Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Spanker]

Hi,

very consistently my MC hangs when scrolling fast through song titles in Theaterview-Audio(and Video).
It happened with MC20 and still happens with current MC 21.0.23
Tapping Esc twice breaks up Theaterview into Standardview, when returning to Theaterview the problem is solved.
Until the next fast scrolling.
Problems exists when using keyboard and IR remotecontrol.

I can reproduce it with video files- Theaterview-Video, too.
Audio files are on a SSD, video files are HDD.

I checked all the common problems and found nothing comparable on the forum.

Attached the log file.

I hope someone can help, thnx in advance.

Henk

[JimH]

Problems like this are usually related to video drivers.

Theater View uses 3D drawing.

[mattkhan]

Same here using latest nvidia drivers (gtx660 card), has been the case for a long time also.

It does seem there are regular reports of issues with theatre view that are filed under " driver issue" yet same driver appears to work fine on other things. This issue and the device lost one are the two obvious ones that spring to mind.

At what point do they become a theatre view issue?

[Spanker]

Hi Jim,

I use a Geforce GTX750 Ti, should I try a different (older) driver?
Or any particular setting?
The Nvidia driver has a whole list of settings concerning 3D.

thnx

[JimH]

I can't help with what to try.  Sorry.

[JimH]

It does seem there are regular reports of issues with theatre view that are filed under " driver issue" yet same driver appears to work fine on other things.

Theater View uses 3D drawing.  And bugs can be triggered by unique sets of circumstances.

I should start a thread on problems that turned out to be caused by video drivers.  There are a few in this thread:
http://yabb.jriver.com/interact/index.php?topic=24031.0

[mojave]

Here is another recent thread on the same issue:

Crashes when scrolling lists in Theatre View

Build 21.0.9 (9/19/2015) tried to fix the problem: 
9. Changed: Tweaked flushing and presenting in 3D rendering for Theater View to avoid it freezing, testing appreciated.

I had some freezing in the past, but haven't had any in the past few weeks. I'm using Geforce Driver 361.43 on two computers with Nvidia GTX 970 graphics cards.

[Spanker]

This is probably what you are referring to, Jim:

http://yabb.jriver.com/interact/index.php?topic=24031.msg697684#msg697684

[Spanker]

Thnx Mojave, I'll give that driver a try.

[JimH]

This is probably what you are referring to, Jim:

http://yabb.jriver.com/interact/index.php?topic=24031.msg697684#msg697684

And this:
http://yabb.jriver.com/interact/index.php?topic=24031.msg704783#msg704783

But there are many more.

[mattkhan]

Here is another recent thread on the same issue:

Crashes when scrolling lists in Theatre View

Build 21.0.9 (9/19/2015) tried to fix the problem: 
9. Changed: Tweaked flushing and presenting in 3D rendering for Theater View to avoid it freezing, testing appreciated.

I had some freezing in the past, but haven't had any in the past few weeks. I'm using Geforce Driver 361.43 on two computers with Nvidia GTX 970 graphics cards.

I'm using 361.43 as well and it's still an issue here

I wonder whether it's a question of how theatre view handles (input) events? Obviously I don't have any understanding of the internals of theatre view but, speaking as a developer, this is a plausible explanation (where plausible in this situation means "logical but easily utterly wrong")

I use a remote usb driven interface (either some RF keyboard to a dongle attached via usb or an RTi remote talking to the MCE remote dongle), holding a button down on either will translate a long or rapid repeat press into a slower event stream. Lets say this event stream arrives at jrmc at 200ms intervals and lets say that theatre view applies a 150ms sized buffer to that event queue (to avoid redraw churn given that 100-150ms is the lower limit of human perception time to a change in the UI) and lets further say that theatre view is a bit slow when it gets moving (perhaps it has cache misses on thumbnails or the background is reloaded repeatedly).

One could imagine, in this situation, the UI freezing *if* the renderer is coupled (deliberately or via contention over some shared state) to the event reactor in some way.

[mattkhan]

Actually I tested this theory a bit more and it does seem to be a remote driven lockup. Using a remote, I can make theatre view stop responding to commands but I can Ctrl+1 back to the main view. I wonder what control mechanism other people (who experience lockups) use?

[mojave]

I was having the issue with my keyboards. One is a USB wired Microsoft keyboard and one is a wireless Logitech keyboard.

[Hendrik]

Actually I tested this theory a bit more and it does seem to be a remote driven lockup. Using a remote, I can make theatre view stop responding to commands but I can Ctrl+1 back to the main view. I wonder what control mechanism other people (who experience lockups) use?

