Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
Melchior FRANZ wrote: Bug #3: Lee threw in an old bug and repeated the long disproved myth that listeners are the cause. They are not! This bug causes fgfs to freeze for a tenth of a second in 8 seconds intervals beginning a few minutes after startup (but not always). The intervals stay the same, but the freezing time becomes longer and longer over time. This happens now since years and on my machine it can reliably be fixed by turning off AI traffic (the c172/pa28 aircraft). I tried to track it down but didn't invest enough time. The fact that this AI part was full of bugs just didn't make it worthwhile. (I just say tower.cxx. :-) sorry to break your state of bliss on bug #3 but I have , had for some time, this bug without AI enabled. as far as I am concerned this is a long standing, semi-serious bug that remains unresolved. No, its not my hardware (popular explanation in months past). - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
* Melchior FRANZ -- 8/31/2007 9:59 AM: Listeners are not involved, unless you have very badly written ones on your harddisk (only). When you brought that up last time I added some logging capabilities for listeners and disproved this claim. See this mail for how to log listener calls: http://marc.info/?l=flightgear-develm=11601627835w=2 I apologize if my response was too harsh. But I'm touchy when it comes to (IMHO) unjustified criticism of Nasal listeners. They have become a central part of FlightGear=Nasal-core interaction, and I don't like it if there are (IMHO) unfounded doubts about their correct working. I fixed all problems with them ever reported and I'll keep doing that. So, if you run into anything where you think listeners cause problems, please report that along with logging information, example code, and command line. Are there any listeners triggered at the same time when such an interval freezing occurs? If so, which? m. - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
This thread started as bug report about a recent problem with dynamic view, but people used it as opportunity to throw in their favorite bugs, although those are apparently unrelated. Let's keep things separated: Bug #1: stutters and pauses with 1-2 sec lenght when dynamic view is enabled (observed by Heiko in a recent build by Thomas). We don't know much about this bug, not even if it's fg/osg or fg/plib and when it started. It didn't happen when I ran fgfs last time (about two weeks ago). Bug #2: Syd added a complaint about the fg/osg *feature* that causes stutters until all scenery objects in 360 degree were loaded. This is annoying, but AFAIK not a bug. Should be resolved nevertheless, and I'm sure Mathias and Tim are aware of it and will change that at some point. (Someone on IRC has a local Nasal script that makes a 360 degree lookaround at the very beginning so that this won't disrupt the flight in the following minutes.) Bug #3: Lee threw in an old bug and repeated the long disproved myth that listeners are the cause. They are not! This bug causes fgfs to freeze for a tenth of a second in 8 seconds intervals beginning a few minutes after startup (but not always). The intervals stay the same, but the freezing time becomes longer and longer over time. This happens now since years and on my machine it can reliably be fixed by turning off AI traffic (the c172/pa28 aircraft). I tried to track it down but didn't invest enough time. The fact that this AI part was full of bugs just didn't make it worthwhile. (I just say tower.cxx. :-) * leee -- 8/31/2007 3:09 AM: I couldn't establish anything consistant about it but re-writing some of the nasal I was using, to use less listeners, which were being called every frame, ameliorated the situation somewhat, iirc, but wasn't an answer. Listeners are not involved, unless you have very badly written ones on your harddisk (only). When you brought that up last time I added some logging capabilities for listeners and disproved this claim. I monitored all triggered listener code and there were none run regularly. (OK, some were, and they got fixed after that, for example those that listened to /gear/gear/wow, which is set in every frame by YASim.) Of course, if you write bad Nasal code, you can create all sorts of problems, and listeners will reliably run that buggy code. But the problem then is the bad Nasal code, and not the listeners by themselves or the number of active listeners. If you bring this up next time, please add some facts: show us the command line and the listeners that you had to comment out to fix the problem. m. - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
On Friday 31 August 2007 09:43, Melchior FRANZ wrote: * Melchior FRANZ -- 8/31/2007 9:59 AM: Listeners are not involved, unless you have very badly written ones on your harddisk (only). When you brought that up last time I added some logging capabilities for listeners and disproved this claim. See this mail for how to log listener calls: http://marc.info/?l=flightgear-develm=11601627835w=2 I apologize if my response was too harsh. But I'm touchy when it comes to (IMHO) unjustified criticism of Nasal listeners. They have become a central part of FlightGear=Nasal-core interaction, and I don't like it if there are (IMHO) unfounded doubts about their correct working. I fixed all problems with them ever reported and I'll keep doing that. So, if you run into anything where you think listeners cause problems, please report that along with logging information, example code, and command line. Are there any listeners triggered at the same time when such an interval freezing occurs? If so, which? m. Hi Melchior, It wasn't my intention to criticise listeners in any way - I was just reporting what was happing, what I tried doing about it and what the results