SweClockers: Geforce GTX 590 burns @ 772MHz & 1.025V

[kevinsbane]

Sorry, but NO.
Temperature is average kinetic energy of the molecules. E=kT
Heat or Thermal Energy is the total kinetic energy of molecule in an object. At temperatures near absolute zero the excitation energies of crystal lattice is an important part of Heat, but at room temperatures and above heat is best approximated by total kinetic energy.

I hope you mean thermal energy is not heat. I realise this is wikipedia, but this sentence is still accurate:

Thermal energy is distinct from heat.

I am not well versed in the physics of thermal energy, but I know heat is not thermal energy.

 

[GaiaHunter]

To be more precise heat is the process of transfer of energy between systems.

Thermal energy is the kinetic energy of a system particles - stuff like vibration and rotational states of molecules, etc.

 

[pcm81]

I hope you mean thermal energy is not heat. I realise this is wikipedia, but this sentence is still accurate:

Thermal energy is distinct from heat.

I am not well versed in the physics of thermal energy, but I know heat is not thermal energy.

To be 100% accurate you are right, Heat is used as a term for heat transfer, that is energy transfer across a boundary. However, units of heat and those of thermal energy are one and the same. So, Heat and thermal energy can be used interchangeably, making a distinction that in that case heat is not an abbreviation for heat transfer. Personnaly I dont like term heat, because it is redundant to energy when that energy is transfered somewhere. In real life all energy is always transferred somewhere, with exception of theoretical physics problems which can never exist in real life, hence all thermal energy in real life is heat. Furthermore, the terms Heat flux or energy flux, as well as time rate of heat or time rate of energy or Power are synonymous in real life.

 

[AnandThenMan]

I think I accidentally logged onto forums.thermodynamics.com.

 

[AtenRa]

Secondly, it should be noted that in the two pictures being compared in this thread by Seero, that the HD 6990 is running on "Standard" mode; from all the reviews I've seen, the HD 6990 consumes less power than the GTX 590. I've seen anywhere from 15-100W less in terms of total system pwoer; unfortunately, xbit labs hasn't come out with their power consumption numbers for the video cards themselves (yet). It is my suspicion that the HD 6990 in O/C mode runs about the same power consumption as the GTX 590 at stock. I can't say for sure, as reviews vary with sites like Anandtech pegging the 6990 OC at higher system consumption than the 590, while other sites like Tom's, Tweaktown, TechReport, etc. as showing 6990OC using less or the same amount of power as the 590. I feel 6990OC wattage~=590 wattage is a reasonable position to take. Thus, in the picture being compared, the HD 6990 is dissipating rather less thermal energy that the gtx. The correct comparison is the 6990 OC vs 590. I will be comparing these two pictures.

If what you saying is correct and HD6990@450W has the same power consumption with the GTX590 then the cores of the GTX590 would must have higher temperatures due to smaller heatsing and lower CFMs. if the temperature readings are true in the review, the actual core temps of the GTX590 are at 80C and 84C way less than HD6990@450W (93W and 102W). How do you explain that ??

HD6990

GTX590

http://www.hardware.fr/articles/747-22/maj-dossier-cartes-graphiques-degagement-thermique.html

http://www.hardware.fr/articles/825-4/dossier-nvidia-repond-amd-avec-geforce-gtx-590.html

Also from the same review the GTX590 power consumption in way lower than HD6990@450W except at Furmark

Granted we don’t know in what bench they did took the readings and the IR pictures but I will bet that in gaming the GTX590 consumes less power than the HD6990@450W.

http://www.hardware.fr/articles/825-3/dossier-nvidia-repond-amd-avec-geforce-gtx-590.html

 

[AnandThenMan]

Actual temperatures don't mean a damn thing, providing the components can handle them. You can create temps that are hotter than the surface of the Sun, but actual energy output is very low because it is extremely localized, and only exists for a few nanoseconds.

 

[pcm81]

Actual temperatures don't mean a damn thing, providing the components can handle them. You can create temps that are hotter than the surface of the Sun, but actual energy output is very low because it is extremely localized, and only exists for a few nanoseconds.

OMG you are so right!!! It all don't matter... 6990 vs 590... IT DOES NOT MATTER. The world is coming to the end in 2012 anyways, I suggest we all go to our local bars and get drunk.

 

[GaiaHunter]

To be 100% accurate you are right, Heat is used as a term for heat transfer, that is energy transfer across a boundary. However, units of heat and those of thermal energy are one and the same. So, Heat and thermal energy can be used interchangeably, making a distinction that in that case heat is not an abbreviation for heat transfer. Personnaly I dont like term heat, because it is redundant to energy when that energy is transfered somewhere. In real life all energy is always transferred somewhere, with exception of theoretical physics problems which can never exist in real life, hence all thermal energy in real life is heat. Furthermore, the terms Heat flux or energy flux, as well as time rate of heat or time rate of energy or Power are synonymous in real life.

