The software decoders don't use multiple buffers because there is no benefit since the decoders run synchronously.
For hardware decoders, you specify the number of buffers during the OMX negotiation. We have used anywhere from 4 to 12 buffers depending on the specific codec (AVC usually needs more) and the hardware capabilities. On Jun 9, 11:07 am, scs sek <scs...@gmail.com> wrote: > Hi All & Packet Video Engineer, > > This is regarding the Opencore data path. > > I can see each node (source/intermediate/sink node has numbers of queues and > buffers to do some level of buffering. But when i see PV logging, I found > one buffer is given for decoder and the same is pushed to renderer also. > This looks like single buffer passing each node and makes data flow more > synchronous. This case, there is no actual benefit of multiple buffers and > buffering in node level. > > Is this because of non tunneling operation? Any suggestion to overcome this? > > Thanks & Regards, > SCSSek -- unsubscribe: android-porting+unsubscr...@googlegroups.com website: http://groups.google.com/group/android-porting
