P3D PMDG 777 memory leak

[Hoang 108]

Hello,

This is probably the first time a memory leak occur in my simulator. I tried to use PMDG 777 to fly other routes and also FSLab A320x and there isn't any OOM.

This particular route from Tokyo Narita RJAA to San Francisco KSFO first give me msvcr120.dll error but later I found out that my P3D was OOM which cause msvcr120.dll and Couatl error. I had autosave with Pro-ATC/X so I load the flight again at cruise and I got OOM again every single time after several hours. No 2D pop up. Aircraft is PMDG 777-300ER United Airlines,I program the FMC with RJAA SID CUPID3 departure runway 16R and STAR BDEGA2, trans AMAKR ILS 28R into San Francisco. I'm using newest nvidia driver with latest version of P3D, Win10.

Today after the 3rd time OOM, I dig deeper into Process Monitor and found 2 lines keep showing up:

 

D:\Game\Lockheed Martin\Prepar3D v3\Fonts\PMDG_777_DU_B.ttf                            SUCCESS

D:\Game\Lockheed Martin\Prepar3D v3\PMDG\PMDG 777X\777X_debug_log.txt         INVALID DEVICE REQUEST

 

I go into 777X_debug_log.txt and this line jumps up continuously 

Cursor counter: 6

Cursor counter: 6

Cursor counter: 6

...

 

I predict this line has to do something with the mouse cursor in the electronic checklist in PMDG 777, so I click that page, move the mouse cursor and then change to another page. Now both the "777X_debug_log.txt  " line in ProcessMonitor and "Cursor counter: 6" in debug log has stopped showing up.

Still flying the route now with 5 hours to go so if there is anything I need to check I can do it now, or I can go back to the earlier stage of the flight where I haven't touch anything.

So far, anyone has idea what is going on in my simulator? I have used PMDG 777 in FSX for months and P3D for days without any OOM, so this is quite unique.

Thank you,
Hoang Le

 

 

 

 

 

 

[Hoang 108]

I has been almost 2 hours and my available memory already dropped from ~1.7GB to ~700MB which is unbelievable. The only difference I can think of right now between this route and other routes that I flew is this one is flying through the night.

 

[scandinavian13 5,192]

3 minutes ago, Hoang said:

I has been almost 2 hours and my available memory already dropped from ~1.7GB to ~800MB which is unbelievable. The only difference I can think of right now between this route and other routes that I flew is this one is flying through the night.

This sounds a lot more like scenery, honestly.

[Hoang 108]

Just now, scandinavian13 said:

This sounds a lot more like scenery, honestly.

I already think of scenery, but if I load the flight in the middle of the ocean and now still fly in the middle of nowhere, how can a scenery/airport affect it?

Moreover, if it is because the scenery.cfg, then why all my other routes are fine except for this one? If one airport is having problem then it should affect all?

I'm just curious because I'm not an expert at this, never had OOM before.

[downscc 1,327]

PMDG already ruled out the "cursor theory."  We ran a special little beta on a couple of recompiled modules with fingers crossed.  800 MB VAS remaining is very reasonable.  The MSVCR120.dll faulting module is also an indication that you are using the left PFD 2D popup..., don't use that in the 777.  This is a known bug, has nothing to do with VAS.

[cannonplug 126]

The ILS28R at SFO is one of those known to have caused memory leaks in the past. I don't know whether this was ever completely resolved. You might want to try the same flight again, but this time wait until you're near the destination before entering the approach into the CDU.

[Hoang 108]

Just now, downscc said:

PMDG already ruled out the "cursor theory."  We ran a special little beta on a couple of recompiled modules with fingers crossed.  800 MB VAS remaining is very reasonable.  The MSVCR120.dll faulting module is also an indication that you are using the left PFD 2D popup..., don't use that in the 777.  This is a known bug, has nothing to do with VAS.

It was 800MB, now it is about 500MB and dropping, as I type this it went from 500 down to 400 now, very fast. If I fly in the middle of the ocean, I usually get around 1GB or more most of the time stable. And I have read the post about MSVCR120.dll fault which you said in there about using the pop up. I don't use it and I never use it because I seat on the FO side :D

 

Just now, wsmeier said:

The ILS28R at SFO is one of those known to have caused memory leaks in the past. I don't know whether this was ever completely resolved. You might want to try the same flight again, but this time wait until you're near the destination before entering the approach into the CDU.

This is what I am concern, I read somewhere about the memory leak cause by certain STAR/arrival procedure when set into the FMC with PMDG 777. Will test this again.

Thank you,
Hoang Le

[pcubine 143]

In some cases NVIDIA driver versions higher than 376.33 can cause significant memory leaks.

[Hoang 108]

9 minutes ago, pcubine said:

In some cases NVIDIA driver versions higher than 376.33 can cause significant memory leaks.

I have been using the latest version without any problem so far, I actually posted a very basic test for the memory leak when the newest driver came out.

The first flight I had OOM I change back to 376.33 and continue the flight again, still get the OOM. Now I am using 376.33

Hoang Le

[pcubine 143]

20 minutes ago, Hoang said:

I have been using the latest version without any problem so far

Well if it is not the driver, I would suspect what Walter Wesmeier has indicated - KSFO 28R. That runway and missed approach procedure cause memory leak topic quite often. All though as fast as the memory is leaking it could be something else.

[Hoang 108]

1 hour ago, pcubine said:

Well if it is not the driver, I would suspect what Walter Wesmeier has indicated - KSFO 28R. That runway and missed approach procedure cause memory leak topic quite often. All though as fast as the memory is leaking it could be something else.

Thank you, I'm redoing the same route again without punching the ILS into FMC.

 

[Hoang 108]

Did the same flight again, without putting KSFO ILS 28R into FMC. Took off with 1.3GB available, during climb I get up to 1.45GB available. After that it starts dropping very fast, now only 45 minutes into the flight and I'm already at 600MB and still continue to drop.

Any suggestion on what should I do? I will do a flight from Taipei RCTP to KSFO later today and see how it goes.

Hoang Le

[downscc 1,327]

Set up a test session without any addons, no weather, no traffic. Turn off bglmanx.dll and couatl.exe if you have FSDT scenery, turn off SODE.dll too.  Get to a configuration where there is just P3D, PMDG and you.  Load the flight plan without the ILS.  Monitor.

If memory doesn't "leak" on the ground then you may have something else besides the known-bug in the PMDG 777 causing problems.  The memory will leak if certain approaches are selected, you may verify that too.

[cannonplug 126]

In addition to what Dan is suggesting above, I would also verify that you are indeed on 376.33 and that Windows or nVidia didn't "upgrade" the driver without your knowledge.

[Hoang 108]

4 hours ago, downscc said:

Set up a test session without any addons, no weather, no traffic. Turn off bglmanx.dll and couatl.exe if you have FSDT scenery, turn off SODE.dll too.  Get to a configuration where there is just P3D, PMDG and you.  Load the flight plan without the ILS.  Monitor.

If memory doesn't "leak" on the ground then you may have something else besides the known-bug in the PMDG 777 causing problems.  The memory will leak if certain approaches are selected, you may verify that too.

 

3 hours ago, wsmeier said:

In addition to what Dan is suggesting above, I would also verify that you are indeed on 376.33 and that Windows or nVidia didn't "upgrade" the driver without your knowledge.

Hello,
I did a basic test similar to what Dan suggested and no memory leak with the latest nvidia driver. I will do a long haul flight today to test again.

Thank you,
Hoang Le