Continuing the discussion from Pose Estimation with DeepStream :
Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson / GPU) • DeepStream Version • JetPack Version (valid for Jetson only)
• TensorRT Version
• NVIDIA GPU Driver Version (valid for GPU only) • Issue Type: Error/Performance/Resources Issue Blog post mentions that TRT engine creation could take between 4-10 minutes. Do I need to free up some resources for this process to complete successfully?
**• How to reproduce the issue ?
To get started, please follow these steps.
Install DeepStream on your platform, verify it is working by running deepstream-app.
Clone the repository preferably in $DEEPSTREAM_DIR/sources/apps/sample_apps .
Download the TRTPose model , convert it to ONNX using this export utility , and set its location in the DeepStream configuration file.
Replace the OSD binaries (x86 or Jetson) in $DEEPSTREAM_DIR/libs with the ones provided in this repository under bin/ . Please note that these are not inter-compatible across platforms.
Compile the program
$ cd deepstream-pose-estimation/
$ sudo make
$ sudo ./deepstream-pose-estimation-app <file-uri> <output-path>
mchi
May 6, 2021, 3:43pm
3
it’s building TensorRT engine from the onnx file, you need wait for some time, maybe 15 ~ 20 mins…
The Jetson Nano is resource constrained. I am running XFCE desktop and have 6GB Swap File, but the the Nano System still killed/terminated the process. Options?
username@nano:/usr/local$ sudo jetson_clocks --show
6> [UID = 1]: Trying to create engine from model files
mchi
May 7, 2021, 1:16am
5
after it’s terminated , could you check the kernel log by command - “sudo dmesg” and check if it’s OOM.
It looks like an OOM issue. I ran task manager and the process nears/reaches max available Memory (3956MB), the system terminates the process. I am not sure if the process is using/leveraging available swap memory. Is there a way to verify? Should I use a USB flash drive for swap instead of the SD Card?
username@nano:~$ free -h
username@nano:/opt/nvidia/deepstream/deepstream-5.1/sources/apps/sample_apps/deepstream_pose_estimation$ sudo jetson_clocks --show
-------------------------------------------------------
[ 1256.804067] Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
mchi
May 8, 2021, 12:59am
7
Maybe you could try to create swap on drive… I suspect it does not work, but I’m not sure…
Or, you could try to remove some applications like below (the data below is not from NANO) to have more free memory.
On system freshly installed by SDKManager, after system just boots up, system consumes ~ 849MB (= MemTotal - MemAvailable = 3924576 KB - 3054852 KB )
# cat /proc/meminfo
MemTotal: 3924576 kB
MemFree: 484128 kB
MemAvailable: 3054852 kB
After removing below packages, system memory consumption reducs to ~ 362 MB (= 3924576 KB - 3554064 KB )
$ sudo apt autoremovepython
# cat /proc/meminfo
MemTotal: 3924576 kB
MemFree: 3506544 kB
MemAvailable: 3554064 kB
How many memory the application consume are close to ’ Pss ’ of procrank .