Possible new VAS issue in v3.4

Any issues, problems or troubleshooting topics related to the Prepar3D client application.
rubster
Posts: 12
Joined: Fri Jul 05, 2013 8:15 am

Re: Possible new VAS issue in v3.4

Post by rubster »

I just disconnected the HTC vive and the VAS is better. Also the rest of the performance. I think although not using the Vive it's still taking a lot of recources, because the lenses boot up with p3d and it has to be instant ready when changing views. Not to mention steam VR. So disconnecting the Vive helps for me for now.
ahuimanu
Posts: 112
Joined: Mon Dec 17, 2012 5:45 am

Re: Possible new VAS issue in v3.4

Post by ahuimanu »

downscc wrote:Things got worse for me arriving at KLAX from VHHH. I've flown the PMDG 777 extensively without VAS problems but suddenly OOM errors appear in v3.4
...
Dan, I noticed a thread from you at AVSIM where you say you sorted things out. How is this the case? Since you seem to do longhaul in the 777, what did you do to ensure completed flights?
User avatar
downscc
Posts: 1623
Joined: Mon Dec 01, 2014 5:46 pm
Location: KCRP

Re: Possible new VAS issue in v3.4

Post by downscc »

ahuimanu wrote:
downscc wrote:Things got worse for me arriving at KLAX from VHHH. I've flown the PMDG 777 extensively without VAS problems but suddenly OOM errors appear in v3.4
...
Dan, I noticed a thread from you at AVSIM where you say you sorted things out. How is this the case? Since you seem to do longhaul in the 777, what did you do to ensure completed flights?
I thought I posted earlier in this thread that I deleted cameras.cfg and let P3D rebuild it. For whatever reason, the cfg file had cameras for Occulus Rift but I've never had such a device by any name. The new camera.cfg file has VR definitions but the views do not cycle through them (pressing S repeatedly) like it did for the Occulus cameras, and it was the view cycling that destroyed VAS.

I now see better VAS management than ever..., during a long haul B77W flight the VAS remaining stays right around 1200 MB whereas in earlier v3.x it would stay about 800 MB remaining. Test flights are LAX-SYD.
Dan Downs
KCRP
vmounier
Posts: 21
Joined: Sun Apr 12, 2015 4:47 pm

Re: Possible new VAS issue in v3.4

Post by vmounier »

Quick update as I just remembered I was actually the OP here...

My testing has been absolutely random, I am ashamed to report, and I have no hard facts to post despite Beau's kind intervention and request for feedback.

That being said, I have not had an OOM since shortly after starting this thread, and they had been happening like fireworks before that. So what have I changed?

I did not delete my camera.cfg file as I use EZDok and was lazy, not wanting to lose its entries, but I did delete the Occulus entries in the file. I did not revert to OPTIMIZE_PARTS=0 as the consensus seems to be it costs a few FPS, and I need them all.

I also started using PTA rather than Reshade/Mastereffects, and through more performance testing, ended up switching from AM 116 back to AM 85. I had a catastrophic update to Windows 10 Anniversary which would not recognize my GPU no matter what and left my PC in limbo between BIOS and OS, and reverted to previous version. I lowered my weather settings a tad, but am still using 2XSGSS. I updated to the latest hotfixes for AS16 and ASCA. I killed all road traffic.

That's about it. I assume that one or a combination of the changes above will have been responsible for the improvement. This is far from scientific testing, but with limited time and a need to enjoy flying rather than spend hours tweaking, that's what I got.

Last night, I was flying the Dash and Jetstream in heavy weather at dawn in Valdez, and the 777 at YVR, and had the best visuals I have ever had, with 40 to 60 FPS around and in clouds in Valdez, and 30 FPS on final in Vancouver, and no VAS issues! I was speechless.
Vincent Mounier
gpickup
Posts: 4
Joined: Sun Sep 23, 2012 3:52 pm

Re: Possible new VAS issue in v3.4

Post by gpickup »

I'm seeing significant issues too.

