My Nvidia GPU doesn't work with full power to train models

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

I don’t know why, but my GPU even after downloading Nvidia kits to use it to train models, it still doesn’t work with the full power. Sometimes, 50% CPU & 10% GPU. I appreciate your help!

2

This is not a course related question. That said, could you share more details about your OS / GPU / tensorflow version to provide more context? You could always try stackoverflow / google.

3

I know, sorry for that. But, I use Jupyter to code so I just asked. I tried google and stackoverflow but nothing worked. I have win11 OS & Nvidia GeForce MX230 GPU.

4

No need to apologize. Let’s see if someone has a windows system to help you out. Good luck.

1 Like
5

This is not a course related question.

6

There is no problem to ask a question like this. Balaji’s point is just that there is no guarantee you will get an answer, but it is perfectly fine to ask. The mentors may not know the answer, but anyone in the community is welcome to answer if they want to help and have any useful information to contribute. A lot of people have experience running training on their own GPUs, so maybe you get lucky and one of them is listening here.

1 Like
7

Thank you for your concern. I really appreciate that. I knew it wasn’t a related question to the course but I just hoped someone could help.

8

I can’t tell how much analysis is behind this one 50% v 10% data point. But I’m not sure data like that tells the whole somewhat complex story. Have you looked at this profiler guideline: عملکرد TensorFlow GPU را با TensorFlow Profiler بهینه کنید  |  TensorFlow Core ? It might help you get more detailed insight. For instance, you might find out that your GPU is input bound.