RX588 temperature

Len4ik00N
I liked the idea very much with plasticine. I will use it

IngwardIn
Thanks for the video👍. I liked how the propellers were disassembled in the second video. Kick-ass, he has a bunch of tools. Just like him, the thermal pad stretched out when removed from it And on the first, for two RX470s, the VRM heats up to 70-71 °, in the third up to 80 °. It turns out that for some RX cards 80 ° on VRM is the norm? As with the plasteline, I'll deal with the thermal spacers). disassemble again. Yes, and I do not have a backplate). And I did not have the old thermal pads.
But for some reason, the video does not talk about the main thing: about the thermal conductivity of thermal pads.

Denis Kyokushin wrote:
And on the first, in two RX470s, the VRM heats up to 70-71 °, in the third to 80 °
80, this is not very sure, but the permissible temperature, over 90 is already bad, probably depends on the state. Does the VRM temperature greatly affect the overall GPU temperature?

I myself am an adherent of low temperatures, I have such a leap ... and I think, since there is anxiety, it is not without reason, apparently the card was colder before? And besides, having looked at the tests on YouTube, I saw that the card can work in the region of 70-75 degrees. Although 80 degrees is not a critical temperature, but if the card can be at 75, why not?))

In my opinion, an error of 3-5 degrees can be allowed between the same specimens, but in general, the temperatures in the drain for all specimens should be + - the same, since the cards initially operate at the same voltage. And if the difference is greater, it probably depends on the state of the graphics chip or some other component. I do not take into account the breathability of the case, as it is obvious.
Denis Kyokushin wrote:
But for some reason they don't talk about the main thing in the video: the thermal conductivity of thermal pads
Well, that's what I managed to google.
Spoiler At the Laird company - the colors used for thermal pads are very scarce. Almost all of their light gray or white gaskets have a wide range of thermal conductivity values: from 1 to 5 W / mK. And only thermal pads with thermal conductivity of 2.8-3.0 W / mK have pink, blue, blue-violet and dark gray colors.
And if I saw the color correctly, then it turns out that the author of the RX470 card used 2.8-3.0 W / mK?
Drift ice generally says 13 W / mK😲

Probably on the power circuit, the softer the better .... maybe something about 20-35 units. Shore.
And if we proceed from the experience of AnPei, then it turns out 35 units. Shore will be sufficient and the greater the thermal conductivity, the better. Or 3-6 W / mK.
You can try drift ice as well, using multiple layers of spacers to achieve the desired thickness.
And on the memory chips I think that the spacers are suitable.

Here are the recommendations for installation .... but I did not find direct recommendations for hardness and thermal conductivity.
Spoiler Installation of the thermal pad should be carried out in the following order:

Cut off the required amount of material, the size of a chip or slightly larger.
Remove the film from the adhesive surface of the thermal pad (if present).
Having previously slightly bent the gasket, like a roll, lay it, starting from the edge, on the surface, i.e. the thermal pad must be rolled out on the surface of the chip. This is necessary to remove air at the point of contact between the thermal pad and the chip.
Holding the gasket by the edge, remove the second protective film (if present).
Install the radiator.
When installing a new thermal pad, you should pay attention to the fact that its thickness should be 0.1 ... 0.5 mm more than the thickness of the deformed part of the old pad.



But what if the thickness of the thermal pad is unknown, or there is a thinner thermal pad? In this case, you can proceed as follows.

Install a 0.5mm thermal pad on the chip as described earlier.
Install and secure the cooling system radiator with screws.
Unscrew and remove the radiator.
Check if the thermal pad was pressed by the heatsink, making sure that there is or is not a deformation area left on the thermal pad by the chip.
If the thermal pad was not pressed, install another thermal pad, on top of the previous one, according to the above instructions.
Repeat steps 2-5 until the thermal pad is pressed.

Source
Spoiler http://www.mirpu.ru/nout/264-thermointerface.html

IngwardIn
IngwardIn wrote:
When installing a new thermal pad, you should pay attention to the fact that its thickness should be 0.1 ... 0.5 mm more than the thickness of the deformed part of the old pad.
That's probably why it was not possible to solve the problem again. Plasticine showed between the VRM and the radiator 1 mm, and between the chokes and the radiator 1 mm. I threw another 0.5 mm (I still had a bit of thermal spacers) on the chokes and VRM, but in the end nothing and did not dare:
Spoiler
I understand that, I need 1.5 mm of exactly the same thermal
pads on the VRM and the throttle ( IngwardIn wrote:
Repeat steps 2-5 until the thermal pad is pressed I
also think about it, since 13W / m * k and 45 bl by 1.5 mm I cannot find

Denis Kyokushin wrote:
I understand that, I need 1.5 mm of exactly the same thermal pads on the VRM and the throttle (
Well, if you buy, preferably softer, like the same ones from the factory.
It's a personal matter, of course, but if there isn't, maybe you shouldn't rush what would these experiments no more money spent?

For information:
Or maybe it is unnecessary to worry I've looked a couple of tests such as the video card, and noticed that it by default it gets warm in many projects to 80+ and relatively open housing in? package with closed muzzle.
Spoiler

IngwardIn
Why didn't it immediately warm up to 80 °? At the start
Spoiler

Denis Kyokushin
Denis Kyokushin wrote:
Why didn't it warm up to 80 ° C at the start
So the cooling system is working properly.
In videos with tests where the temperature reaches 80, the card is already warmed up and in an amicable way the gameplay is shown after a while.
And in Witcher 3, the card quickly reaches its peak, because most of the time the game loads the card at 100%.

