Why Intel CPU's run at 95°C and why AMD's should, also

[mtcn77]

@mtcn77 what is your current CPU?

Not about the point I'm making and not about yours too, if you are slow to make the 7nm jump.

ok...

Here is a custom copper ihs - https://www.gamersnexus.net/guides/3238-custom-copper-ihs-tested-on-intel-i7-8700k-cpu-rockit-cool

It must upset you that you are only referring to my strong suit. I am not sporting an eagerness for better conductors. I'm fancying best practices, so I'm not so easy to be dismissed... I'm crafty.
[Why it seems like you've misses my point on #79]

 

[EarthDog]

Not about the point I'm making and not about yours too, if you are slow to make the 7nm jump.


It must upset you that you are only referring to my strong suit. I am not sporting an eagerness for better conductors. I'm fancying best practices, so I'm not to easily dismissed... I'm crafty.
[Why it seems like you've misses my point on #79]

Crafty... lulz. Too much theory, not enough applied physics is all I see here. Why dont you stop posting and make it happen with your confidence and craftiness? Otherwise, we're just spinning our tires trying to understand w/e TF it is you are trying to really say.

Also, for jack again... silver ihs does little here - http://forum.notebookreview.com/threads/silver-7700k-ihs-prototype-testing.809218/#post-10606102

For the couple/few C thus may make a difference, simply doesnt yield anything real-world. People have tested a copper ihs, silver ihs... and while there are improvements, it isnt worth it for an overwhelming majority of the enthusiast crowd.

 

[mtcn77]

Crafty... lulz. Too much theory, not enough applied physics is all I see here. Why dont you stop posting and make it happen. Otherwise, we're just spinning our tires.

Also, for jack again... silver ihs does little here - http://forum.notebookreview.com/threads/silver-7700k-ihs-prototype-testing.809218/#post-10606102

Silver.

Silver is only 425w/m*k. Vapor chambers are 10000-50000. I'm sure you get the picture.
This isn't to denigrate intel followers, but you are really misinterpreting mTr/mm2 density of these parts.

 

[EarthDog]

What is the difference between a vapor chamber here versus those that have already been in use on gpus???

Silver is only 425w/m*k. Vapor chambers are 10000-50000. I'm sure you get the picture

Build it or can it is where I'm at...

 

[Zach_01]

Not about the point I'm making and not about yours too, if you are slow to make the 7nm jump.

Lets put all this to a good use... enough about the theory...
I would like to see your 7nm CPU working at 95°C and 75°C. To compare values of clock, voltage, wattage and current of the 2 temps I mention, and see what short of benefits will both give...
Up to it?

 

[JackCarver]

Again, enjoy that 100 mhz difference if you're lucky! This isnt for the average enthusiast, even...

I wouldn‘t buy this but to come back to the topic here:
Liquid metal on the die along with an silver IHS would be one of the best solutions for a good junction to case heat dissipation. The case to ambient belongs to the user and his or her heatsink

 

[mtcn77]

Build it or STFU is where I'm at...

I can; however are you sure we on the same page? First post, I make no text walls to keep the paragraphs flowing. One just has to read;

Spoiler: Totally new reddit:

Intel has frequency problems with their 10nm because of the 2.7x density scaling factor over 14nm. Zen 2's density is below 60 MTr/mm2 and more akin to Intel's 14nm than regular N7. Contrast that with Ice Lake which is more than 50% denser than N7 HPC, and it's pretty impressive that it can hit 4.1 GHz single core turbo. That would translate to around 4.3-4.4 GHz in AMD advertised clock speeds.

[I think this is the same reddit thread]
You notice same old same old if the push comes to shove with N7-10nm.

 

[EarthDog]

I wouldn‘t buy this but to come back to the topic here:
Liquid metal on the die along with an silver IHS would be one of the best solutions for a good junction to case heat dissipation. The case to ambient belongs to the user and his or her heatsink

I can; however you sure we on the same page? First post, I make no text walls to keep the paragraphs flowing. One just has to read;
[I think this is the same reddit thread]
You notice same old same old if the push comes to shove with N7-10nm.

Good luck, gents. I'm just too slow for this thread I guess..

 

[TheoneandonlyMrK]

I can; however are you sure we on the same page? First post, I make no text walls to keep the paragraphs flowing. One just has to read;
[I think this is the same reddit thread]
You notice same old same old if the push comes to shove with N7-10nm.

There are many ways to cook an egg, it's how it sits in a belly that matters.

I'm just simplifying the debate here on to save time and brain.

 

[mtcn77]

Lets put all this to a good use... enough about the theory...
I would like to see your 7nm CPU working at 95°C and 75°C. To compare values of clock, voltage, wattage and current of the 2 temps I mention, and see what short of benefits will both give...
Up to it?

