Using DMABUF to obtain zero copy with omxh264enc

classic Classic list List threaded Threaded
2 messages Options
JK
Reply | Threaded
Open this post in threaded view
|

Using DMABUF to obtain zero copy with omxh264enc

JK
Hi

I  need to design a pipeline to be run on the Zynq Ultrascale+ MPSoc
platform which basically has a camera src connected via SDI input.

I will be using the appsrc such that I fetch the frames from SDI path and
push to the appsrc queue, the incoming frame is YUV420/YUV422 10bit/8bit,
(NV12/NV12_10LE32/NV16/NV16_10LE32).

Is to possible to use the dmabuf feature with reference to the v4l2src
plugin which exports the dmabuf fd and wraps in the GstMemory passing to
omxh264enc which then import this fd.

I Observed that the current omxh264enc does a copy of incoming frame data to
the omx layer allocated buffer, we need 4K @ 60fps, so to optimize the
pipeline we wish to implement the zero copy pipeline.

Thanks




--
Sent from: http://gstreamer-devel.966125.n4.nabble.com/
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
Reply | Threaded
Open this post in threaded view
|

Re: Using DMABUF to obtain zero copy with omxh264enc

Guillaume Desmottes-2
Hi,

appsrc does not support dmabuf but you should be able to capture from SDI
using v4l2src directly and so don't need to use appsrc here.

     G.

On 30/06/2020 14:37, JK wrote:

> Hi
>
> I  need to design a pipeline to be run on the Zynq Ultrascale+ MPSoc
> platform which basically has a camera src connected via SDI input.
>
> I will be using the appsrc such that I fetch the frames from SDI path and
> push to the appsrc queue, the incoming frame is YUV420/YUV422 10bit/8bit,
> (NV12/NV12_10LE32/NV16/NV16_10LE32).
>
> Is to possible to use the dmabuf feature with reference to the v4l2src
> plugin which exports the dmabuf fd and wraps in the GstMemory passing to
> omxh264enc which then import this fd.
>
> I Observed that the current omxh264enc does a copy of incoming frame data to
> the omx layer allocated buffer, we need 4K @ 60fps, so to optimize the
> pipeline we wish to implement the zero copy pipeline.
>
> Thanks
>
>
>
>
> --
> Sent from: http://gstreamer-devel.966125.n4.nabble.com/
> _______________________________________________
> gstreamer-devel mailing list
> [hidden email]
> https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel