[windfire]

Quote:

The new cooler is an engineering sample, it boasts six 6mm heatpipes, measures 230mm*146mm*47mm, weights 590g which is 60g heavier than MK-26. There are 20 pieces and 44 pieces of cooling fins in each side covering cooling area of 4670cm2.

Source: http://en.expreview.com/2012/05/08/hands-on-prolimatech-mk-26-pro-vga-cooler-born-for-hd-7970/22975.html

 

[Arni90]

I felt a shiver down my spine when I saw this.
Looks ridiculously big and powerful.

 

[Shiveron]

Meh. I happen to like how my card acts like a secondary space heater. It's like a bonus feature!

 

[Alex132]

Quote:

Originally Posted by Shiveron

Meh. I happen to like how my card acts like a secondary space heater. It's like a bonus feature!

Same amount of heat will be dissipated, this just does it better.

 

[Xyxyll]

Quote:

Originally Posted by Alex132

Same amount of heat will be dissipated, this just does it better.

Heat Transfer 101

 

[funfortehfun]

If you put this on a low-TDP CPU, it would probably act as a pretty good passive 1.5U CPU cooler.

 

[jrbroad77]

Quote:

Originally Posted by Alex132

Same amount of heat will be dissipated, this just does it better.

Not quite. If the core and PCB temperature are hotter with a worse cooler, less heat is dissipated. If one GPU is running hotter, everything being the same except for the cooler and/or fanspeed, simple logic says the hotter one dissipated less heat. If heat is "created", internal delta(heat) + heat dissipated = total heat..

 

[Xyxyll]

Quote:

Originally Posted by jrbroad77

Not quite. If the core and PCB temperature are hotter with a worse cooler, less heat is dissipated. If one GPU is running hotter, everything being the same except for the cooler and/or fanspeed, simple logic says the hotter one dissipated less heat. If heat is "created", internal delta(heat) + heat dissipated = total heat..

Heat has to go somewhere, and since the temperature of the card does not indefinitely rise, heat generated (W) = heat dissipated (W), whether it's through a cooler or through natural convection of the PCB. Better performing coolers simply are more efficient at transferring heat, which results in less of a temperature climb of the card. You're still pumping 200W into the air, whether the card is near room temperature or 100C.

 

[jrbroad77]

Quote:

Originally Posted by Xyxyll

Heat has to go somewhere, and since the temperature of the card does not indefinitely rise, heat generated (W) = heat dissipated (W), whether it's through a cooler or through natural convection of the PCB. Better performing coolers simply are more efficient at transferring heat, which results in less of a temperature climb of the card. You're still pumping 200W into the air, whether the card is near room temperature or 100C.

So your theory is heat generated = heat dissipated? This is false on an instantaneous basis, but start to finish (ie power up system, game, then shut it down), that's the end result. If that's the case instantaneously, your GPU would run at ambient temperatures.

Again, you have 2 GPUs, different coolers, same clocks, same voltages. Assume switching the coolers would verify their heat output etc. Suppose the one with better cooling runs at 50C, the one with worse cooling at 90C. Explain your theory in regards to this.

 

[Xyxyll]

Quote:

Originally Posted by jrbroad77

So your theory is heat generated = heat dissipated? This is false on an instantaneous basis, but start to finish (ie power up system, game, then shut it down), that's the end result. If that's the case instantaneously, your GPU would run at ambient temperatures.
Again, you have 2 GPUs, different coolers, same clocks, same voltages. Assume switching the coolers would verify their heat output etc. Suppose the one with better cooling runs at 50C, the one with worse cooling at 90C. Explain your theory in regards to this.

It's no theory. It's the first law of thermodynamics. "The increase in the amount of energy stored in a control volume must equal the amount of energy that enters the control volume, minus the amount of energy that leaves the control volume."

This sounds remarkably similar to the equation you worded out, and that's right because it's the same! What you're missing is how the equation changes when you reach steady state.

So what happens when steady state core/pcb temperatures are reached (i.e. max temperatures)? Well, the amount of energy stored in the core and pcb (i.e. it's temperature above ambient), no longer increases. It is constant, so all future energy inputed to the system is dispersed.

Now let's try and understand how the coolers can still output the same heat, despite different temperatures.