Installed P3D v3.4 on totally clean install and brand new machine.
Added Orbx PNW
Orbx Bowerman
FSDT Vancouver
REX 4

Very modest settings:
Medium textures (P3D, REX and FSDT Vancouver)
LOD = high
Autogen veg = dense
Autogen bld = normal

I can't fly from Bowerman to Vancouver and get the OOM approaching the Vancouver city coastline from the south using any of the default planes (but mostly the F22 for expediency of testing). This happens every time I try and I always see a spike in VAS once I clear the Olympic Peninsular and start crossing the water to San Juan Island.

Interestingly I can fly the other way from Vancouver to Bowerman and land but will OOM if I then do a couple of circuits at Bowerman.

Wondering if FSDT Vancouver was driving this I just tried Miami to Jamaica in the PMDG 737. Got to within 200 miles of Jamaica and then OOM - flying over nothing but water...

It just doesn't look like the memory/garbage collection that was working in previous releases is working now - very frustrating to get so many OOMs when this was all possible in v3.3.

I tried the camera config delete and it's made no difference to me. I also tried the Optimise Parts change that Rob mentioned and whilst that does buy me and extra 5/10 minutes I still OOM on the Bowerman to Vancouver flight.

Be happy to provide further information if requested.

Thanks,
Graeme
gpickup
Posts: 4
Joined: Sun Sep 23, 2012 3:52 pm

Re: Possible new VAS issue in v3.4

Post by gpickup »

I recalled a comment by Saul some time back about Hyperthreading turned off frees up significant VAS and with me installing to a new machine I double checked if Hyperthreading was on for me. It was.... So I disabled it and tried my test flight of Orbx Bowerman to FSDT Vancouver. Not only was I able to complete the flight but I turned around and flew back to Bowerman - no OOMs. VAS reached about 3.5GB and maintained that level - even when i'd left Vancouver and looked backwards in the VC (the airport had to reload) VAS stayed at 3.5GB.

Disabling hyperthreading has made a difference to me so I wanted to share in case it help anyone else or means something to the developers.

I'll do some more testing over the weekend and see if things are okay on longer flights.

Thanks,
Graeme
Mickeyj
Posts: 320
Joined: Sun Mar 24, 2013 5:50 pm

Re: Possible new VAS issue in v3.4

Post by Mickeyj »

Hi guys,

i just want to share my experience. I just did a flight from KMIA to MWCR in the NGX. Stock airports with ORBX openLC NA as well as REX4 Soft Clouds AS16+ASCA 40% of SKAI Traffic 2.0 (which is not much for SKAI Traffic) and fairly high scenery settings. I started with 2.0 GB used VAS and ended up after landing, parking and deboarding with GSX @2.2 used VAS. I did lots of outside view switching during the flight. I try a flight from UK2000 Humberside over ORBX ENG&SCO to UK2000 Aberdeen in the NGX to see what happens. Btw I left OP=1 for the time. In my case OOM´s occur rarely, i really have to push the sim to give me an OOM.
gpickup
Posts: 4
Joined: Sun Sep 23, 2012 3:52 pm

Re: Possible new VAS issue in v3.4

Post by gpickup »

Turning hyperthreading off has certainly made the difference to me. I'd love to understand why it has such an impact...

Anyway, I've installed more and completed several longer flights and not one OOM:

FSDT Toronto to JFK in Majestic Q400
JFK to FlyTampa Princess Julianna in PMDG 737
Sacramento to FlightBeam San Fran - I thought this one would get the OOM but it was fine (with Orbx California) in Carenado Hawker 850

Thanks,
Graeme
User avatar
Afterburner SST
Posts: 67
Joined: Wed Oct 07, 2015 8:44 am

Re: Possible new VAS issue in v3.4

Post by Afterburner SST »

I have conducted my favorite test flights from KIAD to KJFK using ConcordeX and AS2016 after having updated the Client from v3.2 to 3.4, and I can say that the VAS consumption in v3.4 has NOT been noticeably higher overall than in v3.2 (in some segments it was even lower) despite having OpenLC NA installed in v3.4 versus not having had it on my system in v3.2.

