sample_drivenet crashing badly

Autonomous Machines DRIVE PX2 DriveWorks
#1

Hello,

Today I tried to execute every samples provided with the DriveWorks SDK. All was going well.
But when I tried to execute the sample_drivenet file, this one was crashing after 2s and also crashing my computer (which rebooted).
The command I used on Linux to run it looks like:
./sample_drivenet --video=<rccb.raw> --stopFrame=0.

The only .raw video file provided with the SDK being very short (max 0.5 second of video), I though the crash was due to an excessive looping of this one (the --stopFrame flag provided being 0). So I changed my command into:
./sample_drivenet --video=<rccb.raw> --stopFrame=1.
to play only one frame of video and prevent it from looping very quickly. Still, the video kept lopping very quickly in the display window (3 or 4 time before crash). And from what I had the time to see before my computer shut down, the DriveNet recognition wasn’t occurring.

=> 1: How can I get the console error message, given that the entire computer crash before I even have the time to read more than a single line in the console ?

=> 2: Can it be due to a CPU overcharge ?

=> 3: Can it be due to a GPU orvercharge ?

=> 4: Can it be due to a RAM overcharge ?

Thank you.

(NB: I also tried it with a .h264 video, for the same result, but the drive net recognition was working before the crash)

#2

Dear bcollado-bougeard,
I do not see any problem using same commands on my machine. Can you please share your system details(like RAM, OS, GPU etc). Also, please check if this occurs when running compute intensive samples(matrix multiplication), graphics samples from CUDA samples. You may also check system logs to get information about the crash.

#3

Hi,

My system infos:

  • System: Linux x86_64

  • Kernel version: #40~16.04.1-Ubuntu SMP

  • Kernel release: 4.13.0-36-generic

  • CPUs infos (x2):
    Architecture: x86_64
    CPU op-mode(s): 32-bit, 64-bit
    Byte Order: Little Endian
    CPU(s): 24
    On-line CPU(s) list: 0-23
    Thread(s) per core: 2
    Core(s) per socket: 6
    Socket(s): 2
    NUMA node(s): 2
    Vendor ID: GenuineIntel
    CPU family: 6
    Model: 44
    Model name: Intel® Xeon® CPU E5645 @ 2.40GHz
    Stepping: 2
    CPU MHz: 1596.000
    CPU max MHz: 2395,0000
    CPU min MHz: 1596,0000
    BogoMIPS: 4795.54
    Virtualization: VT-x
    L1d cache: 32K
    L1i cache: 32K
    L2 cache: 256K
    L3 cache: 12288K
    NUMA node0 CPU(s): 0-5,12-17
    NUMA node1 CPU(s): 6-11,18-23

  • More about processors (x24):
    Socket Designation: CPU1 PROCESSOR
    Type: Central Processor
    Family: Multi-Core Xeon
    Manufacturer: Intel
    Signature: Type 0, Family 6, Model 44, Stepping 2
    Version: Intel® Xeon® CPU E5645 @ 2.40GHz
    Voltage: 1.1 V
    External Clock: 5866 MHz
    Max Speed: 6000 MHz
    Current Speed: 2400 MHz
    Core Count: 6
    Core Enabled: 6
    Thread Count: 12
    Characteristics:
    64-bit capable

  • RAM: 24 GB

Thank you for your help.

#4

================================== E D I T ======================================
Hello there,

I managed to modify the DriveNet sample algorithm source-code (in “main.cpp”).
I added a simple < std::cin.get(); > command inside the virtual onProcess() method called at each frame by the run() method. By this way, I must press ENTER to allow the processing of each frame. And now I can run the video 3 times without crash.

I recorded the RAM, CPU, & GPU usage during this time, and indeed, are all far to be overcharged.

Keep investigating.

#5

==========<< PROBLEM SOLVED >>==========*

SOURCE: CPU Overcharge / Too high FPS

SOLUTION: Added a 10ms pause in onProcess() => brings the overall FPS to 23.

Kind regards.