Please provide the following info (tick the boxes after creating this topic):
Software Version
DRIVE OS 6.0.8.1
DRIVE OS 6.0.6
DRIVE OS 6.0.5
DRIVE OS 6.0.4 (rev. 1)
DRIVE OS 6.0.4 SDK
other
Target Operating System
Linux
QNX
other
Hardware Platform
DRIVE AGX Orin Developer Kit (940-63710-0010-300)
DRIVE AGX Orin Developer Kit (940-63710-0010-200)
DRIVE AGX Orin Developer Kit (940-63710-0010-100)
DRIVE AGX Orin Developer Kit (940-63710-0010-D00)
DRIVE AGX Orin Developer Kit (940-63710-0010-C00)
DRIVE AGX Orin Developer Kit (not sure its number)
other
SDK Manager Version
1.9.3.10904
other
Host Machine Version
native Ubuntu Linux 20.04 Host installed with SDK Manager
native Ubuntu Linux 20.04 Host installed with DRIVE OS Docker Containers
native Ubuntu Linux 18.04 Host installed with DRIVE OS Docker Containers
other
Hi,
I’m trying to understand GPU memory utilization for specific application that are run on the Drive Orin device. Currently, I have looked at tegrastats and Nsight Systems. To begin with, does the Orin have a shared memory between CPU and GPU? Because when I use “htop” to monitor processes on the CPU, it shows the same amount of total memory (I assumed to be CPU memory) as the GPU total memory listed on Nsight Systems (both 27.8G).
On tegrastats, I was able to see RAM changes when I run a program that uses the GPU. If the CPU and GPU has a shared memory, then this is enough for monitoring GPU memory utilization (right?). On Nsight Systems, I was able to get a field named “CUDA Memory Usage”. I thought this might be GPU memory usage. But the number I get (for my application around 140MB) from Nsight Systems doesn’t really make sense because we are deploying a rather big neural network with TensorRT.
To conclude, I mainly have 3 questions:
- Does the Orin CPU and GPU have a shared memory?
If not, then - How can I monitor total GPU memory utilization on the Orin device (something similar to nvidia-smi and nvtop)?
Many thanks in advance!