100% utilization of GPU

Autonomous Machines Jetson & Embedded Systems Jetson Xavier NX
1

I’m doing inference of a modal on Jetson Xavier NX. Is there any way to use all the cuda cores on Inference & to make it faster?

3

Hi,

It’s recommended to check the bottleneck of your app first.
It’s common that the pipeline is blocked by the slow input pre-processing.

To improve this, you can use an accelerated multimedia API instead.
For example: https://developer.nvidia.com/deepstream-sdk

Thanks.