747 v3 Steady VAS increase [Due to nVidia drivers - not 747]

[Wise87 1,281]

I'm interested in what drivers those effected are using and if rolling back indeed solved the VAS issue. I'm currently using 376.53 with a 1080 and I'm not seeing the huge VAS increase but I'm interested in what the best Nvida drivers to use. 

[riftin11 2]

I have the same issue that I have on the 777 in terms of VAS, where there is a memory leak on certain runways - VHHH 07R, KSFO 28R. Its discussed in detail here: http://www.avsim.com/topic/477686-p3d-v3-fmc-memory-leak-back/

 

My first ever flight in the 747 was to find out if the same memory leak exists, I tried to do KSFO-VHHH landing on ILS07R - got an OOM about 8 hrs in, same as on the 777. Tried to do the same flight yesterday, but set the approach to ILS07L, and no memory leak, landed with 1GB VAS remaining. There's something PMDG code related that's causing these leaks, as I have them in the exact same scenarios, on both the 747 and 777, can someone look into this - more specifically that thread I linked above?

 

I'm also using the "good" driver - 376.33

[DrumsArt 270]

Were you by chance using Navigraph recent AIRAC? Reason I ask, there was/is an issue with VAS slowly creeping up in the PMDG 777 due to an issue with certain line entries in some but not all STARS. If you were using STAR in your plan in the FMC, perhaps that is the cause. Easy way to find out, refly the flight and wait until TOD to enter the STAR into FMC. If that was the problem, you should be ok on VAS.

 

Just a thought and possible easy work around.

 

Hope you get it worked out.

Hi,

 

Please, have you some references/details about which Navigraph recent Airac ?

 

Thanks and Regards,

 

Richard Portier

[Wise87 1,281]

Hi,

 

Please, have you some references/details about which Navigraph recent Airac ?

 

Thanks and Regards,

 

Richard Portier

 

Made no difference according to others. 

[Harold_Finch 29]

I have the same issue that I have on the 777 in terms of VAS, where there is a memory leak on certain runways - VHHH 07R, KSFO 28R. Its discussed in detail here: http://www.avsim.com/topic/477686-p3d-v3-fmc-memory-leak-back/

 

My first ever flight in the 747 was to find out if the same memory leak exists, I tried to do KSFO-VHHH landing on ILS07R - got an OOM about 8 hrs in, same as on the 777. Tried to do the same flight yesterday, but set the approach to ILS07L, and no memory leak, landed with 1GB VAS remaining. There's something PMDG code related that's causing these leaks, as I have them in the exact same scenarios, on both the 747 and 777, can someone look into this - more specifically that thread I linked above?

 

I'm also using the "good" driver - 376.33

 

Just for the record, my first few flights were indeed KSFO outbound from 28R.

 

Harold Finch

[edpatino 1,964]

Some guys at FSLabs forum reported increasing VAS usage after NVIDIA driver update.

At first, FSLabs update was blamed for VAS increase. But later they found out NVIDIA latest drivers were the cumprit

 

I do not know if I can post the link here, but...

 

http://forums.flightsimlabs.com/index.php?/topic/11263-vas-usage-insane-after-update/.

 

Which "latest" drivers are you referring to, please?. Could you pls indicate version?.

Thanks, Ed

[Jack_Sawyer 4,600]

Guys,

 

We did a *ton* of VAS testing during beta - it was a primary concern. One of the new testers for this cycle (Kyle Weber from this forum) was brought on specifically for his expertise in tracking these types of leaks down. We were all confident, including him, that we'd found and killed any that existed or we wouldn't have released. Thousands of hours were flown testing this.

 

I'm not saying that there isn't still an issue somewhere, but I would be very surprised if there is. In testing for this you guys really need to eliminate all the other possible variables. When we did this during beta, we were using 100% clean builds of the simulator with no other addons present. That is the only way to prove causality here - simple things like a bad AFCAD file thousands of miles away from your route can actually cause issues and you have to make absolutely certain that isn't happening.

 

Kyle made an Excel macro during beta that takes VAS data collected by FSUIPC and automatically formats and graphs it so that we can see instantly whether there's a constant negative slope in the curve. An actual leak will be extremely precise and regular, you won't see the VAS suddenly jump up by a large amount or something like that in a leak scenario. As an example, one leak we found in beta was caused by the vibrometer sound for the analog standby instrument loading into memory over and over while it was active due to a bug in the way the switch from the ISFD to analog was initialized (the initalization essentially happened over and over). That's the kind of thing we're talking about here - if you have a giant sudden increase, that's far more likely to be scenery or AI related than it is the airplane.

 

I'm going to ask Kyle if it's alright to provide his spreadsheet to the forum at large and you guys can use it to run tests. There was a whole methodology we came up with that I will post as well.

 

I just bought this and I have all Orbx scenery.  I never have VAS issues but last night after buying and installing it all I did was a short pattern flight around KPSP.  Yes things were extremely smooth with FPS of 60+.  However, I was astonished to just read the above where you did all your beta testing using clean builds and no add-ons. This isn't very real considering most people have a ton of scenery addons.  After a simple pattern flight my VAS was tanked down to something like 400 MB left.  If I had made another pattern I would have OOM'd.  P3D 3.4 without the latest HF.  Win 7 64.

