i just buy GTX graphics card. nvidia-smi is handy command for details. when i am executing machine learning code and tensorflow GPU details shows it uses high memory (7000/8000) and low compute utils 5% , low electricity 40w/180w. and when i am doing crypto mining memory usage is low(2000/8000) and high compute utils 99% , high electricity 140w/180w. what is the reasons behind this. is it something related to GPU architecture?
Asked
Active
Viewed 91 times
2
-
No, both tasks require different amount of resources (GPU compute and GPU memory), crypto mining requires lots of compute and almost no memory. – Dr. Snoopy Nov 30 '18 at 13:20