RX588 temperature
In a few minutes RX588 warmed up to 80 °:Spoiler
Is this the norm?
Spoiler
Today I changed the thermal interface:
Spoiler
Before that it warmed up even faster:
Spoiler
I would not like to repeat this:
Spoiler
Do you really have to change the thermal pad to this one ???
Spoiler
Len4ik00N The
salt is that the sensor gives the SAME temperature. In the first post the screen shows this. At 7950 the temperatures differ
Denis Kyokushin
It gives not the same, but one, perhaps the general or the highest, as mentioned above. Those. you touched the card in the area of ​​the chip, and 81 degrees could be at vpm. Try to repeat the "experiment", just use another program that will give separate readings from the sensors (for example, Aida), at the same time you will see how many of them are obviously lying and how quickly the chip and the VRM zone heat up / cool down in relation to each other.
Len4ik00N
This card will not cure the same (. The difference from 7950 is noticeable
Denis Kyokushin
The 7950 has less power consumption, according to the screenshots, and the vrm temperature is closer to 80 degrees. Again, 80 degrees is equal in the case with a load in the region of 80 degrees.
Denis Kyokushin
To start "treatment", you first need to make a "diagnosis", and we have been dancing around the "symptoms" for a month already))
It makes no sense to directly compare the temperatures of 7950 and 580 - the cards are very different, so you shouldn't torment the "old woman"
Len4ik00N
Len4ik00N wrote:
To start "treatment", you first need to make a "diagnosis", and we have been dancing around the "symptoms" for a month already)) I
just touched it again, hot RX588. So, the sensor doesn't lie:
Spoiler 
I'm thinking, maybe thin find a plate and put it between the chip and the heatsink?
Marsj
Maybe the comparison is incorrect, but I wonder how hot it is. It turned out to be useless (. Apparently, poor contact between the core and the heatsink. I bought thermal pads softer, 60 Shore. It can be seen that the thermal paste remains both on the core and on the heatsink.) And still the result is the same (
Denis Kyokushin
Of course, over 2300 shots are hot. Make an experiment: put 30 frames in RivaTuner.
If the temperature does not drop, then the heatsink will not reach the chip.
Marsj The case is
not in the frames
Tekeshi_Tetsuo The
BIOS failed to reflash :
Spoiler 
I tried it like this:
Spoiler
Also failed)))
original BIOS 
Thermal paste on the core and heatsink
Denis Kyokushin
I would not go into the BIOS, you can distort the card if there is no programmer. And then what's the point if the bios is native? Even if you sew it under another card, it will be crutches, not a solution.
In general, the temperature of 80 is far from trotting (if they trot at all), but the fact that the card drops frequencies with each step of temperature rise, it seems like the norm ... Like yes, it is throttling, but in fact it does not reflect at work, as is the case with the cpu ...
Xs what to advise))
I honestly already doubt that this is a problem, since according to tests, the card heats up in such games. Run it normally, see what peak it reaches and if there will be a stability problem.
Denis Kyokushin
Denis Kyokushin wrote:
... the sensor gives the SAME temperature ...
Can you make a screen of the latest version of one of these programs under load?
Spoiler
Denis Kyokushin
Now I will throw off the video there everything will explain in detail why you can not re-flash the BIOS, you just need it through the command line
IngwardIn
Up to 1100MHz with kopecks and heated up to 81 °. And all this in about a minute. As I understand it, I will have to take a 0.2 thick thermal pad. Dear, the infection
SonyK_2
Ok, but not now
Tekeshi_Tetsuo
His method of flashing did not help me
Denis Kyokushin
Denis Kyokushin wrote:
I'm thinking, maybe I can find a thin plate and put it between the chip and the heatsink?
This can solve the problem, and once and for all, if you do not guess with the thickness and get chips on the crystal)) Moreover, you do not need it - judging by the photo, and the thermal paste is smeared normally, and there is contact between the chip and the radiator. It turns out that the thickness of the thermal pads was chosen correctly (at least in the larger direction).
I did not see the fan speed on the screen with the witcher's menu, but I understand that the screen with monitoring, where the fan speed is 3.5k, is for him. For such revolutions, 81 degrees in the game menu is a lot. I saw this on laptops when they had problems with the cooling system (more precisely, with heat pipes). Symptoms are similar: with "fresh", normal thermal paste and normal contact between the chip and the heatsink, the "patient" heats up strongly and quickly, even at low loads, and cools down for a long time after they decrease. But laptops usually have one tube per chip, so the "symptoms" are more pronounced and it is easier to diagnose with "improvised" means. If you cannot solve the problem in some other way, or just decide to check the option with heat pipes, you can write me a couple of options in the ls-podkin how this can be done,
PS It is strange that in Aida there is only one sensor, and in HWiNFO 6, but for all the same temperature. I also have a card from Gigi, only 470: it has separate sensors 1 for the GPU and 2 vrm (though one for the number, because it is always stupid to lie). By the way, recently I also had to change the thermal pads on it.
Len4ik00N
Len4ik00N wrote:
I did not see the fan speed on the screen with the witcher's menu, but I understand that the screen with monitoring, where the fan speed is 3.5k, is for him.
Yes, this is 100% rpm. Although, for some reason, in the firmware 100% = 3000 rpm
Len4ik00N wrote: The
symptoms are similar: with "fresh", normal thermal paste and normal contact between the chip and the radiator, the "patient" heats up strongly and quickly, even at low loads, and cools down for a long time, after their decrease.
If you run any FallOut 3, then the "patient" feels quite normal, because the core does not overclock to full. If you turn off MSI Afterburner, the card gates will stop in a quiet state. That is, at rest the card works as expected.
Len4ik00N wrote:
By the way, recently I also had to change the thermal pads on it.
And what did you set, if not a secret?
Hemorrhoids with thermal pads, if you do not know the thickness of the pads from the manufacturer. You need to know the values ​​of the thicknesses of several gaskets and the place of their installation, if you do not guess with one of the thicknesses of the gasket and the installation location, then this will be your case.
Denis Kyokushin
Denis Kyokushin wrote:
And which ones did you set, if not secret?
It's no secret, but I can't say, because I don't remember the name)) I just measured the thickness of the old ones and in the nearest store I chose what I liked according to the characteristics. I remember that they were blue, rather soft (to the touch almost like the factory ones), thermal conductivity in the region of 4 W / (m * K). I set everything up without problems, the flight is still normal