Well, I have grasped your perspective. It is my contention that you haven't grasped whether I highlight latent heat, just not sensible heat. So you aren't taking me at my word when narrowing it down to a temperature differential. We are on different pages.

There are many ways to cook an egg, it's how it sits in a belly that matters.

I'm just simplifying the debate here on to save time and brain.

Totally, which is why you eat it poached.

Okay, I have to make it spin using the Fourier Law of Conduction:
q=k*dT, q=Q/A.
7nm is twice the q, because area(A) is twice smaller. Hence you need to make dT twice higher, or use a higher heat absorbing(-2q) coldplate. It essentially will reach thermal equalibrium, we are negotiating our terms favourably.

 

[TheoneandonlyMrK]

Well, I have grasped your perspective. It is my contention that you haven't grasped whether I highlight latent heat, just not sensible heat. So you aren't taking me at my word when narrowing it down to a temperature differential. We are on different pages.


Totally, which is why you eat it poached.

Okay, I have to make it spin using the Fourier Law of Conduction:
q=k*dT, q=Q/A.
7nm is twice the q, because area(A) is twice smaller. Hence you need to make dT twice higher, or use a higher heat absorbing(-2q) coldplate. It essentially will reach thermal equalibrium, we are negotiating our terms favourably.

Try it ,ie crack on , but do report back, see if YMMV because as I told you been here done that, didn't pan out, and could be pointless.
Especially if you're just gaming two hours a night.
But please do crack on.

 

[eidairaman1]

No lol, almost all of us in this thread are hardware enthusiasts and very much into overclocking. Because of that we all understand the benefits of running cooler, which is why we are all confused as to why you think higher temps are better..

My rig is an example, 55 under gaming, 75 under blender, Air Cooled on a thinner cooler.

 

[mtcn77]

Try it ,ie crack on , but do report back, see if YMMV because as I told you been here done that, didn't pan out, and could be pointless.
Especially if you're just gaming two hours a night.
But please do crack on.

You wouldn't suspect, although I know how horrid a better gpu cooler that surpasses ambient flow restrictions can bring temperatures sky high, but don't quote me on that. I'm wise to the act now. You just don't have to beat a square peg in a round hole - you can produce the same heat load and just expel it more slowly. The temperature will level at a higher threshold. Same heat, lower volume of air, higher exhaust temperature, same case ambients. It is only an orchestration of various parameters using the same airflow.
Notice the tdp limit? We aren't circumventing it any time soon. We are back in square one forever more.

I cannot underestimate CoolerMaster's investment into the development of vapor chamber technology. They are the trend setters in using the latent heat transfer function of working fluids.

 

[JackCarver]

This graph is from 2014....is it yet a technique to think about? Would be interesting how the 2020 graph AiO vs Vapour Chamber will be looking. There is also another technique, where the whole chip is set under liquid, when I understand right, it‘s called phase shift cooler.

 

[mtcn77]

This graph is from 2014....is it yet a technique to think about? Would be interesting how the 2020 graph AiO vs Vapour Chamber will be looking. There is also another technique, where the whole chip is set under liquid, when I understand right, it‘s called phase shift cooler.

I haven't found any good examples. It is a good irony, they are only now on the verge of significance yet nowhere to be found.

 

[JackCarver]

Will definitely be interesting what cooling technique will be the best in the near future or will replace the classic techniques like air or water cooling. But for me, most brands out there invent new AiOs, nearly every day you can read about it. Also CoolerMaster has new AiOs in program, one with two pumps as I read. So AiOs seem to be still a big point out there in cooling, especially cpu cooling. Personally I‘m a water cooling fan, mostly because of the look of a custom loop with hard tubes and because you get really good temps, especially for the GPU. In my opinion water cooling a GPU makes much more sense than water cooling a cpu, only look at their power dissipation of about 250-400W, if you use a 2080 or 2080Ti, in benchmarks or games.

 

[EarthDog]

my opinion water cooling a GPU makes much more sense than water cooling a cpu, only look at their power dissipation of about 250-400W, if you use a 2080 or 2080Ti, in benchmarks or games.

I agree with your point, but for a more detailed reason.

As a side note, those GPUs are 215W and 250W cards. Power limits are typically set around 5-10% with outliers on the edge there. 400W out of these cards wont happen without at least a modded bios for significantly increased power and voltage limits.

As noted somewhere previously, the GPU temps are typically a lot lower with water. To the point where they can run before the card stops dropping boost bins due to temperatures (55C??). The difference between a few C isnt much, but when you able to run full boost due to 25C+ differences... a performance bump is there.

 

[Zach_01]

Well, I have grasped your perspective. It is my contention that you haven't grasped whether I highlight latent heat, just not sensible heat. So you aren't taking me at my word when narrowing it down to a temperature differential. We are on different pages.

