On Mon, 4 Aug 2003, Trent Piepho wrote: >I take it you do the format conversion and copying into the mmapped buffer on >the sync call?
Yes. >Have you considered doing the format conversion earlier? Have Not seriously. I can't think of any advantage of that. >just one bounce buffer for the hardware, and convert the data as soon as the >hardware buffer is full and an empty mmap buffer is available? I don't think this is so efficient with just one mmap buffer. The empty mmap buffer is not available before application calls CAPTURE, because before that it's used for the previous frame. SYNC will be likely called just after CAPTURE, and if the conversion from internal buffer would be done to the mmap buffer when the internal buffer is captured, it should happen in the practically non-existing time window between CAPTURE and SYNC. mmap buffer usage: frame0 -> CAPTURE -> available -> SYNC -> frame1 Using multiple mmap buffers, yes, but where's the point? -- video4linux-list mailing list Unsubscribe mailto:[EMAIL PROTECTED] https://www.redhat.com/mailman/listinfo/video4linux-list
