nVidia video driver - Any reports?

[GSalden 4,656]

1 hour ago, willy647 said:

I'm on 376.53 and all is well. So why should I bother to update to newer releases ? I understand the new releases are to support the latest games, but bring no benefit what so ever to P3D. Reading the comments here, they only cause headache. I rather spend my time flying than solving driver problems. Just my thoughts

 

 

 

 

You are absolutely right.

Only with the newest cards ( improving the drivers ) it will help.

 

[RancidViper 37]

1 hour ago, willy647 said:

I'm on 376.53 and all is well. So why should I bother to update to newer releases ? I understand the new releases are to support the latest games, but bring no benefit what so ever to P3D. Reading the comments here, they only cause headache. I rather spend my time flying than solving driver problems. Just my thoughts

 

 

 

 

Well probably because not everyone here uses their computers for flight simming only.. Just a thought? :)

[gboz 157]

Deleted.

gb.

[Schlotterknie 35]

For those still having trouble with VAS-leaks and OOMs with Prepar3d v3:
I´ve had more or less heavy VAS-leak issues with all these GeForce drivers between that "legendary" (and meanwhile pretty old) driver 376.33 and the newer 381.65.

As 381.65 was the first revision that fixed that odd VAS-leak completely at least on my system (i7-4790K, GTX980) I was curious to test the brand new version 381.89 which was released yesterday. For that test I left the PMDG 747 v2 standing still at EHAM Gate F9 with p3d-config sliders at maximum, AI traffic at 100% (UT2) and time lapse at a rate of 4x.

The VAS increase after 15min:
381.89: +50MB (and still kept rising after that)
Bad surprise so I rolled back to previous 381.65 and did exactly the same test again:
381.65: +0MB

That plus 50MB after 15min doesn´t sound like a disaster at first view but considering that the plane was not moving and that I didn´t change cameras or view angle during that 15 minutes it might be a much bigger problem when flying where much more objects are loaded and (hopefully) released from memory. That´s why I´ll definitely stay with 381.65...

Btw. I strongly recommend to turn off Hyperthreading in your PCs bios as well as the "Shader Cache" and "Threaded optimization" using Nvidia Control Panel or Nvidia inspector. Doing so saves a lot of VAS.

[Alberto Zanot 53]

So, neither this 381.89 solved the vas leak?

Someone else could report?

[Wolf0 31]

Well, my experience with this driver is different. I'll copy the report I've made on FSLabs forum regarding this driver yesterday:

381.89 Seems to be working fine.

Test results: Nvidia 980Ti, Win7 64, P3Dv3.4 (latest), FSLabs A320 (latest), T2G LFPG at the gate, no weather, no AI traffic, time of day noon, water setting: ultra + bathymetry.

Load the simulator with 376.33 on VC, wait until VAS stabilizes. VAS usage: 2852MB.

Load the simulator with 381.89 on VC, wait until VAS stabilizes (after a driver clean install and reboot). VAS usage: 2818MB. VAS remains stable indefinitely.

2042Z - VAS usage: 2818MB;

2055Z - VAS usage: 2817MB;

2110Z - VAS usage: 2816MB;

2146Z - VAS usage: 2817MB.

To summarize VAS usage is roughly the same as 376.33 and VAS usage is stable with time (no leak) on my setup.

Today I've already made two flights: LFPG->LIMC->EDDF without restarting the simulator, using live AS16 WX and GSX and arrived at EDDF with +500MB of VAS left. VAS usage was as expected for an well behaved driver.

I guess this driver drama depends a lot on the combination of hardware/software and I still couldn't pinpoint any obvious correlation. I've suffered from the infamous VAS leak since 376.33 but as of now I have everything back to normal with the current driver.

[georgio1952 52]

I don't have any problem with NVidia 381.89 driver ...I made a flight "KSDF to KPDX"  it's smooth as butter...

[Alberto Zanot 53]

3 hours ago, Schlotterknie said:

Btw. I strongly recommend to turn off Hyperthreading in your PCs bios as well as the "Shader Cache" and "Threaded optimization" using Nvidia Control Panel or Nvidia inspector. Doing so saves a lot of VAS.

WoW i really have to thank you. Doing this it improves a lot on my system, and with the latest drivers.

Thanks.

[fluffyflops 1,858]

Is it me or are the problems linked to 980ti cards as suggested above.

I've got a 980ti and had to go back to 376.33 drivers as I Was getting the leak big time.  Going back has sorted the issue.

 

Schlotterknie can you screenshot me your ndivia control panel settings as I have the same card..

Many thanks in advance 

 

[Pete Dowson 558]

On 4/25/2017 at 5:12 PM, GSalden said:

... and everything is working with W10.

Yes, I know it is okay for lots of folks. Not here.

I've had to make changes in FSUIPC because some Windows APIs it uses causes crashes -- of those programs you use there are probably more than just FSUIPC that have had to be modified to suit W10. It is most certainly not a fully backward compatible version of Windows at all -- in fact it seems more different than even moving from XP to Vista to W7.

The problem API in FSUIPC which I had to avoid has up to now caused me approaching 4 weeks of solid work, and i still haven't reached the "fully working in all cases" solution, one which worked fine on Win7. And it is definitely a standard, documented Win32 API function which is crashing and I am having to find work-arounds for.

I have similar problems with different API functions on my assorted drivers which are written by me, and for me, for additional hardware in my cockpit. I use Flight Illusion, Cockpitsonic, Sismo and PFC equiipment and all with my drivers only, not theirs. I like to be in full control of what is going on hardware-wise.

I expect documented Windows APIs to work properly on Win10 as they always have on XP, Vista and Win7 -- unless they are marked as no longer supported, which these most certainly are not. And, yes, I am using the most recent on-line referenced available for them..

Carry on happily with W10. I am not decrying that. I just have very good reasons to hate it!

Pete

 

[Pete Dowson 558]

On 4/25/2017 at 7:17 PM, onebob said:

Hi Pete update 381.89 popped up on my w7 this afternoon.

Aha! Good. Thank you!

Pete

 

[Schlotterknie 35]

1 hour ago, tooting said:

Is it me or are the problems linked to 980ti cards as suggested above.

I've got a 980ti and had to go back to 376.33 drivers as I Was getting the leak big time.  Going back has sorted the issue.

 

Schlotterknie can you screenshot me your ndivia control panel settings as I have the same card..

Many thanks in advance 

 

I use a GTX980 not a 980ti but I don´t think that makes a huge difference in the way P3D handles VAS anyway.
However I don´t use the Nvidia control panel for that configurations, I do that with the Nvidia Inspector. It doesn´t make much difference as the concerned settings are equally labeled in both tools.
So here´s what I do with the nv-inspector:
 

[bean_sprout 1,201]

On W10 with 381.65 no issues.  Running REALLY WELL.  I especially like the new NVidia CP menu item to disable the "Windows Experience Improvement" Option.  Had VAS issues with earlier revs.   

 

Cheers

jja

[A340Bangla 40]

2 x GTX980ti here, W10, 381.65 running well. No VAS issues, even SLI seems to work again.

[willy647 193]

23 hours ago, Schlotterknie said:

Btw. I strongly recommend to turn off Hyperthreading in your PCs bios

I have an i7 7700K (4 cores + HT) and not using any AM .  Looking at the resource monitor (perfmon.exe)  I noticed that when the sim is loading scenery, it uses all 8 cores. On my setup, disabling HT makes it noticeable slower and produces stutter in flight. Just to show that every system is different. Its a matter of try and error