[ExtractStream] Re: Status of Windows tools ==> Linux Solutions

[Dale Reed]

Kyle Davenport wrote:

> 

> > > 4. ConvertStream doesn't work for me (I'm at TiVo 2.0.1).
> >
> > Converstream should work fine with 2.0.1. I re-worked some of the
> > code and pulled some other code from Extractstream. If you'd like
> > a copy that works with 2.0, let me know.
> 
> I thought I downloaded the latest from the author's web site - please do
> send.


Well, there is the official latest on the website, and lots of patches
and updates floating around. I ported it over to Win32 and updated some
of the code based on the CVS ExtractStream (among others). AFAIK, that
code has not been released. You can get the new win32 executable from
http://www.ntworld.com/tivo.


> > > 5. Finally! got nfs mounting to work. A little documentation would
> > have
> > > been nice. It's for mounting remote server's filesystems on the
> > TiVo.
> >
> > nfs is REALLY slow. I couldn't get over 300K/sec with it. I still
> > use
> > the single netcat connection and pull almost 950K/sec with netcat.
> > The nice thing about nfs is that you don't have to set any listeners
> > up on the PC side, though. :)
> 
> hmmm, I'm getting only 230K/sec. Why won't the netcat thing work for
> me?


Most people have speed issues with full duplex. You have to either put
your nic in a PC and manually set it to half duplex, use a hub (rather than
a switch) or have a switch that you can configure half duplex on the port.
I used a HUB and now my switch that I can configure the port on.


> Check it out: http://www.sigmadesigns.com/products/netstream2000.htm


Ahh. The dxr3 is based on the Hollywood card. Its a pretty good card
as well.


> After pulling a large stream from the TiVo, I too noticed the audio was
> behind the video about 1/2 second by the end. It seems mplex was
> complaining about the frame id's - could ExtractStream be inserting
> incompatible/non-standard frame id's?


Well, there is a common assumption that the audio and video streams
after being split are exact. When in reality, I don't think they are.
IMHO, the way they stay synced on the tivo is because thay are
intertwined into one stream. The tivo reads the single stream, sends
the audio to the audio hardware and the video to the video hardware as
its reading the respective parts of the stream. Even if there is bad
data in the audio or video, it doesn't matter, as the output stays
synced no matter what.

The project that I'm contributing to takes this theory and rather
than split the streams and then try to independently put then back
together, it puts the mpg together just like the tivo does (on the
fly audio/video staying synced).

Dale