>  > On 1.5.2 (customized config but standard sources), when I boot my DVD
>  > player is recognized as a UDMA 2 device.
>  >
>  > When I try to play a DVD movie, it degrades through UDMA 1, DMA 2, to PIO
>  > 4, claiming lost interrupts:
>
> [...]
>
>  > My questions are:
>  >
>  >  * Is this at all normal or expected?
>
> I dont't know this particular Maxtor IDE controller, but it seems it has
> a Promise chipset. Some of these IDE controller chipsets which can be

Yes, I think so.

I tried swapping cables around but didn't improve the situation.  I know
that at least at one point I had an improperly seated cable, though.  But
I also know that I have yet to improve on the situation.


> found on add-on PCI boards are not really ATAPI controllers but instead
> just feature the command set neccessary to control hard disk drives.
> Thus, it may be that you simply can't use this controller with your DVD
> drive (although even if it is no ATAPI controller many CD-ROM drives do

Err, I *can*.  It just doesn't work in any of its DMA modes.  (^&


>  >  * Is DMA or UDMA important for DVD playback?  (Even on my PCI video
> card,
>
> It is. In fact, when watching DVD movies on systems without DMA you
> often experience a "stop" for maybe half a second or so everytime the

I don't have any half-second delays.  Though there is visible uneaveness,
I've attributed this to the frame drops.  (Which, in turn, seem to be tied
to the bandwidth to the PCI card...the only quantitative difference in
playing to 16bpp vs. 24bpp is the amount of data going out per frame
(2/3rds as much) and the frame rate (*very* close to 3/2 as much).

Still, I'll try a little bit more fiddling.

But, I guess chalk up one more previously unwritten caveat to DVD's.
(Well, someone mentioned something about DMA before, but I had no idea
that a drive/card combination that support DMA won't support DMA in
combination...(^&)  Oh well...is there any way that the kernel could
determine this kind of information in advance (or even diagnose the
chipset's capabilities and tell you that ATAPI won't work)?


> DVD is accessed (which usually is quite often). On DMA-enabled systems
> you don't face this problem. (Well, at least not due to the lack of DMA
> transfers :-)
>
>  >    I get 18fps in 16bpp, or 12fps in 24bpp, according to Ogle.  I
> probably
>  >    won't be doing much else with my computer while watching a movie,
>  >    so as long as I can get full frame rates with an eventual AGP card,
>  >    this isn't a huge matter.)
>
> Equipped with *any* modern AGP card, your system should be more than
> able to playback DVD movies without frame drops.

That's the thing to which I've been ascribing the performance problems up
till now.


Thanks for the helpful information, though.  I hope that I can get the
drive to talk to my motherboard's ATAPI, if DMA is really important.  If
it is a significant contributor to my fps problems, I might get over 20fps
on 16bpp with my current hardware, which would be nice.  (I've decided
that I'll continue living with my present video card until DRI appears for
NetBSD or I give up and set up a dual-boot machine again, and DVD playback
will just have to live with the PCI card.  (^&)


  ``I probably don't know what I'm talking about.'' --rauch@math.rice.edu