I’ve just stated toying with gstreamer and deepstream (ngc container 6.1) on a machine with a A100, and cannot figure what to change in the second cmd below to get it to work like the first:
cuGraphicsGLRegisterBuffer failed with error(219) gst_eglglessink_cuda_init texture = 1
A100 is not display card, it’s compute card. you can try to setup virtual display, following Deepstream/FAQ - eLinux.org 5A, but it’s for Tesla series. you can give it a try.
Thanks,
is it possible to run the deepstream examples with this setup by letting the cpu handle the encoding part? If so could you point me to where I can see the code for it?
You can refer to deepstream-app, which support hardware encoder and software encode, you can set in config. #encoder type 0=Hardware 1=Software
enc-type=1
create_encode_file_bin in sources/apps/apps-common/src/deepstream_sink_bin.c