how to run 12 gmsl camrea with DriveSDK? (Solved)

Hi,guys:
i found a sample program for 12 gmsl camera,/usr/local/driveworks/bin, but those are based on some wared “dw” SDK ,i want use the Nvidia Drive works SDK: Drive/5050bL_SDK/DriveSDK/drive-t186ref-linux/samples,but i only can find quad camera program in one group,someone can help me? Nvidia engineer guys.

wangxuhui.

Dear wangxuhui,
We have Driveworks API which provide APIs to communicates with different seonsors. We already have a Driveworks sample for this purpose(sample_camera_multiple_gmsl). Can you please check that sample

hi,SivaRamaKrishna
thanks for your reply;
right now ,i use a custom camera,is the Driveworks API support custom Camera?,i already try the source named img_cap to run my custom camera and it worked when i use only one group,the problem is when i use two threads to run two groups camera,one group can get the data from interface NvMediaICPGetFrameEx,but another one will return time out from NvMediaICPGetFrameEx.can you help me on this?
wangxuhui

Dear wangxuhui,
Could you please share the camera details. Did you check NVMEDIA_IMAGE_CAPTURE_TIMEOUT_INFINITE for timeout parameter in NvMediaICPGetFrameEx

hi,SivaRamaKrishna
I just tried the NVMEDIA_IMAGE_CAPTURE_TIMEOUT_INFINITE.Just TIMEOUT disappered,still only on group working.My camera is ov2718,I think it is kind of same as ae0231.well ,thank for your help

Dear wangxuhui,

Each camera group AB/CD/EF support 4 cameras each.
Each instance of nvmimg_cap supports capture of 4 cameras (one group), and displays it on a display.
So to launch 12 cameras simultaneously, 3 instances of nvmimg_cap should be launched, and displayed on 3 available displays. Thanks.

Sorry SteveNV & SivaRamaKrishna,i solved the problem last week.It’s my fault camera doesn’t work.I was running 8 cameras use two threads to manage two groups of the cameras.Due to i forgot modify the gloable variable to be the local variable of the thread,the program can not output the data.Thanks for you guys.

Dear wangxuhui,

Thank you for you update.