A difference like night and day!
First I cleaned the card as usual, i.e. first wiped the roughest parts dry with a soft cloth from the outside to the inside, then cleaned the cracks and edges around the GPU with cotton buds and then wiped everything again with a soft cloth and isopropyl alcohol. Once everything is really clean again and the alcohol has completely evaporated, you can apply a paste that is not too liquid but also not too viscous. The package has a slight curvature on the long side, so I used a little more paste. Thanks to the clamping cross with spring screws and the resulting good contact pressure, this is easily pushed out to the side.
Temperature comparison of old against new
Now it gets exciting and oha! You can really let that go! Let’s start right away with the torture loop, i.e. the real hard case, and this time both measurements are taken in an open setup for direct comparability with the old results. The hotspot drops from up to 100 °C to a full 72 to 73 °C, i.e. by 27 to 28 degrees! The new delta to the GPU edge temperature of a comfortable 60 °C is only 12 to 13 degrees instead of 26 degrees, i.e. halved. The visible drops and dropouts of the card in its original state are also eliminated. That really is a brutal difference!
Let’s now compare the socket and the temperatures on the polymer capacitor on the GPU socket as the hotspot on the back of the board. This drops from just under 66 °C…
… to just under 59 °C, which is also around 7 degrees lower. This automatically reduces all other temperatures in the socket area, right down to the RAM modules.
And what about the noise development?
Here I have some very bad news for the buyer, because the card is so quiet that you can even hear the coils again. Let’s first look at and listen to the card in its original state with its abnormal almost 44 dBA in the Torture Loop. The broadband frequency spectrum also stands for the exorbitant air movement of the three whorls. Below you also have the non-normalized and non-amplified audio recording from the measurement microphone.
And now we come to the conversion result. As the fan speeds are lower, the frequencies of the engine noise are also somewhat lower. The air movement is almost completely absent and the less than 33 dBA is nothing that would disturb you in any way. Mind you, this is the torture loop after 30 minutes! We are now over 11 dBA below the initial value, frightening. Only the coils can be heard again.
Summary and conclusion
Normally I don’t have much time to repair cards from readers. In this case, however, the situation was somewhat different, because (a) I was curious and (b) such investigations and repairs can also be used as practical articles for other readers, who may then be able to help themselves. Then everyone has a nice win-win situation. I have something interesting for the website, the reader can finally gamble undisturbed and potential imitators now know what to look out for. And Asus can only be advised to tighten up its quality management a little more, as I’ve now had to give my RTX 4090 Strix a new paste because the old one had completely bled out. So it’s anything but an isolated case, but an unnecessary cost-down on such expensive cards. Nowadays there are good phase change pads made of corundum and zinc oxide that perform well in the long term and are much more durable. This is really anything but witchcraft, but unfortunately also a bit expensive. Well, not much.
ASUS TUF Gaming GeForce RTX 4080 SUPER OC, TUF-RTX4080S-O16G-GAMING, 16GB GDDR6X, 2x HDMI, 3x DP (90YV0KA0-M0NA00)
 | Lieferzeit 3-6 Werktage | 1219,99 €*Stand: 17.05.24 15:10 |
 | Ab Zentrallager sofort lieferbar | 1273,42 €*Stand: 17.05.24 16:22 |
| nullprozentshop.de | siehe Shop | 1273,43 €*Stand: 17.05.24 15:58 |
135 Antworten
Kommentar
Lade neue Kommentare
Veteran
Urgestein
Veteran
Urgestein
Urgestein
Urgestein
Urgestein
Urgestein
Urgestein
1
Urgestein
1
Urgestein
Urgestein
Urgestein
Mitglied
1
Urgestein
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →