I am writing to seek clarification regarding the QUADRO RTX 8000 GPUs. I currently own four QUADRO RTX 8000 GPUs for neural network (NN) training with tensorflow.
Versions:
Driver 535
CUDA: 12.2
Tensorflow 2.15
During the NN training and optimization, I’ve observed that the temperature of the GPUs tends to stabilize around 82/83 degrees Celsius. These training sessions often take long hours, or even 2-3 days, I am concerned about the long-term effects of running code for long hours at this temperature on the GPUs.
Could you please advise if maintaining a temperature within this range for extended periods can damage the GPUs? I would appreciate any recommendations or strategies to optimize the GPU temperature to ensure longevity.