I do not insist and I do not take the responsibility to draw conclusions, since I am a humanist.)) I just gave some information.
Of course, it's more correct to redo everything as expected. Moreover, the cooling of the power circuit is provided from the factory. But I admit that for some time you can use the card like this. Since, even if the power supply circuits do not reach the radiators, it will also be cooled indirectly by the air flow from the second turntable.

In general, am I thinking correctly, but if the power circuits do not touch the radiator, then the values ​​in the afterburner should be less, since they do not transfer heat to it.
For example, Palit was so cunning on a 1650super with one fan, only with memory chips. The radiator simply did not lie on them and the afterberner gave out 65 degrees.

IngwardIn
IngwardIn wrote:
Since, even if the power supply circuits do not reach the radiators, it will also be cooled indirectly by the air flow from the second turntable.
Will not:
Spoiler

IngwardIn wrote:
Of course correct to change everything as it is necessary
only that the sign in the DNS, bought all godnotu), not rigid uspel.Ostalos guano

Denis Kyokushin
Denis Kyokushin wrote:
... why did it not immediately warm up to 80 °? ...
The temperature in the room (where the system unit is located) is "right away" and now it is different? - if it has increased, there is nothing surprising.
... why there are no such temperature jumps on the old HD7950 ... The
cards are different. And do not forget that every year manufacturers are trying to save more and more on anything - perhaps they used a lower-quality thermal interface ...

SonyK_2
We recently turned off the heating. As if, testing not at an air temperature of 30 °
SonyK_2 wrote:
Different cards
The 7950 from HIS has one valve
SonyK_2 wrote: a
lower quality thermal interface ... The thermal
paste is the same, there
are no thermal pads on it Here is an old video , played for three hours, the temperature rises to 75 °:
Spoiler
Quite different indicators. The radiator is not applied to the VRM, there is just air under the casing the cooler drives the
old cards are more solid, and the new ones are fragile

Denis Kyokushin
Denis Kyokushin wrote:
... the test is not at an air temperature of 30 °
How much is it different from the tests before?
... Old cards are more solid, but new ones are some fragile I
wrote about this ...
SonyK_2 wrote:
... every year manufacturers are trying to save more and more on anything ...

SonyK_2
SonyK_2 wrote:
How much is it different from the tests before?
I will not name the degrees exactly, but it became cooler

Denis Kyokushin
Denis Kyokushin wrote:
... it has become colder
If the temperature is lower than then (to be sure, it is better to check with the room thermometer), the cooling of the vidyuhi does not cope. Provided that there are no changes with the general cooling - for example, if one of the case fans has become worse to work.

Rumor has it that the new port, Days Gone, will allow the chip to warm up. Test, do not test, cooling will not help.

Marsj
That is, the game will change the stock value of the TPP of the GP? Well genius 🤣 how are you so smart? ))

Denis Kyokushin
Ordering is not an option? In the meantime, the pads are on the way, during use, for your peace of mind, you can reduce the power limit.
In general, have you tried to watch the temperature of the VRM in the same state as the cooling system?
I wonder what temperatures are there.

IngwardIn
IngwardIn wrote: And I
wonder what temperatures are there.
Exactly the same as the GPU: 80 °. As shown by HWinfo
IngwardIn wrote:
Order is not an option?
1100r. I need it for 730r. She is not

Denis Kyokushin
If the load does not rise any more, like norms, for temporary use. Some cards from the factory have such values.

IngwardIn
Marsj
Tekeshi_Tetsuo
Listoman
KENIG_39
semferot48
SonyK_2
Thanks to everyone who replied. I reassembled my "bike", bought thermal pads, but it turned out that there is a lot of 1.5 mm thickness on the VRM and throttles. I put 1mm on them, and for memory 0.5mm, I knocked a couple valve 120, and in the time of the test in The Witcher 3, first a couple of times a black screen, and a third GPU frequency dropped to 700 kopecks MGTs.Vskryl, while GPU card bad contact with the heat sink:
Spoiler

That would be brutal, if the GPU degraded

AnPei
Now I did the same as you did: leaving 0.5 mm each on the memory banks and thermal paste on the GPU. Here's what came of it:
Spoiler
It seems that the GPU again does not really reach the heatsink. On the go, the memory banks need a 0.2mm thermal pad, or some super soft 0.5🤦‍♂️

Denis Kyokushin
Super soft at 0.5-1.0mm. before the mining boom there was no and now even more so I had an idea to put a soft one, but I put what was in stores because I didn't want to wait long. You have poor contact of the crystal with the radiator, you can do this - stick small pieces of thermal pads on memory banks on them threads with ends to release so that you can pull, small drops of thermal paste in the center and in the corners (while only the contact of the crystal with the radiator is of interest) we collect a little by pressing radiator to the board, we need to pull the threads so that the threads do not stretch. Next, we disassemble if the thermal paste is smeared as it should, then we take on the thermal pads, for all cases (mining, games), thermal pads are needed softer, but for games you can try to use what is, put them on a flat surface by pressing them with a flat plate and beat them to the desired thickness.

Denis Kyokushin
Denis Kyokushin wrote:
It seems that the GPU again does not really reach the heatsink. In the course, the memory banks need a 0.2mm thermal pad, or some super soft 0.5🤦‍♂️ Is the
contact between the chip and the heatsink normal without thermal pads ? And what about the textolite: are there any distortions (especially in the GPU area)?