Question87 Celcius Hotspot and 2500 rpm fan speed on a stock brand new GPU ?
Question87 Celcius Hotspot and 2500 rpm fan speed on a stock brand new GPU ?
If you choose the new fan route, it would be wise to opt for improved fans overall, especially if the ones included with the case are the ones that came with it. If they're not available locally, the Arctic P12 Pro multi-packs (both ARGB and non-ARGB) should offer the best value for performance.
Based on what you shared earlier, it seems the issue likely lies in the front panel's ability to contain air effectively. Improving the fans might help, but if the panel is limiting airflow significantly, the new fans could still be noticeable. A quieter fan design might not solve the problem if thermal performance remains similar. Glass insulates heat well, and many aftermarket cooler designs direct heat away from the side panel and motherboard, reducing energy loss. Using a thermal pad instead of paste could also be a better approach than what Asus implemented.
I don't believe it would significantly improve things since the performance was already weak and didn't noticeably changed afterward. Still, if I had to repaste my GPU again, I'd probably choose one of the PTM sheets I own—they appear to work well for bare die and last longer than paste. It's unfortunate that availability varies so much by region.
So you believe the main issue is the case airflow. In my new XPG case, CPU temperatures are 4-5°C cooler, and even with side glass removed and using stock settings, GPU fans still spin up to 2300 rpm while the hotspot stays around 85°C. Removing the side panel doesn’t really change the temperatures much. I wouldn’t mind having an 87°C hotspot if my fans stayed between 1600-1800 RPM. It’s just frustrating because I get 87°C with 2500 RPM, and refunding or replacing it in Turkey is very tough even if you’re right. That’s why I chose to void the warranty by repasting.
If eliminating the side panel results in little impact on performance, it's unlikely the issue involves airflow. Have you adjusted the video card fans to a speed between 1600-1800 RPM and observed any changes?
with the 50% max fan curve I achieve 1675 RPM which is acceptable. The GPU runs at 82 hotspots, 58 GPUs, and interestingly the memory stays around 70°C. I was surprised by the lower VRAM temperatures—this seems like a compromise. Some cards have extremely high VRAM temps while others have strong hotspots, but I’m managing these results with -15% PL and -50 mV. If I switch to stock settings with 50% fan speed, the hotspot drops to 92-93°C. I might add another fan at the bottom for better airflow and consider using a side glass panel off. Now I really dislike using a GPU when it lacks stock settings because I often encounter errors during gameplay.
I wouldn't stress too much about the hotspot since the delta remains stable. It appears you might experience improved temperatures with increased airflow, but you'll probably still need to adjust the fan curve similarly.
It seems there could be a contact problem with the cooler as well. Although the cards sometimes show a large delta difference, your setup seems to be having trouble maintaining overall cooling.
Is the room where the system is located hot enough? That wouldn't seem to be the main concern either.
It's around 20°C, and there seems to be a contact issue. You can try fixing it yourself or rule it out. Should I just take it apart again to check if the thermal paste spread properly? When I first opened it, it felt like it was pumping out, with a very thin layer on the chip that looked dry. I was confident the temperatures would improve significantly after repasting, but nothing changed. Since I used the Arctic MX6 for pumpouts, what does a hot spot mean? Is it just a random temperature reading on the actual chip where I applied the paste, or is it related to the VRAM? I haven't used any DDU from 2060S to 9070 or 9070 to 9070XT, and instead manually deleted the driver to install for the 9070XT. Is there any chance this could be causing high RPMs or a hot spot?
It should represent the highest temperature reading on the GPU die. GPU temperatures reflect all sensor data and identifying the main hotspot helps spot poor contact issues. For the 9070 XT, a significant difference appears, so it's important to check that the change in temperature remains stable across various fan speeds during operation.
MX-6 should be sufficiently robust to avoid such concerns. I previously used MX-5 with my 1660 Ti and later 3080 models without any problems.
This wouldn't cause much strain on the cooler contact, but slight pressure could disrupt it. On my 1660 Ti, two VRAM chips weren't fully covered, prompting me to explore fixes since the cooler layout didn't accommodate them well. A tiny 1mm shim on one heatpipe caused a noticeable imbalance, raising temperatures by about 2°C under load.
As I mentioned in some discussions, using a PTM pad instead of applying paste directly to the die could be better. Thermal Grizzly KryoSheet is another option, though they're difficult to handle, expensive, and availability can be limited. Just ensure you're confident it won't damage the components before applying it.
This approach might affect fan settings, so resetting all drivers and reinstalling could help without impacting the hotspot.
I updated my PC to W11 and installed W11 instead of 10. A Reddit user from Polish Tech shared a benchmark comparison for all models, showing similar temperatures around 2500 RPM with an 85+ hotspot. It seems I wasn’t overreacting too much.