1080i

[misterflibble]

I hadn't realized that the 8xxx series worked well in regards to syncing 1080i content; I thought everyone was just referring to the various split screen bugs that have cropped up where you can't get any usable 1080i modelines. If that's the case, then you may very well be right about there being a problem with the refresh rate reporting being a source of problems, although I think there are still other issues with mythfrontend and syncing to the wrong field in addition to this one of losing sync.

[tld]

Yup...apparently 1080i worked flawlessly in the older drivers, and some (who can) are still using them. Obviously the old drivers don't support newer chipsets (like my GeForce 7100GS), so many can't use them. I actually toyed with getting a card with an older chipset for that purpose, but my mythtv frontend has only PCI-e, and no such card exists.

Tom

[asdere]

Just tried the new 173.14.05 driver. The 10-15 secs interlaced drift/jitter/tearing bug is still there.

[asdere]

177.70 update:

Interlacing output "drift" every ~15 secs. nvidia-settings -q RefreshRate still reports 60.05, regardless of what the actual refresh is as reported in Xorg.0.log.

Some fuzzy math, maybe someone more technical can break it down. Actual TV refresh rate from TV and Xorg(and let's assume this is what the hardware is pushing): 60.10 Hz. Nvidia software driver hardcoded refreshrate at 60.05.

60.10 - 60.05 = .05 difference

60secs / .05 difference = 12 secs

I may be way off base, but this makes sense to me.

[tld]

Interesting. By the way...I never get any sign in my Xorg.0.log as to the actual refresh rate that X is using, and for the life of me I can't find any utility that reports that. Any idea how I could get that?

In another one of your posts:

http://www.nvnews.net/vbulletin/show...ighlight=60.05

...you had indicated that the real refresh rate was 59.9. If that were the case, this might make even more sense. The actual technical video aspects of this are way way out of my league, but here's something that makes sense to me:

60.05 - 59.9 = .06 = the difference in cycles/second.

Converting Hz to period...that is the number of seconds for one cycle...is the inverse: 1/Hz. For example, 60Hz = 1/60 or .016666 seconds per cycle. I believe that, if you do that with the above difference:

1/.06 = 16.66667

That "period", I believe, would be the amount of time it would take for that difference to wrap around and effectively line up again....even closer to what we're seeing...something like the very slow beat frequency you hear when two musical notes are just slightly different.

Maybe I'm totally off base there as well...

Tom