Cool, also check the archives since I've mostly covered this actually. But the 2 frame method seems really interesting, I couldn't get 3+ ever to work well at all and maybe skipped over too, so very very promising :-).
I'm glad someone is understanding and playing around with this too, seems
to be something we need to look at closer, also newer firmwares may act
different, they force frame based DMA and change it alot, so I'm still not
sure if there's a *perfect* combination of firmware and DMA settings possibly that we could discover. The bigger chunks of DMA mode was tried
and worked pretty well for a long time, but as of recent the frame based
mode seemed to be working better mainly for the buffer sizes allowed to
be smaller. So the buffer size may influence this, seems 16k ones
work well now, 128k ones or the size of the bulk DMA mode, something else
to explore too :-).
Oh, just so you know (since I haven't posted system stats):
AMD Athlon XP 2400+, 256MB
ASUS A7V8X-X (with VIA KT400/VT8235 chipset), using onboard sound, network
Riva TNT2-based video card
2x 160GB UDMA133 hard drives, raid0 on the MythTV recordings and swap partitions
WinTV PVR-250 with iTVC16 encoder, tuner type 50
Using ivtv 0.2.0-rc3a and 02040011 firmware
Well, a bit of additional info - I tried 3 and 4 frames, but it seemed not really to make much of a difference in the block sizes the card wanted to transfer versus 2 frames. What ended up happening is that eventually the card would decide to do a really small DMA (like 4k or so), and that would lock things up.
On the other hand, I'm not sure now exactly how relevant that is anyway. I tried out the 128k DMA blocks some more, and was able to get the system to lock up like that after a few minutes using MythTV.
Is there any hardware documentation I should be reading? I have some questions, things that might be best answered by documentation. I'm also wondering whether Hauppauge has made any changelogs for their firmware available - maybe there's some older version that pushes DMA a little less. I should probably find out more about VIA's crappy chipset, too :)
I'm also going to try out some other ideas, like unloading the network driver before testing, seeing how long I can get cat /dev/video0 > test.mpg to work (which isn't as rough as recording and watching a ringbuffer simultaneously), bumping down UDMA on the hard drives to mode 4 or 3 (66 or 33), etc. From what I understand, using PIO on the hard drives will make it work, but then you are sorely lacking in the necessary bandwidth to do simultaneous read/write (such as with MythTV Live TV).
Barry
--
==========--------- - - - - - - - - - - - - - -
Barry Drennan [EMAIL PROTECTED]
Dynamics of Adaptive Behavior Research Group
Case Western Reserve University
http://vorlon.cwru.edu/~beer/group.html
- - - - - - - - - - - - - - ---------==========-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now. http://productguide.itmanagersjournal.com/
_______________________________________________
ivtv-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/ivtv-devel