(To give you some numbers: Using MyTraffic 5.4B with 25% airliner traffic, very high autogen settings and having everything activated in ORBX Vector, I started with 3000 MB VAS and ended up with almost 3500 MB VAS at the gate in JFK). I think ConcordeX is a sufficiently complex airliner to push the VAS. However, I should let you know that I deleted the cameras.cfg file before starting the sim and also manually erased all entries following "CameraDefinition.011". I left OptimizeParts=1 and have a quad-core CPU without HT.
User avatar
Raphael_Chacon
Posts: 82
Joined: Tue Apr 07, 2015 5:40 pm

Re: Possible new VAS issue in v3.4

Post by Raphael_Chacon »

Well,

Sorry to say, but +1 Really ugly VAS performance on V3.4

For expl, i used to fly KSFO-KIAD(ORB+Flightbean+PMDG777+AS16+REX Clouds) without any issue. Sometimes i had issues at KLAX+SCA, specially when arriving there, but i could still plan, takeoff and land at another payware airport after that without further issues.

Then, after updated to V3.4(Full install), im taking off from KIAD-KSFO(FlightBeam), and after 2 hours of flight, i get the VAS error(Which used to be a frequent route for me). Im monitoring it, and it only goes up and up and up.... it increases until the error. If i try to open again the sim and restart from previous point(Lets say, 38,000FT, near Denver), well, same history, VAS increases and increases, and increases, even with very very limited camera view manipulation. Dont men tion if im near KSFO, i would immediately get a VAS CTD.....

W10 64bits
i7-6700k OC 4.4Mhz
32GB RAM 3200Mhz
AM=85
TextureMaxLoad=12
UseGlobalTerrainView=True
SHADER_CACHE_VERSION=1
LOD_RADIUS=8.5
1024 Texture resolution(Used to have 4096 without issues, now 1024 and getting OOMs?)
My P3D Settings / Default AS16 Settings / Low resolution clouds for REX TD + this Nvidia Profile(Inspector)
ORBX (Global+LC+Vector) + FSGlobal 2010

I havent ever had issues with this new PC, in fact, not like that even with my older i5+GTX580. Im a long hauler mostly, and there is definitely something about VAS with the P3DV3.4 that i can not even perform a 2 hours flight.

Will try some recommendations seen here and see what happen, but seems like a band-aid to me...

Will keep in touch... Txs
Image
INV-2071056
bcrawley57
Posts: 12
Joined: Fri Sep 12, 2014 8:07 pm

Re: Possible new VAS issue in v3.4

Post by bcrawley57 »

I have the same VAS issues, to the point where it is now impossible to complete a heavy flight with the system dumbed down so much that there were even no buildings at the airport. This is on a new machine that was running great until the latest client update. I tried to revert to the previous client but the problem stayed. So have reverted to the latest client in the hope that a fix will be released soon.

Sticking to very limited VFR flights now, even those are struggling with the dreaded VAS.

If you need any further information, please let me know.
Adam Breed
Lockheed Martin
Posts: 1382
Joined: Mon Sep 20, 2010 6:17 pm

Re: Possible new VAS issue in v3.4

Post by Adam Breed »

All,

In the upcoming v3.4 hotfix 2 we have reworked an area of code to more aggressively purge data when traveling large distances. We believe it might improve some of these VAS reports. Feel free to continue to report use cases and any specific reproduction steps.

Regards,
Adam
Prepar3D® Engineering Project Manager
funkyaz77
Posts: 96
Joined: Wed Mar 26, 2014 9:07 am

Re: Possible new VAS issue in v3.4

Post by funkyaz77 »

Thanks Adam,

That is what the community were mainly concerned about the actual VAS memory management which can greatly reduce OOMs so thank you for addressing this.

Blessings

Duncan
User avatar
BG2518
Posts: 731
Joined: Sat Nov 30, 2013 7:35 pm
Location: London

Re: Possible new VAS issue in v3.4

Post by BG2518 »

+1
lholdridge
Posts: 3
Joined: Wed Jun 27, 2012 1:56 pm

Re: Possible new VAS issue in v3.4

Post by lholdridge »

+1
Locked