Nvidia 378.78

[jabloomf1230 2,445]

2 hours ago, Jim Young said:

These are supposedly the best drivers put out by Nvidia, performance wise for DX12 - http://www.guru3d.com/news-story/quick-comparison-geforce-geforce-378-66-versus-378-78-directx-12-performance.html

"We do have to state that we are using a fast clocked (8 cores @ 4.3 GHz) 16 thread processors (5960X) this might hinder the threading performance that DX12 optimizes so much, Have a peek at our results."

The fact that the driver is optimized for CPU threading in DX12 might also explain why it is a terrible VAS  hog in DX11. Or not...

[Wolf0 31]

Back to the OP question. 

On my system this driver still leaks VAS.

Test setup: P3Dv3 (latest build), FSLabs A320 (latest build). Start simulator (no weather addons, cameras, anything else) and wait for the simulator to stop loading textures and stabilize. Place the camera on virtual cockpit and never touch any key for the next hour. Airplane is cold and dark. This happened at time 1511Z. These are the VAS usage statistics I've made:

1511Z - VAS usage 2772MB;

1527Z - VAS usage 2861MB;

1549Z - VAS usage 2987MB;

1554Z - VAS usage 3019MB;

1615Z - VAS usage 3145MB.

After 1h and 4 minutes VAS usage rose by 373MB. The initial conditions remained the same. No key was pressed during this period, no weather injected, no settings have been changed.

Let me just add that many users confuse VAS leaks with the VAS usage rising while performing a flight. VAS usage throughout the flight is no way to ascertain anything about the driver itself. It may only indicate that the simulator isn't releasing unnecessary VAS as quicker with a particular driver comparing with another.

The point is that a driver leaking VAS will consume a steady amount of VAS even with the simulator placed on tower view looking at the tarmac. While we may benefit from a driver using VAS more rationally and seeing this manifestation as the ability to finish a flight to a complex airport with on driver instead of another, this does not represent evidence of leaks. 

I'm back to driver 376.33 which was the latest build that didn't leak VAS on my combination of hardware/software.

Best regards.

 

[Boeing or not going 578]

5 hours ago, penta_a said:

I have a flight that I do frequently EGLL-LOWI-EGLL PMDG 737NGX, tried all drivers (376.33 to the latest) and found the following,

1- I get OOM warning 30-40min inflight using driver higher than 376.33/53 (I'm using Windows 10 Pro 64bit insider build 15048, it enforces driver update to 376.53 that I didn't find at nVidia website, but it works like 376.33 with no issues).

2- with driver 376.53 I can make it back to EGLL with no issues.

3- Activating Windows Game mode under any driver will cause P3D to crash after about 20min inflight, I did not isolate running programs that work with P3D (AS16, PSXseeconTraffic, EZdoc.. etc) to find out what causes the crash.

4- Performance is good, around 25-28 FPS at busy airports and about 55 FPS during cruise (my P3D settings are moderate) to avoid constant OOM over Europe. I have a very bad pauses/stutters during AI PSXseeconTraffic loading, it disappears after I turn away from busy traffic airports, EGLL in this case.

 

Something is wrong with your setup. Hitting an OOM in the PMDG 737 the lightest vas usage airplane they released is not normal. Probably caused by that traffic program.

[nicksort 8]

6 hours ago, penta_a said:

I have a flight that I do frequently EGLL-LOWI-EGLL PMDG 737NGX, tried all drivers (376.33 to the latest) and found the following,

3- Activating Windows Game mode under any driver

 

I am missing something here, what does "Windows Game Mode"?

Where do I look for this as I cannot complete any flight longer than 20-30 min without OOM!

[penta_a 428]

3 hours ago, nicksort said:

I am missing something here, what does "Windows Game Mode"?

Where do I look for this as I cannot complete any flight longer than 20-30 min without OOM!

that is a new mode under windows settings coming out with next windows 10 update, called creatives update I guess, I'm running insider preview and it comes with it.

 

[penta_a 428]

3 hours ago, 777200lrf said:

Something is wrong with your setup. Hitting an OOM in the PMDG 737 the lightest vas usage airplane they released is not normal. Probably caused by that traffic program.

for the early OOM warning, I have disabled all traffic with no luck.

[TheBoom 31]

Updated to this driver to test VAS usage. OOM-ed before I could even takeoff lol. 

LIRP X with AS A320 and all other regular addons. 

[virtualstuff 752]

Currently testing (1070) at this point no VAS drama here local, and actually performing the same as 376.33.

Have to do some more tests to decide to keep the driver or not...

[G-YMML1 582]

On 3/12/2017 at 5:54 PM, awf said:

Currently testing (1070) at this point no VAS drama here local, and actually performing the same as 376.33.

Have to do some more tests to decide to keep the driver or not...

Andre, 

 

Are you on W7 or W10?

[virtualstuff 752]

W10 but after a longer test had VAS problems so reverted back...

[jabloomf1230 2,445]

NVidia now is up to 378.92 if somebody  wants to be a test subject.

[vortex681 806]

3 hours ago, jabloomf1230 said:

NVidia now is up to 378.92 if somebody  wants to be a test subject.

After you!

Seriously though, I gave up asking what people thought of the latest drivers about a year ago when I read that a particular driver was the best ever for my card. When I installed it I thought that I'd downloaded the wrong one. Noticeably worse performance than my old driver. So much seems to depend on your specific system/components that the only way to really know how good a driver is is to try it yourself. That said, I'm a great believer in the old adage "if it ain't broke, don't fix it!".

[Wolf0 31]

Not good news on my system.

Since 376.33 all releases leak VAS on my setup. This one is no exception. 

Same test as always: load P3Dv3.4 with no weather software and no additional add-ons. Let the plane sit with engines off at the gate and don't touch anything else. After loading VAS usage was 3007MB. After 3:30h VAS usage was 3907MB...

I'm leaking VAS at approximately 300MB/hour. Enough for some users to complete a small flight and report "no problems".

Driver 376.33 and older ones don't leak VAS, i.e., VAS usage remains the same in this test scenario.

[Marius_S 4]

So, as it looks at the moment, we need to wait until Prepar3D v4 before we can update the driver...

No big deal for me though, 376.33 is working fine.

[Alberto Zanot 53]

Same here. GTX980ti Amp Extreme, Win10 64, P3D 3.4.22, no addons, no AI, default F22 Raptor.

I did my usual test, OOM for my setup after 3 circles over NY from KJFK.

Back to 376.33. Thanks Nvidia for supporting us simmers ;-)