Eiswolf 2 on Strix 3080 Ti with high memory junction temperatures

Asteroid

New member
I installed Eiswolf 2 on my Strix 3080 Ti (purchased late August) yesterday and I have been seeing an abnormally high memory junction temperature after the installation. The memory junction temperature is sitting at 60 C when idle, and reaches 90 C when under 100% load during benchmarks. I am certain that I correctly installed all the thermal pads (I had checked multiple times and I also had my friend helping me double check), and I am confused why the memory junction temperature is still high. However, all other components on board seem to be fine, with GPU core only sitting at low 30s and GPU hot spot sitting at low 40s.

If anyone has any suggestions on potential fixes, or if anyone could let me know whether this is an expected temperature or not, it would be most helpful! Thanks!
 

Gorin

New member
Hi. Same here. I've also double checked pads alignment etc. Even changed the pads itself, but VRAM temps are still sitting at 90s under load. On the other hand, this doesn't affect stability. I've ran multiple heavy games for hours and everything is fine.
 

Asteroid

New member
Yeah technically this should be within the official specs of VRAM which thermal throttles at 110 C, but my friend is only seeing around 35 C for the VRAM on his 3090 with an EVGA water block. I also noticed that the stock Eiswolf 2 thermal pads for the VRAM (1 mm) are thinner than the stock Strix thermal pads (2 mm), and I am curious whether this is an intended change by Alphacool engineers, or this is simply a design oversight.
 

Gorin

New member
Exactly! My guess also is that pads' size in current Esiwolf revision is wrong. I've noticed some threads here with similar complaints, so lets hope that this issue will be resolved.
 

Thomas_S

... the nice marketing guy next door
Staff member
Hi, when you checked the heat conducting pads, did you also check the imprints of the components? Was there enough contact? @Asteroid , for design reasons, heat conducting pads are not changed. Changes there always have a functional background.
 

Gorin

New member
Hi, when you checked the heat conducting pads, did you also check the imprints of the components? Was there enough contact? @Asteroid , for design reasons, heat conducting pads are not changed. Changes there always have a functional background.
Yeap, they had visible and clear imprint. Unfortunately I forgot to take pictures (
 

Thomas_S

... the nice marketing guy next door
Staff member
Hi. Same here. I've also double checked pads alignment etc. Even changed the pads itself, but VRAM temps are still sitting at 90s under load. On the other hand, this doesn't affect stability. I've ran multiple heavy games for hours and everything is fine.
In the end, it's the result that counts. If your system is stable, then the cooling system is doing its job. One must not forget that 3080/3090 are true "heat monsters" and the ambient conditions also play a significant role. (Airflow, ambient temperature etc...)
 

Asteroid

New member
Hi, when you checked the heat conducting pads, did you also check the imprints of the components? Was there enough contact? @Asteroid , for design reasons, heat conducting pads are not changed. Changes there always have a functional background.
Hi Thomas, sorry I didn't take apart the cooler once I assembled it. But I'm sure I placed the right thermal pads in the correct location, and I also tightened the screws as much as I could. If there is no design issue with the stock thermal pads then I assume the memory junction temperature I saw is expected. Thanks for the help!
 

Asteroid

New member
Just a follow up on this topic: I checked my Nvidia Control Panel setting, and it seemed that I selected "prefer maximum performance" for the "power management mode" item. And all previous temps were measured under that setting.

After I restored default setting which selected "normal" for that item, I saw a significant drop in all temp measurements across the board. Now I'm only measuring around 40 C on the memory junction sensor when idle, and the core temps dropped even further down to about 31 C when idle. So I guess the power setting really does make a huge difference in terms of temperatures and power consumption.
 
Last edited:
Top