Hi,

sorry, this question may have been discussed already, but as far as I found under a different understanding. Particularly our mate martinlest2 has posted back in 2008, but somehow the original text seems to have been disappeared and generally these old threads are hard to read due to strange formatting.
http://forums1.avsim.net/topic/235923-dealing-with-out-of-memory-errors/?hl=martinlest2

 

So this is my Problem. I have read many times that the Scenery.cfg has a limitation of thousand (=999) entries. Recently I had OOMs in my sim at a flight Tbilisi-Dubai without finding a cause, I only assumed that AI traffic could have been the culprit. It was the first time I had OOMs, though I never applied the 3GByte patch as I have never had had OOMs before. I had particularly done this flight already in spring without any problems. 

 

I then found a test scenario (placing my a/c at OMDB (FlyTampa), changing to EDDF (Aerosoft), changing back to OMDB) which caused this OOM very reliably. When I disabled my heavy AI traffic, OOM didn't occur so I still believed it was a AI traffic problem.

 

With AI traffic active again, I then deinstalled some sceneries which had no geographical relation to the OOM-plagued sceneries. These which I removed were mostly older sceneries which I either never visited or which weren't up to todays standard anymore. The important thing is, my sim worked proper for years with these sceneries. I "merged" other sceneries into one scenery area; I did so especially with meshes. So I brought my Scenery.cfg from 713 down to 684 entries. Guess what happened: Now the OOM was gone even with full traffic at least with my test scenario; I repeated the test several times always having no OOM (a dedicated long haul test flight will follow).

 

So my question is if there are experiences with too many entries in Scenery.cfg causing FS2004 to behave heavy and causing OOMs. Is there a certain limit below the known limitation of 999 entries, depending on PC specs and OS? Any thoughts were helpful. Perhaps martinlest2 can step in and recapitulate his thoughts or add new experiences.

 

Note: I wrote some words about this problem in my recent thread "OOM and CTD at Dubai" but it seems this question should be asked in a dedicated thread because the other thread concentrates on fault finding regarding to AI traffic or a particular scenery.

 

Thanks in advance

and best regards

Harald

Harald!

 

Just a short input

 

Interesting approach, but imho I can´t imagine that the number of entries is of any relevance with dedicated OOM/CTD occurence.

Another idea would be a "bug" in any of your AI-Traffic Bgls. Even if I don´t know what traffic system you´re currently using, (especially checking relevant traffic bgls in Traffic Explorer (double files?)), it´s never wrong to recheck traffic within ACA2005 (or 2007) and search for any misbehavior in your AI traffic structure.

 

Further down the road: have you checked your areas around OMDB/EDDF for overlapping (wrongly coordinated) scenery boundaries with LWMviewer? Think, "loaded bgl for this area" would be the function...

 

God luck for now!

So my question is if there are experiences with too many entries in Scenery.cfg causing FS2004 to behave heavy and causing OOMs. Is there a certain limit below the known limitation of 999 entries, depending on PC specs and OS? Any thoughts were helpful. Perhaps martinlest2 can step in and recapitulate his thoughts or add new experiences.

I don't know any issues as you describe, but most importantly there is not a limit of 999 entries. I currently have 1301 entries with no problems whatsoever.

 

I concur it is probably an AI issue. The problem with testing such problems is that aircraft are in different places at different times of the day so it isn't always easy to re-create a situation.

 

John

Hi Alex and John,

 

thanks for your input. The check with LWMViewer would be a thing I didn't think about so far.

 

My traffic structure contains only four traffic files - each one for commercial, GA, military and heli traffic. They are compiled from WOAI packages which I didn't want to install their own file path. I may compile them once again.

 

So if the problem is AI-related, there must be a faulty texture, as I checked already for missing textures.

 

But the the question arises again why my OOM has gone when I deinstalled some sceneries in the central Pacific while letting my AI traffic and the other sceneries - especially these where my OOMs occured - unchanged. May it be that uninstalling some sceneries decreases memory usage so that an existing AI-traffic- or scenery-related fault won't get evident?

 

By the way I did my tests always at strictly the same time at the same day. I created a dedicated new flight only for this purpose.

 

Maybe the following observation is useful: I monitored Memory usage via Windows' Task Manager. If I'm for instance at EDDF and then I change to OMDB there is a sudden peak of memory usage after hitting "OK", when the Progress bar is still at 0 %. This was the moment when usually OOM occured. But approaching OMDB in flight, the OOM occured with the cityline and even the airport already in sight. For me, all this makes no sense.

 

By the way, I didn't mention that the OOM was mostly followed by a CTD message that "facilities.dll" caused a problem. Is there a connection between facilities.dll and the facilities indices which are built every time Scenery.cfg was changed?

Facilities.dll isn't discussed in your - Johns - very helpful FS2004 fault finding tutorial.

 

 

I currently have 1301 entries with no problems whatsoever.

 

And all of them are active at the same time?

May I ask you if you have a 32bit- or 64bit-OS?

 

Thanks

Harald

I would suggest to deactivate all sceneries and reactivate them by 20 or 30 until you found whether OOM is caused by one of the sceneries. Best and easiest way is using FS Scenery Manager. If you get an OOM with the first reactivated sceneries, the cause could be your AI traffic. Therefore you have to deleted it, and reinstall it step by step  from the basket.to your traffic folder.  

 

Bernard

 

 

And all of them are active at the same time?

May I ask you if you have a 32bit- or 64bit-OS?

A quick check shows all but three active.

 

I'm currently 64-bit but my FS setup is still on my old 32-bit machine for testing.

 

John

However, I repeated the flight on which I recently got OOM with my "decreased" Scenery.cfg. Now I was able to end the flight without OOM, letting all things else and especially my AI-traffic untouched. A check with ACA2005 had resulted in some potentially missing lightmaps, which turned out to be "false positives", so there was no indication of one of these obvious faults which ACA2005 can detect. 

 

Nevertheless, I ran ProcMon which indicated two missing textures in sceneries, one in Matthew Ministry's Sharjah scenery nearby and over southern Iran ("HEATA.BMP").

 

And:

What I saw else in ProcMon was FS9 constantly reading ".bgl" files of other sceneries all over the world. Maybe this behaviour is one contributor to my OOM problems.

 

Thanks so far

Harald

ACA only looks for filenames. If a texture file is corrupt, it won't identify it.

 

Whilst Leuen has given you a way of investigating deeper I think I would start at the other end and disable all AI to start with, then progressively reintroduce them. Over the years I have found four or five dodgy AI aircraft, either through corrupt textures or some kind of faulty model. Having just thought of the latter, if you have any AI helicopters try disabling them first. I do recall a faulty model but don't ask me which!

 

John

 

Hi John,

 

I also prefer disabling AI traffic first. It seems that if there is a faulty plane, it is within the commercial AI traffic, as I already investigated. I think I will then remove the AI planes, put them back model by model and see what happens.

 

My only "problem" ist that since I deinstalled some sceneries, the OOM didn't occur again...!

 

Thanks for your help and your contribution to the flightsimming topic in general!

 

Harald