were after I tried a couple of different things, including re-writing some of the nasal to use less listeners. There were two main areas where I was using listeners that were being called every frame and this was intentional because it seemed appropriate to me. First of all, I needed to intercept stick control inputs so that I could use the stick in different 'modes' i.e. use the stick input to set pitch angles, roll angles or climb rates and alternatively provide proportional surface deflections instead of direct control surface deflection. To do this it also meant that I had to monitor the aircraft orientation, accelerations and speed. Like I say, using listeners for these functions seemed appropriate because they would/could be constantly changing. Tied in with this, I also had to use a number of A/P filters to provide properties that I could actually set some of the listeners against, as some of these properties wouldn't work directly with listeners. All this meant that the listeners would be called each frame but that was exactly what I thought I needed. The re-writes I did were to use timers for many of these functions instead of the listeners and while this worked it was less than ideal because it set a limit to the resolution of the things I needed to monitor. The result was that instead of spotting a deviation as soon as it occurred it wouldn't be spotted until the next sample, which may have been as long as the sample period and this meant that the deviation would have grown and been greater than it needed to be before it could be corrected. This in turn required a greater correction than it would have done if the deviation could have been corrected when it was smaller. Perhaps I had taken the wrong approach for this or perhaps I was just trying to do too much with insufficient hardware, but anyway, that's what I was trying to do and for the most part it seemed to be working, at least in principle. In the end though, what with some of the A/P PID controllers refusing to start without first resetting the A/P and then finding that the PID controller sampling rates were not working at the Ts sample rate but were tied to the frame rate, which made it impossible to tune them for consistant operation, I gave it up. LeeE - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
Bug #2: Syd added a complaint about the fg/osg *feature* that causes stutters until all scenery objects in 360 degree were loaded. This is annoying, but AFAIK not a bug. Should be resolved nevertheless, and I'm sure Mathias and Tim are aware of it and will change that at some point. (Someone on IRC has a local Nasal script that makes a 360 degree lookaround at the very beginning so that this won't disrupt the flight in the following minutes.) My apologies , from the description I thought it was the scenery loading stutter.I never encountered the dynamic view problem... -- SydSandy [EMAIL PROTECTED] - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
Robert Black Sent: 29 August 2007 03:41 To: FlightGear developers discussions Subject: Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view On Tuesday 28 August 2007 18:44, Laurence Vanek wrote: also had this problem in general (not with dynamic view). only thing that helped was adding the following to my .fgfsrc file: --prop:/sim/frame-rate-throttle-hz=75 as far I know no one has solved this problem. apparently only a few of us seem to have the issue. I thought everyone had this behavior. Most videos I have seen of FG have the stutter and pause. Robert I certainly do. I suspect that people have given up complaining about it and have come to regard it as normal. At its worst it makes FG almost impossible to use. Vivian - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
Quoting Vivian Meazza : Robert Black Sent: 29 August 2007 03:41 To: FlightGear developers discussions Subject: Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view On Tuesday 28 August 2007 18:44, Laurence Vanek wrote: also had this problem in general (not with dynamic view). only thing that helped was adding the following to my .fgfsrc file: --prop:/sim/frame-rate-throttle-hz=75 as far I know no one has solved this problem. apparently only a few of us seem to have the issue. I thought everyone had this behavior. Most videos I have seen of FG have the stutter and pause. Robert I certainly do. I suspect that people have given up complaining about it and have come to regard it as normal. At its worst it makes FG almost impossible to use. I think stutter comes from the threaded scenery tile loader. When you change view direction, you ask the loader to load more tiles, and when all required tiles are loaded for a given position, the stutter stops. Also, when there is an insane frame rate, there should be no CPU time allocated to the tile loader, and thus, the stutter, so limiting the frame rate, either by using the throttling feature, or activating vsync, should lessen the problem. -Fred -- Frédéric Bouvier http://frfoto.free.fr Photo gallery - album photo http://www.fotolia.fr/p/2278/partner/2278 Other photo gallery http://fgsd.sourceforge.net/ FlightGear Scenery Designer - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
