counts, partially due to IO Die and additional base consumption but at around 8 cores or more Zen2 simply gets ahead.

TPU, among others, test efficiency (i.e. amount of energy used to complete a task). Here's the graph. No futher comment needed, hopefully.

The gap is still huge in single-core because of how much IF pulls (but that's a fundament of Zen architecture).

AMD Ryzen 7 3700X Review

AMD's $330 Ryzen 7 3700X is an 8-core, 16-thread CPU that's clocked high enough to compete with Intel's offerings. Actually, its application performance matches even the more expensive Intel Core i9-9900K. Gaming performance has been increased significantly, too, thanks to the improved...

www.techpowerup.com

 

[Crackong]

Indeed none of that post makes sense.
9900K die size is 174mm2.
I'm not sure why you're looking at IHS size, let alone multiply it by 0.65.

Please check #86 for die size -> heat density calculation.


IHS size is also one important aspect in the equation.
Since the path is: Die -> TIM -> IHS -> TIM ->Cooler .
The horizontal heat transfer within the IHS will always be faster than heat transfer through TIM, so the IHS is an important part of the system.

 

[Berfs1]

I think the most astonishing phenomenon here is how common it is for individuals to couple their sense of social identity to a brand to such an extant that they actually appoint themselves as ready defenders of that respective brand's "honor". Intel and AMD must be happy to know that the unrelenting hivemind of their fanboy armies are out their on a daily basis doing PR work for them for free.

Perhaps some individuals *cough, cough* should take a few minutes to research the intersection between Social Identity Theory and brand loyalty so as to get a grasp upon their own behavior.

I do like the progress and performance AMD offers, however I apologize if I came off sounding like an AMD fanboy, I am not. I do not give either company an unfair advantage just by “liking them more”, I however factor in reliability and truthfulness, and AMD has been more honest than intel. Intel never mentions power draw in their performance charts, not even in the footnotes. At least AMD does, through total system power usage. AMD is just the more sensible option for most budget gamers, because you can settle for a much cheaper Ryzen 5 3600, and use the extra money saved for a better GPU. CPUs are getting outdated every year, so don’t even bother, take the maybe 5% cpu hit and get a stronger GPU. Oh and not to mention everything else AMD offers, such as tuning with cheap motherboards; you need to spend 150$+ for a quality intel motherboard that has good features, but with AMD, just get the B450M Steel Legend and you are good to go. No, I am not denying Intel has more maximum performance. Yes, I am saying Ryzen is better because it makes no sense to pay 20,000$ more for an 15 horsepower upgrade for a car for most drivers. Long story short, if you are a COMPETITIVE gamer, as in someone doing esports or something, then yea Intel is the better choice by far. But if you are like 98% of the gaming population that doesn’t have 5000$ every month to upgrade a computer, AMD is the more sensible option in almost all other ways, power efficiency too.