Here's a general heat rate equation for convection: q"=hΔT, where q" is the convective heat flux, h is the convection heat transfer coefficient (determined by the cooler's materials, surface area, fan speed, shape, etc), and ΔT is the temperature difference between the card and the surrounding air.

Once steady state is reached (your 50C and 90C card example), both constant-temperature cards transfer exactly the same energy to the air (heat in = heat out). The better cooler transfers the heat with a lower temperature delta because it's h value is higher, and the poorer cooler transfers the heat with a higher temperature delta because it's h value is lower.

 

[jrbroad77]

Quote:

Originally Posted by Xyxyll

It's no theory. It's the first law of thermodynamics. "The increase in the amount of energy stored in a control volume must equal the amount of energy that enters the control volume, minus the amount of energy that leaves the control volume."
This sounds remarkably similar to the equation you worded out, and that's right because it's the same! What you're missing is how the equation changes when you reach steady state.
So what happens when steady state core/pcb temperatures are reached (i.e. max temperatures)? Well, the amount of energy stored in the core and pcb (i.e. it's temperature above ambient), no longer increases. It is constant, so all future energy inputed to the system is dispersed.
Now let's try and understand how the coolers can still output the same heat, despite different temperatures.
Here's a general heat rate equation for convection: q"=hΔT, where q" is the convective heat flux, h is the convection heat transfer coefficient (determined by the cooler's materials, surface area, fan speed, shape, etc), and ΔT is the temperature difference between the card and the surrounding air.
Once steady state is reached (your 50C and 90C card example), both constant-temperature cards transfer exactly the same energy to the air (heat in = heat out). The better cooler transfers the heat with a lower temperature delta because it's h value is higher, and the poorer cooler transfers the heat with a higher temperature delta because it's h value is lower.

But you're missing the point. If you plot out the heat dissipation for both after firing up the computer, running say Furmark for 5 minutes on each, the total heat dissipation from the first 5 minutes will be lower on the card with the hotter GPU core; assuming the heat stored in the heatsink and PCB are at least equal to the cooler card. In steady state, they both dissipate the same amount of heat, yes. All systems don't necessarily reach steady state . So, it's not accurate to say that the heat dissipated = heat generated on an instantaneous basis, unless the system is in steady state.

Note that 1. I agree, heat dissipated = heat generated, start to finish (and in a steady-state system), but 2. Heat dissipated doesn't = heat generated on an instantaneous basis until you get to a steady state - as I've already laid out. So, steady-state, yes heat in = heat out on instantaneous basis. If it's not steady state, or wasn't steady state since the start of the process, you can't say total heat in = total heat out, until the process completes (by process completing, I mean heat generated stops, the system is given time to return to ambient).

 

[prava]

Quote:

Originally Posted by jrbroad77

Not quite. If the core and PCB temperature are hotter with a worse cooler, less heat is dissipated. If one GPU is running hotter, everything being the same except for the cooler and/or fanspeed, simple logic says the hotter one dissipated less heat. If heat is "created", internal delta(heat) + heat dissipated = total heat..

Quote:

Originally Posted by jrbroad77

So your theory is heat generated = heat dissipated? This is false on an instantaneous basis, but start to finish (ie power up system, game, then shut it down), that's the end result. If that's the case instantaneously, your GPU would run at ambient temperatures.
Again, you have 2 GPUs, different coolers, same clocks, same voltages. Assume switching the coolers would verify their heat output etc. Suppose the one with better cooling runs at 50C, the one with worse cooling at 90C. Explain your theory in regards to this.

No, no. Temperature is only the quantity of heat per mass unit, and nothing else. Once the equilibrium is reached (eq. means that temperature stays constant) it means that all incoming heat will be dissipated to the ambient, and thus why no matter what cooler you use, the heat output will be exactly the same. In this regard, once you turn off your computer, its actually the heatsink you are using the one that makes your chip stay at the same temperature for a much larger time. Why? A computer chip is very small, and thus even just a few calories will make its temperature to skyrocket (and thus why you should never ever ever turn on your computer without a heatsink) but, at the same time, given it has very small mass once you stop the computer it will cool down almost inmediately. But, the heatsink is very big and thus it will carry a ton of heat even if you feel its not very hot (like I said, "hot" is a perception we have. A very hot object, if very small, will carry very little heat compared to a very big object that is not so hot).