Frederic Bouvier wrote: I think stutter comes from the threaded scenery tile loader. When you change view direction, you ask the loader to load more tiles, and when all required tiles are loaded for a given position, the stutter stops. That seems unlikely. The tile loader is very old code, and this is a new symptom. It's also, as you point out, threaded precicely for the purpose of *not* blocking the main rendering thread. Andy - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Frederic Bouvier wrote: Quoting Vivian Meazza : Robert Black Sent: 29 August 2007 03:41 To: FlightGear developers discussions Subject: Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view On Tuesday 28 August 2007 18:44, Laurence Vanek wrote: also had this problem in general (not with dynamic view). only thing that helped was adding the following to my .fgfsrc file: --prop:/sim/frame-rate-throttle-hz=75 I think stutter comes from the threaded scenery tile loader. When you change view direction, you ask the loader to load more tiles, and when all required tiles are loaded for a given position, the stutter stops. Also, when there is an insane frame rate, there should be no CPU time allocated to the tile loader, and thus, the stutter, so limiting the frame rate, either by using the throttling feature, or activating vsync, should lessen the problem. -Fred You can blame some stuttering on the incomplete integration of the tile loader with OSG, something that I'm working on. In particular, display lists are now compiled and textures are now loaded when a tile first becomes visible; the OSG database pager schedules this work over several frames to avoid affecting the frame rate. Also, it's true that activating vsync will leave some predictable time for the pager thread to run when it won't interrupt the main rendering thread. Tim -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org iD8DBQFG1YtreDhWHdXrDRURArtXAJ9iZaxcutfB33/geBAr8jjXn+7AygCfZq4G nY9J867TOdCadCHKVMCrZH8= =KsQG -END PGP SIGNATURE- - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
Quoting Tim Moore [EMAIL PROTECTED]:
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Frederic Bouvier wrote:
Quoting Vivian Meazza :
Robert Black
Sent: 29 August 2007 03:41
To: FlightGear developers discussions
Subject: Re: [Flightgear-devel] [Bug-Report] Stutterer and
pauses withdynamic-view
On Tuesday 28 August 2007 18:44, Laurence Vanek wrote:
also had this problem in general (not with dynamic view).
only thing
that helped was adding the following to my .fgfsrc file:
--prop:/sim/frame-rate-throttle-hz=75
I think stutter comes from the threaded scenery tile loader. When you
change
view direction, you ask the loader to load more tiles, and when all
required
tiles are loaded for a given position, the stutter stops. Also, when there
is an
insane frame rate, there should be no CPU time allocated to the tile
loader, and
thus, the stutter, so limiting the frame rate, either by using the
throttling
feature, or activating vsync, should lessen the problem.
-Fred
You can blame some stuttering on the incomplete integration of the tile
loader with
OSG, something that I'm working on. In particular, display lists are now
compiled and
textures are now loaded when a tile first becomes visible; the OSG database
pager
schedules this work over several frames to avoid affecting the frame rate.
Also, it's
true that activating vsync will leave some predictable time for the pager
thread to run
when it won't interrupt the main rendering thread.
Did someone experiment changing the number of loader threads ?
This is possible by changing line 120 of Scenery/FGTileLoader.hxx
In CVS, the line is now :
enum { MAX_THREADS = 1 };
It won't cure the fact that OGL resources needs to be allocated in the rendering
thread, but would enable multiple cores to load more than one tile at a time.
-Fred
--
Frédéric Bouvier
http://frfoto.free.fr Photo gallery - album photo
http://www.fotolia.fr/p/2278/partner/2278 Other photo gallery
http://fgsd.sourceforge.net/ FlightGear Scenery Designer
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now http://get.splunk.com/
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
On Wednesday 29 August 2007 15:43, Andy Ross wrote: Frederic Bouvier wrote: I think stutter comes from the threaded scenery tile loader. When you change view direction, you ask the loader to load more tiles, and when all required tiles are loaded for a given position, the stutter stops. That seems unlikely. The tile loader is very old code, and this is a new symptom. It's also, as you point out, threaded precicely for the purpose of *not* blocking the main rendering thread. Andy I'm not so sure I'd regard this as a new problem. I first reported it on the 8th Dec 2006 and Melchior responded that he'd been experiencing it for some time before then. At that time, Melchior thought it was related to the AI sub-system. The thread subject was Periodic 'stutter' in FG LeeE - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] [Bug-Report] Stutterer and pauses withdynamic-view
On Wednesday 29 August 2007 17:53, leee wrote: I'm not so sure I'd regard this as a new problem. I first reported it on the 8th Dec 2006 and Melchior responded that he'd been experiencing it for some time before then. At that time, Melchior thought it was related to the AI sub-system. The thread subject was Periodic 'stutter' in FG Speaking of which; there is one part of the AI system that causes some (currently unavoidable) stutter. Models are loaded on demand, and this causes some stutter. Mathias has indicated trying to move the loading process out of the main render thread, or something like that. I'm not sure how much that relates to the issues Tim mentioned in this thread. In addition, there are a few other candidates for causing stutter in the AI system, in particular the route finding algorithm and the traffic separation detection code, because these are requiring peaks in CPU activity. In particular when two aircraft request these functions in succession, there can be peaks of activity. However, having said that, these AI related issues are quite unlikely to cause the observed stutter, for two reasons 1). Heiko's original report in this thread explicitly mentions that he has turned off traffic manager and AI, and 2) I'm also seeing these stutters occur even before the AI traffic system starts loading. Cheers, Durk - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