But heat transferred doesn't always translate in thermal energy like phase changes.

And you also have other types of energy that can be transferred as heat like chemical energy.

Although is many cases it can be used interchangeably.

 

[thilanliyan]

If what you saying is correct and HD6990@450W has the same power consumption with the GTX590 then the cores of the GTX590 would must have higher temperatures due to smaller heatsing and lower CFMs. if the temperature readings are true in the review, the actual core temps of the GTX590 are at 80C and 84C way less than HD6990@450W (93W and 102W). How do you explain that ??

You are taking the temperatures measured at the back of the PCB (!!...not even the core itself) and taking that as the core temperature?! That in itself in incorrect. The back of the PCB =/= core temperature. You are making some big assumptions in your analysis.

And something else to add, apparently the coolers work in different ways:
http://www.xbitlabs.com/articles/video/display/geforce-gtx-590_3.html#sect0

"The GPUs of the Radeon HD 6990 are cooled one by one: the air flows from the rear heatsink to the front one and is then exhausted. So, we can suppose that the GPUs are going to differ in temperature but most of the hot air will be exhausted out of the system case. The heatsinks of the GeForce GTX 590 are cooled equally because the fan drives the air towards both heatsinks, leaving half of the hot air inside the system case."

I'm not 100% sure if the fan differences they mention are correct (but the large temperature differences in the 6990 GPUs point to that), but that can also explain the temperature differences in the case.

 

[GaiaHunter]

If what you saying is correct and HD6990@450W has the same power consumption with the GTX590 then the cores of the GTX590 would must have higher temperatures due to smaller heatsing and lower CFMs. if the temperature readings are true in the review, the actual core temps of the GTX590 are at 80C and 84C way less than HD6990@450W (93W and 102W). How do you explain that ??

That depends on where exactly the readings are taken.

And remember - the GPUs aren't the only place dissipating energy.

 

[Arkadrel]

OMG you are so right!!! It all don't matter... 6990 vs 590... IT DOES NOT MATTER. The world is coming to the end in 2012 anyways, I suggest we all go to our local bars and get drunk.

I ll remind of you of this quote sometime into 2013

 

[Seero]

OMG you are so right!!! It all don't matter... 6990 vs 590... IT DOES NOT MATTER. The world is coming to the end in 2012 anyways, I suggest we all go to our local bars and get drunk.

I second that.

 

[pcm81]

But heat transferred doesn't always translate in thermal energy like phase changes.

And you also have other types of energy that can be transferred as heat like chemical energy.

Although is many cases it can be used interchangeably.

2. Chemical energy does not exist. To be most general, only 2 energies exist, potential energy and kinetic energy. All other types of commonly refereed to energies are either potential, kinetic or both. Chemical, or nuclear energy are potential energies, because they are stored as binding energies of atoms or nucleons.

1.Phase change is a form of potential energy, while the the energy which comes from motion of molecules is kinetic energy. Excitation of crystal lattice, a component of thermal energy is also potential energy.

So as you see ALL forms of energy are either potential or kinetic and as such all definitions of forms of energy are derived terms and therefore are context dependent and often overlap. For example lets take a hot plate and blow cool air on 1 side of the plate. Does that plate store thermal energy or heat? I could make an argument that the thermal energy stored in the plate is actually heat, because the surface of the plate on which the air is blowing is cooler than the center of the plate and therefore there is energy transfer over an invisible boundary separating center of the plate from is surfaces.

 

[BFG10K]

The difference in thermal colors could simply be the difference in which the two coolers direct hot air. Maybe the GTX590 leaves more hot air in the case.

 

[Arkadrel]

The difference in thermal colors could simply be the difference in which the two coolers direct hot air. Maybe the GTX590 blows more hot air into the front of the case.

No no.. that sounds way to logical...

To me it makes sense that the 6990 blows most of its air outsides the pc case, and the 590, blows most of its air back into the case, which results in the front of the case with the 590 being hotter.

 

[GaiaHunter]

So as you see ALL forms of energy are either potential or kinetic and as such all definitions of forms of energy are derived terms and therefore are context dependent and often overlap. For example lets take a hot plate and blow cool air on 1 side of the plate. Does that plate store thermal energy or heat? I could make an argument that the thermal energy stored in the plate is actually heat, because the surface of the plate on which the air is blowing is cooler than the center of the plate and therefore there is energy transfer over an invisible boundary separating center of the plate from is surfaces.

