• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 97 subscribers
  • Views 254 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TDA4VM: TDA4VM-SDK7.3: Capture/Display 5 sensor simultaneously apper crash

chen rui
Prodigy 60 points
Part Number: TDA4VM

Hello TI，

    I am trying to capture/display 5 cameras on TDA4VM-SDK7.3, but i running the program of multi_camera apper error, as shown in the picture below.

    

    Currently, I try to run and display five sensors simultaneously on the TDA4VM.  I  set the sensorObj->num_cameras_enabled is 5, and i set the capture two CSI.Three sensors are connected to CSI0, and two sensors are connected to CSI1. Unfortunately the PSDK crashes when it reaches the function of ownGraphCreateIntermediateDataReferenceQueues()

    Can you help me answer the following questions:

(1)Does the capture module on TDA4 support simultaneous capture of more than 4 channels of image data? If you support it, can you help me with my current situation?

(2)If i want capture more than 4 channels of image data, I shoule use two capture module in one program？

    thanks,

    rchen.

  • 0
    TI__Guru**** 476595 points

    rchen,

    Yes, multi-channel example does support more than 4 camera, infact, it supports upto 8 cameras and have been tested with the fusion board.

    If you have 3 cameras on CSI0 and 2 on CSI1, then this requires slight changes in the example, because example assumes upto 4 camera on one CSI port and rest in other CSI port. 

    Also from the above error, it seems to be failing while allocating memory. What is the resolution of the image that you are trying to capture? We have a good amount of buffer memory, so it should not fail. But somehow memory allocation is failing. Can you first try lets 3 cameras and then 4 and see where it fails?

    Regards,

    Brijesh

  • 0 in reply to Brijesh Jadav
    Prodigy 60 points

    HI Brijesh, 

        The resolution of the image I try to capture is 3840*1920(NV12). I try to  use 4 cameras which 2 cameras on CSI0 and 2 cameras on CSI1, and the 4 cameras is running sucessful.  But the memory allocation fails when I use 5 cameras.

        I see what you mentioned above this requires slight changes in the example. I think example assumes upto 4 camera on one CSI port and rest in other CSI port,  whether I should to revised the capture node or CSI drv? Could you please tell me in detail which part should be revised.

    Thanks,

    rchen.

        

  • 0 in reply to chen rui
    Prodigy 60 points

    Correct it, the resolution of the image is 3840*2160

  • +1 in reply to chen rui
    TI__Guru**** 476595 points

    Hi Chen,

    3840x2160 is quite a big resolution. I guess you are capturing raw data, so single frame would require around 16.5MB and if you have storing 4 frames at the output of capture, it would take 66.3MB and 5 camera would require 332MB. This is just for capture. VISS, LDC/MSC would require additional buffer. This is a possibility that it might run out of 512MB buffer space that is reserved in vision apps. 

    Few options,

    - You could try keeping the number of buffers to 3 for all outputs..

    -  You could try removing unused nodes, like LDC or MSC..

    With the above changes, you should be able to run the graph with 5 camera. But if it does not, you would probably have to increase buffer space.. 

    Regards,

    Brijesh