[Harold_Finch 29]

Guys,

 

We did a *ton* of VAS testing during beta - it was a primary concern. One of the new testers for this cycle (Kyle Weber from this forum) was brought on specifically for his expertise in tracking these types of leaks down. We were all confident, including him, that we'd found and killed any that existed or we wouldn't have released. Thousands of hours were flown testing this.

 

I'm not saying that there isn't still an issue somewhere, but I would be very surprised if there is. In testing for this you guys really need to eliminate all the other possible variables. When we did this during beta, we were using 100% clean builds of the simulator with no other addons present. That is the only way to prove causality here - simple things like a bad AFCAD file thousands of miles away from your route can actually cause issues and you have to make absolutely certain that isn't happening.

 

Kyle made an Excel macro during beta that takes VAS data collected by FSUIPC and automatically formats and graphs it so that we can see instantly whether there's a constant negative slope in the curve. An actual leak will be extremely precise and regular, you won't see the VAS suddenly jump up by a large amount or something like that in a leak scenario. As an example, one leak we found in beta was caused by the vibrometer sound for the analog standby instrument loading into memory over and over while it was active due to a bug in the way the switch from the ISFD to analog was initialized (the initalization essentially happened over and over). That's the kind of thing we're talking about here - if you have a giant sudden increase, that's far more likely to be scenery or AI related than it is the airplane.

 

I'm going to ask Kyle if it's alright to provide his spreadsheet to the forum at large and you guys can use it to run tests. There was a whole methodology we came up with that I will post as well.

Ryan,

 

     I certainly do appreciate the assistance.  I do not believe that there is any general "problem" with the 747, just that somehow I have a "semi-borked " configuration that presents issues with regards to to VAS usage.

 

Harold Finch

[denthibbe 18]

Some guys at FSLabs forum reported increasing VAS usage after NVIDIA driver update.

At first, FSLabs update was blamed for VAS increase. But later they found out NVIDIA latest drivers were the cumprit

 

I do not know if I can post the link here, but...

 

http://forums.flightsimlabs.com/index.php?/topic/11263-vas-usage-insane-after-update/.

 

 

yep the lastest driver leak VAS

I can confirm this issue.

Last week I suddenly had the VAS leaking like hell (also with Aerosoft's A320 for instance) making a 3 hour flight the absolute maximum.

 

Turned out I had indeed updated my nvidia drivers to the latest version (jan24th iirc). I just re-downloaded the previous one (somewhere december 2016) uninstalled the latest one followed by driver sweeper and made a clean install of the downloaded nvidia driver.

 

It solved the leaking for me.

 

YMMV

[Jack_Sawyer 4,600]

I can confirm this issue.

Last week I suddenly had the VAS leaking like hell (also with Aerosoft's A320 for instance) making a 3 hour flight the absolute maximum.

 

Turned out I had indeed updated my nvidia drivers to the latest version (jan24th iirc). I just re-downloaded the previous one (somewhere december 2016) uninstalled the latest one followed by driver sweeper and made a clean install of the downloaded nvidia driver.

 

It solved the leaking for me.

 

YMMV

 

Please tell us what version it was that you installed?  I have NVIDIA version 376.33 on a GTX 980.

[edpatino 1,964]

 

 

I'm going to ask Kyle if it's alright to provide his spreadsheet to the forum at large and you guys can use it to run tests. There was a whole methodology we came up with that I will post as well.

 

Hi Ryan this would be an excellent support tool to better test the VAS.

Cheers, Ed

[denthibbe 18]

Please tell us what version it was that you installed?  I have NVIDIA version 376.33 on a GTX 980.

 

I had updated to 378.49 (24jan2017) and reverted back to 376.33 (14dec2016). After that my VAS leak was gone.

 

Hope this info is useful to anyone.

[Paul_Smith 92]

So most people can have the latest PMDG aircraft, the latest Orbx scenery, and the latest Nvidia drivers, but some people can only have two out of three. How long before two becomes the norm and some people can have only one of the above?

 

Is there anyone who still believes that the FSX platform is the future of flightsimming?

[riftin11 2]

Just for the record, my first few flights were indeed KSFO outbound from 28R.

 

Harold Finch

 

Interesting... in my case departure runway is irrelevant, just arrival runway. Took of on 28R both times, OOMed selecting ILS07R at HKG, with a steady decrease over the flight, while wiht ILS07L selected my VAS held steady at 1.1GB

[squerble 300]

I just bought this and I have all Orbx scenery.  I never have VAS issues but last night after buying and installing it all I did was a short pattern flight around KPSP.  Yes things were extremely smooth with FPS of 60+.  However, I was astonished to just read the above where you did all your beta testing using clean builds and no add-ons. This isn't very real considering most people have a ton of scenery addons.  After a simple pattern flight my VAS was tanked down to something like 400 MB left.  If I had made another pattern I would have OOM'd.  P3D 3.4 without the latest HF.  Win 7 64.

 

I'm not sure why you're so astonished; how else are you supposed to know exactly how much VAS the add-on you're developing is using?