I have nothing against you calling heat to it.

Quite the opposite.

I always call heat to the kinetic component. Have no use for thermal energy.

Heat as energy measured in Joules and heat as rate of energy transfer in Watts (J/s).

I picked on it because while thermal energy can be described as heat, heat isn't simply described as thermal energy.

 

[Zanovar]

I think I accidentally logged onto forums.thermodynamics.com.

OH cmon you like it really...

 

[Qianglong]

Wow..so many passionate debate on this issue! No wonder is this is a kick ass community! I am guessing the rate of cards blowing up has dropped dramatically?

 

[thilanliyan]

The difference in thermal colors could simply be the difference in which the two coolers direct hot air. Maybe the GTX590 leaves more hot air in the case.

That's what most of us have been trying to get across, but it doesn't seem to be logical to some people.

Xbitlabs said the same thing:
http://forums.anandtech.com/showpost.php?p=31495013&postcount=559

 

[pcm81]

Wow..so many passionate debate on this issue! No wonder is this is a kick ass community! I am guessing the rate of cards blowing up has dropped dramatically?

And eventually it will be zero, after all bad cards blow up.

Failure rate = N*F

where:
N is the number of cards in existence and
F is the probability of failure normalized per unit time of a randomly selected card.

 

[Silverforce11]

That's what most of us have been trying to get across, but it doesn't seem to be logical to some people.

Xbitlabs said the same thing:
http://forums.anandtech.com/showpost.php?p=31495013&postcount=559

The AMD fan is good for moving heat outside your case, but at the expense of noise as its pushing a lot of airflow in one direction, out a small outlet, you can hear the rush of air and its loud.

NVs setup moves half the heat into your case but the airflow is not as noisy. It's tough to call which is better.

Enthusiasts setup their rigs with adequate airflow already so it matters less if extra heat is going into the case. But, also the added heat may limit other component's ability to OC so thats also a negative.

 

[kevinsbane]

To be 100% accurate you are right, Heat is used as a term for heat transfer, that is energy transfer across a boundary. However, units of heat and those of thermal energy are one and the same. So, Heat and thermal energy can be used interchangeably, making a distinction that in that case heat is not an abbreviation for heat transfer. Personnaly I dont like term heat, because it is redundant to energy when that energy is transfered somewhere. In real life all energy is always transferred somewhere, with exception of theoretical physics problems which can never exist in real life, hence all thermal energy in real life is heat. Furthermore, the terms Heat flux or energy flux, as well as time rate of heat or time rate of energy or Power are synonymous in real life.

I thought that thermal energy is described in joules (calories?), and heat is measured in watts (j/s)? I concede that this particular case, it doesn't matter as a video card is a steady-state system. It just makes it easier for me to visualize a known quantity of energy and follow it around as it moves and is dissipated.

If what you saying is correct and HD6990@450W has the same power consumption with the GTX590 then the cores of the GTX590 would must have higher temperatures due to smaller heatsing and lower CFMs. if the temperature readings are true in the review, the actual core temps of the GTX590 are at 80C and 84C way less than HD6990@450W (93W and 102W). How do you explain that ??

]I assumed that the GTX 590 has CFM ~= to HD 6990. Perhaps you are reading a different review, as that review shows the GTX 590 has GPU temperatures of 96C and 90C, versus the 6990 of 102C and 93C. I wasn't using the GPU temperatures as a reference though (temperatures, by themselves, don't mean much), just the thermal images. The 590 seemed to be "redder" over a larger area than the 6990. I fully admit that it isn't scientific or anything, I just feel that it is a reasonable assumption given the absence of any other data. Given that it seems the GTX 590 may be using less power than the 6990 OC in this particular test (see next paragraph) it would seem to me this assumption is justified.

Granted we don’t know in what bench they did took the readings and the IR pictures but I will bet that in gaming the GTX590 consumes less power than the HD6990@450W.

They were running "3DMark 2006 Pixel Shader" on the GPU and Prime95 on the CPU. Don't know if there is a difference between that and 3DMark 2006. Their methodology page describes what they did exactly: 30 minutes of this, open briefly to take measurements; run another 15 minutes, then open and take thermal image. If we take their power consumption figures at face value, then the thermal energy difference is 17 watts, along with the thermal energy difference between the top and bottom of the case, which is unknown (I made a WAG @30 watts, feel free to provide your own). So the 6990 has ~50 watts less at the bottom; it might explain the extra blue colour in the lower right corner. So, not as cut and dry as before, but still not indicative that there is such an extra system fan in the case of the 6990. I hope that explains that while an extra system fan cannot be ruled out, there is little reason to believe that any such fan does exist. Putting the extra fan there doesn't fit with the facts we know, and doesn't serve any purpose that I know of unless the reviewer is purposely trying to skew his results (ie, lying).

