Hi Hank,
I've got problems performing capture with a couple of computers. Can
you give me a brief description of how you compile in chipset-specific
support? Are there obvious steps in 'make config', or obvious error
messages in the kernel make output to work with?
Thanks,
BJ
Hank Barta wrote:
On Sun, 2003-11-30 at 23:17, suriya mohan wrote:
Hi steven,
If at all any frame dropping occurs, then it is not due to the
driver we use. If you try frame capturing in desktop linux then frame
dropping will occur rather should occur. Because the processor is not
dedicated. If u try this in any embedded device like set-top box or
PVR then you will probably have good result. So I think there will not
be any problem with the patches we apply to the kernel source.
Any comments from others in the group regarding this ?
Is there any indication why the capture is dropping frames? I'm a rank
newbie at this, having just tried DV capture for the first time
yesterday. I had problems with 'dvgrab' keeping up until I discovered
that I had neglected to compile in specific support for the motherboard
chipset. Once that was done, I was able to capture without dropping
frames until my hard drive was full. But that's a different problem. In
any case, the warnings produced by 'dvgrab' clearly indicated that
writing to the hard drive was not keeping up. Also compiling the correct
chipset drivers allow disk I/O to use DMA and this reduced CPU load from
about 50% to about 5% as reported by 'top.'
For reference this is a Sony Vaio laptop with 750 MHz P-III, 256MB RAM
and 20 GB hard drive. The kernel is 2.4.22 custom compiled but no
patches applied save a one line fix to scanner.c (USB scanner bug.)