Are you trying to deliberatly confuse us?
What latent heat has to do with running a CPU at 95°C instead of 75 or 60.
Yes vapor chambers, heatpipes, chillers, LN2... all working on the latent heat principal. They absorb large amounts of heat by changing a liquid to vapor. Water cooling is working on sensible heat. SO?
Can you tell us what benefit will a CPU see in its actual function and operation by working on 95°C instead of 75°C. You cant! Because there isnt any...

Dont you understand what are you telling us here?
Its like heating the interior of your refrigerator beyond its nominal levels "just" to maximize the heat absorption and dissipation of its heat pump system. What actual benefit does this have, other than sour food?

 

[trickson]

Intel Because someone has to cook the eggs!

 

[eidairaman1]

Are you trying to deliberatly confuse us?
What latent heat has to do with running a CPU at 95°C instead of 75 or 60.
Yes vapor chambers, heatpipes, chillers, LN2... all working on the latent heat principal. They absorb large amounts of heat by changing a liquid to vapor. Water cooling is working on sensible heat. SO?
Can you tell us what benefit will a CPU see in its actual function and operation working on 95°C instead of 75°C. You cant! Because there isnt any...

Dont you understand what are you telling us here?
Its like heating the interior of your refrigerator beyond its nominal levels just to maximize the heat dissipation of its heat pump system. What benefit does this have other than sour food?

The A6 3650 did not like running at 78C it would auto shutdown, low and behold max is 72C

 

[Zach_01]

The A6 3650 did not like running at 78C it would auto shutdown, low and behold max is 72C

Yes I understand this and I'm with you... Other(s) suggest to run CPUs as hot as possible for the best heat dissipation(by amount, and latent or sensible) but eventually cant tell us simple things like how is this actually can improve the performance of it.
Sorry @eidairaman1, but are you following this discussion? (God make it one, because its not...)

 

[JackCarver]

As a side note, those GPUs are 215W and 250W cards. Power limits are typically set around 5-10% with outliers on the edge there. 400W out of these cards wont happen without at least a modded bios for significantly increased power and voltage limits.

Yes I meant the power hungrier cards out there, like the 2080Ti Lightning. But also my card (ASUS 2080 ROG STRIX) consumes up to 280W in Firestrike. The cards with default power limit are less hungry.

The difference between a few C isnt much, but when you able to run full boost due to 25C+ differences... a performance bump is there.

Absolute agree here

 

[eidairaman1]

Yes I understand this and I'm with you... Other(s) suggest to run CPUs as hot as possible for the best heat dissipation(by amount, and latent or sensible) but eventually cant tell us simple things like how is this actually can improve the performance of it.
Sorry @eidairaman1, but are you following this discussion? (God make it one, because its not...)

Yes I am and I find running any electronic at max temperature is ridiculous and shortens its life and causes more rma and shortened mtbf. Look at gpus used for mining...

Avionics bays in aircraft use forced cooling even

 

[Zach_01]

Yes I am and I find running any electronic at max temperature is ridiculous and shortens its life and causes more rma and shortened mtbf. Look at gpus used for mining...

Avionics bays in aircraft use forced cooling even

Yeah, tell me about it...
I work in a Oil Refinery as a ControlRoom and field operator (Cycling on seven different Duties, 2 CR and 5 Field). The Complex functions under a DCS (Distributed Control System) with a couple of dozens of PC like computers as Input/Output that implies hundreads of microcontrollers placed in racks, server like, in tall cabins, in a large room with an industrial size air conditioner keeping room temp under 18C at all cost.

 

[mtcn77]

What latent heat has to do with running a CPU at 95°C instead of 75 or 60.

It is to demonstrate how your cooler can operate at a more efficient profile. You don't run any cpu, just ryzen 3000 cpus as they are the only group included in this topic to demonstrate heatsink transmittance issues.
Temperature as you know, has thermal equilibrium in its basis and every heat transfer works according to the temperature vector. Ideal conductors would run at equal heat density, a.k.a temperature - with the ihs on contact. That does not mean run it ideally at maximum heat.
Just run it with a supposedly better contact interface and see if it has the required provisions whenever you might need it. It is better to have a constant boost cpu than throttly throttly, no?

Yeah, tell me about it...
I work in a Oil Refinery as a ControlRoom and field operator (Cycling on seven different Duties, 2 CR and 5 Field). The Complex functions under a DCS (Distributed Control System) with a couple of dozens of PC like computers as Input/Output that implies hundreads of microcontrollers placed in racks, server like, in tall cabins, in a large room with an industrial size air conditioner keeping room temp under 18C at all cost.

This is not a live demonstration. You do it under supervision in controlled settings. This is to see cooler thermal sufficiency and running the cpu, that would be ryzen 3000 exclusively, being served by the cpu cooler as required by heat pressure to have a linear heat dissipation work efficiency. It is strictly mission critical averse. Only for exploring cooler capability headroom.