 

[pcm81]

I thought that thermal energy is described in joules (calories?), and heat is measured in watts (j/s)? I concede that this particular case, it doesn't matter as a video card is a steady-state system. It just makes it easier for me to visualize a known quantity of energy and follow it around as it moves and is dissipated.

Heat Rate is in watts.
http://en.wikipedia.org/wiki/Heat
3rd paragraph "Notation and Units"

For the sake of videocard, think about it this way:
1. GPU produces X number of Joules of thermal energy per second.
2. It must give up to the surrounding area X number of Joules per second.
3. Heat Rate or rather rate of thermal energy transfer from the card to surroundings is defined by:
Q'=A*dT*k = X Joules per second
Where:
Q' is the time derivative of heat (Q'=Thermal energy leaving the card per second)
A is the total area of the cards heat sink fins
dT is the difference in temperature between heat sink fins and air.
k is a constant which depends on heat sink design, density of air, velocity of airflow.

Assuming that for both cards A and k are near equal and ambient temperature is the same, that would mean that the ratio of dTs for the cards is the same as the ratio of their rates of thermal energy (Heat) generation.

Edit:
Now the fun part:
Radeon is at 96C and GTX is at 111.9C.
Assuming the same ambient temperature of 30 degrees ( this may be too high for average temperature of the room, but a good guess for average temperature near the card) we have:
Qr'~(96-30) = 66
Qg'~(111.9-30) 81.9

which means that either GTX is producing 81.9/66 times more thermal energy (Heat) per second, or that the cooler/fan combination on GTX590 is only 66/81.9 as efficient as that on Radeon 6990.

Either case, GTX590 is a HOT item... (Pun intended)

Aditional reading on forced convection, the "k" term in above given equations:
http://canteach.candu.org/library/20043805.pdf

WARNING, above link is not for the faint of heart.

 

[AtenRa]

You are taking the temperatures measured at the back of the PCB (!!...not even the core itself) and taking that as the core temperature?! That in itself in incorrect. The back of the PCB =/= core temperature. You are making some big assumptions in your analysis.

And something else to add, apparently the coolers work in different ways:
http://www.xbitlabs.com/articles/video/display/geforce-gtx-590_3.html#sect0

"The GPUs of the Radeon HD 6990 are cooled one by one: the air flows from the rear heatsink to the front one and is then exhausted. So, we can suppose that the GPUs are going to differ in temperature but most of the hot air will be exhausted out of the system case. The heatsinks of the GeForce GTX 590 are cooled equally because the fan drives the air towards both heatsinks, leaving half of the hot air inside the system case."

I'm not 100% sure if the fan differences they mention are correct (but the large temperature differences in the 6990 GPUs point to that), but that can also explain the temperature differences in the case.

They took the measurements not me and if im not mistaken they used the same method for both cards..

Theirs no way the air flow inside the HD6990 will be re-circulate back after passing the right heatsing and be directed backwards to pass the second (Left) heatsing and then be exhausted out of the back panel and in to the atmosphere. The design doesn’t allow this to happen because all the air volume passing through the right heatsing only have one exit and that is out of the card through the large opening (right side). The same goes for GTX590 as the lower metallic plate directs the air flow left and right in to the two heatsinks and out of the card.

Both cards 12V 8-Pins are shielded and air cant pass through that area.

HD6990

GTX590

Edit: grammar

 

[GaiaHunter]

They took the measurements not me and if im not mistaken they used the same method for both cards..

Theirs no way the air flow inside the HD6990 will be re-circulate back after passing the right heatsing and be directed backwards to pass the second (Left) heatsing and then be exhausted out of the back panel and in to the atmosphere. The design doesn’t allow this to happen because all the air volume passing through the right heatsing only have one exit and that is out of the card through the large opening (right side). The same goes for GTX590 as the lower metallic plate directs the air flow left and right in to the two heatsinks and out of the card.

Both cards 12V 8-Pins are shielded and air cant pass through that area.

HD6990

GTX590

Edit: grammar

The air flow in of the HD6990 isn't like that.

The 6990 fan is a centrifugal fan - that means the air flow changes direction twice - once when entering and second when leaving (forward curved, backward curved or inclined, radial) - the 6990 seems to me forward curved.

The GTX590 is an axial fan - air enters and leaves the fan with no change in direction.