I don't know if any of you have experienced such troubles?
After opening the task manager of the laptop, I found that my game is obviously an independent graphics card, such as my RTX4060, especially when running a deep learning model, specifying the device as cuda:0, the process shows that the independent graphics GPU1 has no utilization rate, and Has the nuclear display been highly utilized? Even the code will report an error, suggesting that there is no CUDA available?
In fact, the working modes of laptops and desktops are not exactly the same.
Desktop: After the image is loaded by the independent graphics card, it is directly sent to the monitor;
while for a notebook, it is: independent graphics card -> core display -> finally to the monitor.
In that case, the performance of your favorite machine will certainly be affected!
My own savior Y9000P bought it and felt that it was not silky, so the problem lies here.
After enabling the direct connection function of the independent display in LEGION ZONE (different for each brand):
It can be seen that only one discrete graphics card is working as GPU0.
