I am doing deep learning computations and after a lot of computations the input freezes. This means that there is no reaction on mouse and keyboard while the system merrily computes along. However, when I do alt-prtscr-REISUB the system reboots. Does anyone what is going wrong and how I can prevent this?
The CPU is an AMD 1950X and during the computations all 32 threads are utilized for about 45%. The temperature rises to 76 Celsius. 2 Nvidia 1080Ti's are used for the computations. Both utilized for 75% on average.
Using ubuntu 18.04 with Mate, anaconda spyder, tensorflow 1.9/keras 2.2.2 or tensorflow 1.12/keras 2.2.4 (both combinations result in the same behavior) and Nvidia drivers version 415.
0 Answers