HWiNFO: Monitor the temperature of the hottest sensor on an Nvidia graphics card

HWiNFO: Monitor the temperature of the hottest sensor on an Nvidia graphics card

In the latest beta version of the utility HWiNFO v6.43-4380 a number of improvements and several innovations are now available. The most interesting new feature of the program is the ability to monitor temperature HotSpot GPU for video cards from NVidia. Not so long ago, this free utility made it possible for the owners of Nvidia 3000 series video cards (RTX 3080 and RTX 3090) to monitor the operating temperature of the GDDR6X memory chips (Usually this part of the 3000 series video cards is the hottest when mining Ethereum). GDDR6X memory chips start throttling at about 110 degrees Celsius. The new function is not intended for memory, but for monitoring the GPU (more precisely, the hottest spot on numerous temperature sensors integrated into the GPU itself)

List of changes in the new version of HWiNFO v6.43-4380 Beta:

  • Improved sensor monitoring on MSI H510 motherboards.
  • Fixed + 12V monitoring on some ASUS Z590 motherboards.
  • Improved support for LPDDR4, LPDDR4, LPDDR4X memory.
  • Fixed possible WHEA / system crash during extended monitoring of AMD RX 6000 series GPUs.
  • Improved sensor monitoring on AsRock Z590 Taichi and Z590 PG Velocita.
  • Consolidated AMD NAVI monitoring, added effective frequency and TGP power for NAVI21.
  • Added tooltips for sensors to provide more detailed information on them.
  • Improved the responsiveness of the program interface.
  • Improved handling of hiding and showing sensors individually.
  • Fixed possible freeze when logging sensors with a large set of disabled elements.
  • Added GPU HotSpot temperature monitoring for NVIDIA GPUs.

Tools like MSI Afterburner show the operating temperature of the GPU based on the average of several sensors residing in the GPU. A GPU is a large crystal with many transistors inside, and different parts of it can be under different loads, and some parts can be completely idle from time to time. This all leads to a temperature difference between different parts of the crystal. There are also several thermal sensors inside the GPU that monitor temperatures in certain places on the crystal, and a special algorithm averages these values ​​into one number, which you see as the “real” operating temperature of the GPU.

HWiNFO: Monitor the temperature of the hottest sensor on an Nvidia graphics card
HWiNFO: Monitor the temperature of the hottest sensor on an Nvidia graphics card

The HotSpot temperature measurement, which HWiNFO now reports in the latest beta version, shows the hottest point measured by the sensors on the GPU, and this value is definitely higher than the usual average temperature that you can see in other monitoring programs such as MSI Afreburner. ... Note that it is common to see a large difference (10-15 degrees) between HotSpot and normal average temperatures, however it can be useful for detecting potential problems with the GPU or graphics card cooler while running. If you see a temperature difference of more than 15 degrees and your GPU is not stable, you should pay attention to this. There may be a problem with fans, thermal grease, or some other related to uneven GPU cooling.

As a result, we got a great new tool to help diagnose possible problems with NVIDIA GPUs not working properly when mining. We ran a quick test of the ASUS RTX 3080 TUF Gaming where MSI Afterburner reports 43 degrees Celsius, and HWiNFO talks about 56.8 degrees of maximum temperature at 100% load with a power consumption of about 230 watts. This is the measured difference in 13.8 degrees Celsius between midpoint and hottest point.