Imagine you have 1000W heater in your room. Imagine it has a dimension of X. Now, imagine its triple the size. Which one will be hotter to the touch? Its obvious that the smaller one will be hotter because it has less surface area to dissipate and thus it will carry much more heat per mass unit...but this has nothing to do with the total heat it is outputting, which is what we are talking about here. The card itself is the resistor that makes it run hot, and the body of the heater is the heatsink (and its exactly like that): the bigger the heatsink the more efficient it gets (efficient means that it needs a lower temperature difference between ambient and itself, what we call a Delta), but the heat output remains exactly the same...

Did I say the same? Well, its not that easy

On the contrary, on chips its not weird to see total heatput gets REDUCED the better you cool them. Why? Circuits are more efficient when they run cool, which means that at a certain temperature they will require less power to run at the same clocks and thus why, funnily enough, if you improve its cooling capacity you can even improve its efficiency (as hot transistors get leakier, and the leakier, the more current they need, the leakier they get and so on).

PS: but still, its not useful to analyse systems that aren't in an equilibrium state. We know that if 100W go into a computer, they will go out as heat. Of course they won't as soon as such heat is produced, but we know they will, and thus why for heating purposes it doesn't matter how good the cooling in your pc is, the heat is the same, in the end.

 

[Blameless]

Actually, slightly more heat is dissipated as hotter parts are less efficient and draw slightly more power.

But yeah, you cannot keep a room cooler by cooling parts less efficiently, at least once equilibrium is reached, as mentioned.

Edit: Anyone else notice that the base of this heatsink has a plate soldered to it to contact the core, instead of having a milled protrusion? Probably works fine, but still seems like a half-assed shortcut to me.

 

[JedixJarf]

Quote:

Originally Posted by Arni90

I felt a shiver down my spine when I saw this.
Looks ridiculously big and powerful.

Thats pretty much what she said.

 

[Alex132]

Basically it is the same heat output over a normal period of time. It will not be significantly cooler or hotter. One thing is for sure, it will make the internals of your case hotter if you project the fans upwards.

 

[homestyle]

AMD why oh why did you make the gpu have that stupid recess.

This will go down in history as one of the most boneheaded moves ever.

 

[kikkO]

Why spend so much money on a aftermarket cooler when you do this? Get a MSI Lightning then get the UN Rad Brackets and add any fan you want.

 

[B!0HaZard]

Quote:

Originally Posted by kikkO

Why spend so much money on a aftermarket cooler when you do this? Get a MSI Lightning then get the UN Rad Brackets and add any fan you want.

Because that solution cools worse and looks worse (arguably). Some of the bigger aftermarket heatsinks are sexy.

 

[kikkO]

Quote:

Originally Posted by B!0HaZard

Because that solution cools worse and looks worse (arguably). Some of the bigger aftermarket heatsinks are sexy.

And you know this how?

It reduced my Lightning temps a further 6C @ 100% load while pushing hot air out the rear.

You can find this concept in the NZXT Switch. I don't know what you're getting at.



Here's another, Lian Li V2120:

 

[B!0HaZard]

Quote:

Originally Posted by kikkO

And you know this how?
It reduced my Lightning temps a further 6C @ 100% load while pushing hot air out the rear.
You can find this concept in the NZXT Switch. I don't know what you're getting at.
Here's another, Lian Li V2120:

Great, people use the concept, so what? It works for reducing temps, but not as well as a bigger cooler. It is still more effective to use a big aftermarket cooler (like an Arctic Cooling monster) and if you want you can do both things. It is very simple, the huge area of a big sink simply allows more heat to be transferred to the air. You can either use this to slow down fans (most users prefer not having a 3500+ RPM fan running) or you can match the noise and get better cooling. As good as the Twin Frozr (alledgedly) is, it still operates at a high fan speed (1200-4500 RPM on most cards). That is the same speed as the default fan (although it's still less noisy because of the different fan designs and the lower speeds needed to cool). The comparisons you see most reviewers do, involves two low-med speed 120 mm fans against a 3000 RPM stock cooler, so they're not really comparable. An example is the MK-26 test linked in this article. The fans used on the MK-26 are only 1200 RPM fans while the stock cooler has auto mode (can scale to 3000+ RPM).