Zed-ros-wrapper: jetson TX2 framedrop when image topic subscribed by remote machine

[rhklite]

I'm trying to run RTABMAP with a ZED Mini Camera, with zed running on a Jetson TX2, publishing image_topics to a remote machine. I get frame drops when I subscribe to the image_topics from the remote machine, but I don't know how to improve my frame rate.

Setup

• I have a remote PC and a Nvidia Jetson TX2 that's installed with ZED SDK 2.7 and the latest wrapper. The two machines are connected via a 1gigabit D-Link unmanaged network switch. Jetson is running ZED, with the remote machine subscribing to a image and viewing.
• remote message publishing is done via the ros multimaster_fkie package.

Trouble Shooting Process

• On Jetson, fps is checked with rostopic hz /zed/rgb/image_rect. The values shown 720HD@60fps/30fps, VGA@60fps/30fps was good, no frame drops.
• On the remote machine, fps is also check with rostopic hz /zed/rgb/image_rect_color. When testing with VGA@60fps/30fps, frame rate is stable. But drops to 15/10 hz for 720HD@60fps/30fps.
• On the remote machine, when both rostopic hz /zed/rgb/image_rect and rosrun image_view image_view image:=/zed/rgb/image_rect_color is running, the frame rate drops to ~15/10hz for any resolution/frame rate.

My question

1. Is this because of jetson limitation or network switch limitations?
2. Is there anything I could do to improve this on the zed_wrapper side?

I've viewed this thread https://github.com/stereolabs/zed-ros-wrapper/issues/323

• sudo ~/jetson_clocks.sh --show output

SOC family:tegra186  Machine:quill
Online CPUs: 0,3-5
CPU Cluster Switching: Disabled
cpu0: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu1: Gonvernor=schedutil MinFreq=345600 MaxFreq=2035200 CurrentFreq=2035200
cpu2: Gonvernor=schedutil MinFreq=345600 MaxFreq=2035200 CurrentFreq=2035200
cpu3: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu4: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu5: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
GPU MinFreq=1122000000 MaxFreq=1122000000 CurrentFreq=1122000000
EMC MinFreq=40800000 MaxFreq=1600000000 CurrentFreq=1600000000 FreqOverride=1
Fan: speed=255

i understand that computational resource on jetson is an issue when some process is ran natively on jetson. However, I wonder jetson's computation power would still be an limitation if the process is ran remotely?

[Myzhar]

How are the Jetson and the remote machine connected?

If you are using the WiFi, can you please make the same tests using a direct cable connection?

I confirm that your clock configuration is ok

[adujardin]

@rhklite CPU 1 and 2 are offline (https://devtalk.nvidia.com/default/topic/1000345/jetson-tx2/two-cores-disabled-/)
To enable them run :

sudo su
echo 1 > /sys/devices/system/cpu/cpu1/online
echo 1 > /sys/devices/system/cpu/cpu2/online
~/jetson_clocks.sh

When the maximum performance mode is enabled you should see this :

SOC family:tegra186  Machine:quill
Online CPUs: 0-5
CPU Cluster Switching: Disabled
cpu0: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu1: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu2: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu3: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu4: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu5: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
GPU MinFreq=1300500000 MaxFreq=1300500000 CurrentFreq=1300500000
EMC MinFreq=40800000 MaxFreq=1866000000 CurrentFreq=1866000000 FreqOverride=1
Fan: speed=255

[rhklite]

Thank you for the replies

@Myzhar The two machines are connected via a 1gigabit D-Link unmanaged network switch.
@adujardin I enabled the cpus as you described, though there is no improvement.

now the output of ~/jetson_clocks.sh --show is

SOC family:tegra186  Machine:quill
Online CPUs: 0,3-5
CPU Cluster Switching: Disabled
cpu0: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu1: Gonvernor=schedutil MinFreq=345600 MaxFreq=2035200 CurrentFreq=2035200
cpu2: Gonvernor=schedutil MinFreq=345600 MaxFreq=2035200 CurrentFreq=2035200
cpu3: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu4: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu5: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
GPU MinFreq=1122000000 MaxFreq=1122000000 CurrentFreq=1122000000
EMC MinFreq=40800000 MaxFreq=1600000000 CurrentFreq=1600000000 FreqOverride=1
Fan: speed=255

[Myzhar]

Online CPUs: 0,3-5
The cores 1 and 2 are still offline
You will notice it also with the command $ ~/tegrastats

[rhklite]

Sorry, I copy and pasted the wrong stats. The retest was done with core 1 and 2 online.

SOC family:tegra186  Machine:quill
Online CPUs: 0,3-5
CPU Cluster Switching: Disabled
cpu0: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu1: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu2: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu3: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu4: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
cpu5: Gonvernor=schedutil MinFreq=2035200 MaxFreq=2035200 CurrentFreq=2035200
GPU MinFreq=1122000000 MaxFreq=1122000000 CurrentFreq=1122000000
EMC MinFreq=40800000 MaxFreq=1600000000 CurrentFreq=1600000000 FreqOverride=1
Fan: speed=255

[Myzhar]

Still offline:
Online CPUs: 0,3-5
you must get
Online CPUs: 0-5

[rhklite]

you are right!

Though I dont think it complete solved my problem, but it's better than before now. Thank you!