Hamish Moffatt wrote: > On Wed, Nov 02, 2005 at 09:55:32AM -0800, jignesh@xxxxxxxxxxxxxxxxxxxxx wrote: > >>As far as I understand, the Lite has the old Brooktree chip (for the NTSC >>decoding part of it), whereas the Gold has the newer cx88 chip. I'm sure >>the cx88 driver will eventually mature to be as stable and proven as the >>bttv driver, so no worries there. bttv driver seems to have been around >>forever, and hence I can understand your preference of Lite over Gold. > > > On the contrary - the cx88 is exactly why you should prefer the Gold! > It has a decent sized FIFO for digital data transfer, while the bt878 > was not designed for it. > > DVICO's DVB cards come in Lite and regular too, with the bridge chip > being the difference again (and perhaps the software). Hamish- In theory, you are correct... But in practice: I ha[ve/d] both of these cards, and under linux, the Lite (bt878-based) model produces a much steadier stream. There is no argument that bt8xx ANALOG support under linux is much more mature that cx88 analog support, but this is besides the point. For ATSC/DVB ... cx88 SHOULD be better, but I must say that it is not, at least not in *my* experience. Someone else might pipe in and prove me wrong. (please do, if you know any different) When using high-level applications such as xine and mythtv, it is nearly impossible to tell the difference between the streams from the Gold card Vs the Lite card, however, when testing using a combination of azap + mplayer, the stream stays steady for a *much* longer time on the bt878-based card as opposed to the cx88-based card. I have had similar experiences on a saa7134-based card. I have a theory that might help to explain this... Both saa7134-dvb and cx88-dvb depend on the video-buf-dvb module, while dvb-bt8xx does not. video-buf-dvb lives in v4l-kernel cvs with saa7134-dvb and cx88-dvb, while dvb-bt8xx lives completely in dvb-kernel cvs, with the exception of the analog bttv-* code. video-buf-dvb has not been touched in months. There are probably some improvements that can and should be made to this module, but it might not be getting the attention that it deserves due to the fact of it being in v4l-kernel instead of dvb-kernel. This is just a guess -- I dont know the real reason for the unsteadiness of the ts like I have described. I can also say that most of the errors I get from mplayer are from the a52 audio codec, and if I disable it, the stream becomes steadier... but not as steady as that of the FusionHDTV5 Lite board. I intend to take a close look at video-buf-dvb one of these days, but it could very well possibly be a red herring. All I am saying is that I've had better luck with the Lite card under linux. I hereby open this discussion for people to tell me why it is that I am wrong. (I can only assume that this is what everyone will say.) Enlighten me. :-)