Thats not necessarily related to what you think it is. Switching out of Theater View destroys the 3D device, which just happens to unblock the rendering.

The way it freezes, it appears to be a NVIDIA driver bug. In fact people have reported that using older drives fixed it for them, since the problem started somewhat recently without any changes in MC.
On driver versions, other users reported that 347.88 worked fine, while 350.12 started showing the issue.

[mattkhan]

Thats not necessarily related to what you think it is. Switching out of Theater View destroys the 3D device, which just happens to unblock the rendering.

The way it freezes, it appears to be a NVIDIA driver bug. In fact people have reported that using older drives fixed it for them, since the problem started somewhat recently without any changes in MC.
On driver versions, other users reported that 347.88 worked fine, while 350.12 started showing the issue.

My point there was just to state that jriver itself remains responsive and that it doesn't just white out as a whole. I can't tell if the issue is in delivering key press events or the renderer or whatever else might be "stuck".

What indicates it as a driver bug btw?

[JimH]

If using an older driver solves the problem, it's probably a driver issue.  Please try it.

[flac.rules]

Thats not necessarily related to what you think it is. Switching out of Theater View destroys the 3D device, which just happens to unblock the rendering.

The way it freezes, it appears to be a NVIDIA driver bug. In fact people have reported that using older drives fixed it for them, since the problem started somewhat recently without any changes in MC.
On driver versions, other users reported that 347.88 worked fine, while 350.12 started showing the issue.

Yeah, but having an old driver for the most popular 3D-card on the market isn't a very satisfying solution either. Even if it is a bug on nvidias side, si there nothing that can be done?

[mattkhan]

If using an older driver solves the problem, it's probably a driver issue.  Please try it.

Thanks (to you and hendrik) for the suggested fix, I rolled back to 347.88 and theatre view now works using both keyboard and RTi (MCE) remote.

I agree with Elvis133 though in that this isn't a sustainable solution if this is a driver "feature" rather than defect. IIRC 347.88 was the last good driver for certain madvr features so I guess there is a non zero probability you will need to fix this rather than nvidia. At least it might be worth stating somewhere that this is the last known good nvidia driver (for those who have issues with theatre view and use nvidia cards)

[JimH]

Please tell NVIDIA how you feel.

[flac.rules]

The problem is that telling nvidia how you feel is probably not going to do anything, and using old drivers is a major problem for other uses of the machine.

[mattkhan]

Please tell NVIDIA how you feel.

leaving the efficacy of that approach to one side, there isn't enough info to report a bug to them. I imagine we would need you to specify how theatre view is using the device in order to understand what is going wrong.

Practically speaking though having a known defect that affects all WHQL drivers for Windows 10 is not a good look for a consumer facing app. I would think that gamers generally expect this sort of thing, media player users probably don't.

[JimH]

You seem to insist that JRiver could fix a problem that, to me at least, is an obvious driver bug.  These kinds of problems tend to come and go.  Eventually, NVIDIA will fix it.

What you suggest is similar to proposing that JRiver fix a problem with the OS.  It isn't going to happen.

There is plenty of information to give to NVIDIA.  Just point them to this thread and tell them we'll provide licenses so they can duplicate the problem.  However, the chance of that happening isn't very great.

[mattkhan]

I'm not insisting on anything, I'm just clarifying a different point of view. I am well aware that it's always going to be entirely up to you what you do about it and that nagging doesn't get v far (so on that note I shall exit this thread stage left)

[flac.rules]

You seem to insist that JRiver could fix a problem that, to me at least, is an obvious driver bug.  These kinds of problems tend to come and go.  Eventually, NVIDIA will fix it.

What you suggest is similar to proposing that JRiver fix a problem with the OS.  It isn't going to happen.

There is plenty of information to give to NVIDIA.  Just point them to this thread and tell them we'll provide licenses so they can duplicate the problem.  However, the chance of that happening isn't very great.

Programs have to work around weird things for instance windows does all the time. I understand that it sucks to "fix someone elses problem", but Nvidia probably doesn't care about a small program like this, and other programs seems to do things in a way that doesn't cause this problem, so from the outside it seems solvable.

In the middle stands the consumer with 3 choices

1. Downgrade (with major negative cosequences for everything but MC
2. Hope that nvidia fixes it (they haven't in almost a year, and might never)
3. Hope that MC works around it.

Although it is not "fair", the best shot for the customer probably is choice 3.