Re: [V4L] BTTV2

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]



Jon Pennington wrote:
> 
> On Fri, 27 October 2000, Justin Schoeman wrote:
> 
> >
> > Hi everybody,
> >
> > This is just to let you know that the bttv2 driver is now basically
> > finished.  There are still a few issues, but they are fairly minor.  The
> > driver has been fairly heavily tested, and was found to be very stable.
> 
> I have a nagging question that nobody seems to be able to give a straight answer to (everyone wants me to have a PHD in mathematics and come up with my own answer).  Let's assume for a moment that I have a very fast CPU, or a pair of very fast CPUs.  Let's also assume that I have infinite hard disk bandwidth to capture to.  What is the largest frame (in pixels) that I can grab with a bt878 at +29.97fps?  Do I have to go with 1/4 size, or suffer at 8-bit color?
> 
> I want to do distribution-quality video capture with my bt878 or Zoran 36120 (which I can't get to work at the moment), and then put that on a VCD format media.  Am I asking too much to get NTSC-spec MPEG1 from this/these hardware card/s?

The Zoran should be easy?  Although I seem to rmember it only being
capable of compressed capture.

Anyway, I can't guarantee this, but my latest measurements show that it
should be possible to record at full rate and full size from a Bt8x8
based card.  The problem is that you will probably start getting line
drops.  I think full rate NTSC was around 18MB/s.  DMA'ing this to
system memory is no great problem.  HOWEVER:  When you start to write to
HDD at the same time, you push up your PCI bus requirements to
2*18=36MB's  which is still well within spec, but due to arbitration
latency of the PCI bus, you could have problems...

So, you will just have to try and see!  Make sure your harddrives can
sustain > 25MB/s (a little overhead for seeks, etc.), install the rawio
patches for the kernel, write a capture driver that can use rawio to
capture to a raw partition on the HDD without too much kernel overhead,
get a motherboard with really low latency PCI arbitration, and hope...

Sorry, in the end I haven't really answered your question.  In theory it
is definitely possible,  but I have a horrible suspision that the
arbitration latency on the PCI bus is going to keep theory from
becomming practice...

-justin





[Index of Archives]     [Linux DVB]     [Video Disk Recorder]     [Asterisk]     [Photo]     [DCCP]     [Netdev]     [Xorg]     [Util Linux NG]     [Xfree86]     [Free Photo Albums]     [Fedora Users]     [Fedora Women]     [ALSA Users]     [ALSA Devel]     [Linux USB]

Powered by Linux