GPU is getting utilized 99 % with face detection code using caffe model

Autonomous Vehicles DRIVE AGX Orin DRIVE AGX Orin General
1

Please provide the following info (tick the boxes after creating this topic):
Software Version
DRIVE OS 6.0.10.0
[1] 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
[1] 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)
[1] DRIVE AGX Orin Developer Kit (not sure its number)
other

SDK Manager Version
2.1.0
[1] 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

Issue Description
Hi team, I used cuda backend for face detection using Caffe model. Using tegrastats, I monitored its CPU and GPU utilization. I observed CPU utilization is 21.11% and GPU utlization is 99%. How can it be possible?

3

Dear @akshay.tupkar ,
GPU utilization means how many sampled cycles are active during a period.
so even GPU utilization shows 99%, it doesn’t mean GPU has achieved its maximum computing power.

4

Dear @SivaRamaKrishnaNV
I am sharing a tegrastats report while the code was being executed. Could you please take a look at it?
face_detection)cuda_tegrastats.txt (15.2 KB)

5

Dear @akshay.tupkar,
I see several CPU nodes are in use. Is your application has any CPU executable code? Can you profile the application using nsys for more insights?

6

Yes dear @SivaRamaKrishnaNV , some part of code is cpu executable. CUDA is geting used in backend.
face_detection_cuda_backend.txt (3.8 KB)

7

I notice this is a opencv sample and net.forward seems to use GPU and other functions seems to use CPU. So both CPU/GPU utilization is expected. Regarding the utilization percentage by tegrastats, it is clarified already.

10

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.