[linux-dvb] Origin of datarate limitation?

[Ralph Metzler <rjkm@metzlerbros.de>]

Sven Karlsson writes:
 > I've read at numerous places that there is a datarate limitation on the
 > parts of the TS that can be read from the AV711x via the PCI bus. I also
 > understand that the Nova does not have that limitation and that the Nova
 > does not have any hardware MPEG playback capabilities although it is based
 > on the AV711x.

That's wrong. The Nova card has no AV711x.
It basically consists of a DVB-Tuner, an SAA7146 to DMA the TS into
main memory and some glue logic.


 > Could someone (possibly yet again ;-)) clarify the origins of the data rate
 > limitation on MPEG capable cards? It appears to me that this could be a
 > firmware issue. Is that so? Are there any efforts focusing on patching the
 > firmware or otherwise remove the datarate limitation at the possible expense
 > of the MPEG capabilities?

It is more a hardware issue than a firmware issue.
All data first is moved by hardware (AV711x) to the somewhat limited
(in speed and size) RAM next to the AV711x. Then it has to be moved in
software by the 40MHz ARM7TDMI core of the AV711x to the dual ported
RAM which is also connected to the DEBI port of the SAA7146.
Then it is moved by the SAA7146 via DMA into main PC memory.


 > I, for one, would gladly skip all MPEG capabilities of the board if I
 > instead could process and possibly record several channels at once. My DVB-C
 > card is sitting in my firewall/network server and I intend to send the MPEG
 > stream over my home lan to my other boxes which all of them are powerful
 > enough to decode MPEG-II streams in realtime.

It might be possible to speed up the transfer a little bit by using
the video SDRAM as a transfer buffer but no effort in this direction is
planned right now.
If you want to stream the whole TS better use a NOVA card.


Ralph




-- 
Info:
To unsubscribe send a mail to listar@linuxtv.org with "unsubscribe linux-dvb" as subject.