Jump to content
1 post in this topic

Recommended Posts

Issue synopsis

Hi, I have an issue with my GPUs that I haven't managed to resolve as of now. I have two GPUs, and when I render something they start to overheat (especially one of the two since it is drawing the heat from the other below it). When I render in macOS due to overheating for some reason all my USB devices and Bluetooth stop responding. The OS does not freeze since I can see the progress of the rendering moving normally and I also can remotely log in with Apple Remote Desktop since all my input devices are not working, I suspect due to high temperatures reaching inside the PC. The temperatures of GPU are going high as much as 90+ and 100+ ºC for the GPUs. I tried doing the same render in Windows (Bootcamp dual boot), and not only did it work fine, but it worked almost 3 times faster. Both GPUs didn't go above 80ºC and the average temp was 68 - 75ºC during the whole rendering process. That is of course because in Windows fan curves are working and when temps are rising the fans' RPM ramps up. Under macOS, both GPU fans and all other fans are not ramping up when heat rises.

My build

The parts of my Hackintosh are listed here, the main components are:
 

  • Intel i7 13700k
  • ASRock Z790 LiveMixer (BIOS v13.02)
  • 4x16GB DDR5 Kingston Renegate Silver 6000MT/s (Running at 3600MHz since all four slots are occupied)
  • 2x RX 6600 XT 8GB PowerColor Red Devil
  • Focusrite Scarlett Solo 3rd Gen
  • WD Blue SN570 2TB M.2 (as main storage)

 

I'm currently sitting on macOS Sonoma v14.3 but those issues have been happening since Monterrey. I have uploaded my EFI here for anyone interested to see my work. I have changed any sensitive info like MLB, Serial etc.

My actions so far

I have tried using the Henbury patch in DP, I noticed that one GPU is indeed spinning the fans all the time, but, the other is spinning the fans at a very very low speed, like they are about to stop. Under Windows, I don't have this behaviour. I have tried to extract the softPowerPlayTable via TechPowerUp GPU-Z and MorePowerTool where I disabled Zero RPM for both my GPUs from the Features and Fan tab. Then, I placed the PP_PhmSoftPowerPlayTable key with the equivalent hex data under each GPU device path. That didn't do anything, therefore I removed it. Maybe I should build an SSDT file instead of placing that in the OC config, but I'm kind of not sure how to properly do that, therefore I asked a question about that here. Also, for testing purposes, I have tried disconnecting all monitors and leaving only one without plugging in two GPUs to isolate each. I also tried this, but it didn't change anything.

 


Thank you all in advance!

PS: To avoid any arguing about having two GPUs is not that efficient (and in most cases I would agree), the reason behind my having two GPUs is that I'm a multi-monitor addict (the more the better), so far I have occupied 3 slots for my second GPU (2x4k@60Hz monitors and one ultrawide 2560x1080@60Hz) and my main GPU I have two more monitors (1x4k@60Hz and one ultrawide 2560x1080@60Hz). In addition, I use some apps like Final Cut Pro and Topaz Video/Photo AI and I also do some academic research where I train some simple AI models, therefore, all these benefit from multi GPU setup.

Edited by panosru
Link to comment
https://www.insanelymac.com/forum/topic/358503-gpu-fan-curve-overheating/
Share on other sites

×
×
  • Create New...