Topic: Titan Driver Crashes

[mporter]

Do you have a specific driver version recommendation for the Nvidia GTX Titan card?  We have experienced some crashes using the newest driver version.

Consumer-grade Nvidia GTX and AMD/ATI HD Radeon cards work fine with Lumion.

However, I'd avoid Nvidia Quadro cards and AMD/ATI FirePro cards as they are not as fast as GTX/Radeon cards, and they use non-standard driver software which can sometimes cause problems.

Here are some benchmarks for GTX and Radeon cards:
http://www.videocardbenchmark.net/high_end_gpus.html
http://www.3dmark.com/

[peterm]

No, just that the latest tend to be preferred. 

You would expect the new drivers to be more stable due to the bug fixes but occasionally there are some problems that require going back to a prior version whilst they work out the new issues.

Are the crashes with Lumion, with a certain scene or generally with other applications (3d)?

[wschley]

No, just that the latest tend to be preferred.  

You would expect the new drivers to be more stable due to the bug fixes but occasionally there are some problems that require going back to a prior version whilst they work out the new issues.

Are the crashes with Lumion, with a certain scene or generally with other applications (3d)?

Thanks for your response. I am working with mporter on this issue. The crash never happened during a rendering. There was another application running in the background. Are there any log files that can tell us more about crashes?

[peterm]

Just for confirmation, the crash has only been occurring since the latest NVIDIA drivers were installed?

Some further questions then:
1.  Which product and version of Lumion are you using?

2.  Do you get any NVIDIA driver messages at crash time, including auto-driver recovery?

3.  Do you get any further Windows messages on hardware failure that includes logs?

You could have a look at the Windows Event Manager and scan through the error and application crash logs.

[wschley]

Just for confirmation, the crash has only been occurring since the latest NVIDIA drivers were installed?

Some further questions then:
1.  Which product and version of Lumion are you using?

2.  Do you get any NVIDIA driver messages at crash time, including auto-driver recovery?

3.  Do you get any further Windows messages on hardware failure that includes logs?

You could have a look at the Windows Event Manager and scan through the error and application crash logs.

We have been using Lumion 3.2.1 via remote desktop. So far we haven't lost anything because of the auto-save feature. Lumion did not crash during the render process. We didn't see anything significant in the Windows log files. We will now disable remote desktop and monitor the behavior to see if it happens again when people work directly on the workstation. Thank you for your help.

[peterm]

Ok, that's the problem. Remote desktop is not supported.  See Product Details and Specifications bottom of page (also relates to v3).

[johnboy]

Hey all,
I know this thread hasn't been commented on in a while but we are running 5.7.1 on Nvidia 6GB titans and there has been a lot of background and reliability issues over the past two versions of Nvidia driver updates. We are not using remote desktop and the errors about "the display driver has crashed stopped responding" has been happening very frequently. The black screening and driver crash usually happens while switching between applications say 3DS max and Lumion, and while it doesn't crash Lumion it is a right pain as it will continue come up with the same symptoms repeatedly after that.

Due to Lumion utilising gamiing engine tech and graphics I believe there will always be an intermittent experience and Geforce tries to service and keep it's largest (gaming) market happy by addressing all of the bugs within the latest game titles. Perhaps one day they will fix this...maybe

Just a shame to get high spec cards and then having to deal with constant bugs

Lumion Version 5.7.1
Video Card: Nvidia Titan 6GB
Nvidia Driver version: 353.30

[peterm]

Hi johnboy

Thanks for the information.

We have other users with similar Titan, Lumion, 3ds Max environments and have not been getting any reports of similar.  This indicates to me that it's likely to be specific to your setup. That is, either driver conflicts, resource limitations being reached due to multiple requests to same driver (from software), or perhaps a hardware problem with the card.

The hardware problem may not have surfaced in the past with older drivers, simply because of the age of the hardware or other configuration changes.  Some hardware will work well for an early period then show signs later on. 

We'll need more details about when and what tasks are being done, as well as getting you to run some hardware tests on the Titan to see if there is anything that we can help with from this side of things.

I may split this post to a new topic, but will leave here for meantime.

Note Titan drivers should now be at 355.60, so please update.  Also, please update to Lumion v5.7.2.

1. What hardware tests have been done?

2. What display driver settings are you using for 3ds Max?

3. Is it only happening in one specific scene, or also others.  If it is frequent enough, then can you test by doing the switching between Lumion and 3ds Max with the Lumion Example scene open - Villa Wegner.

Also, please confirm it also happens at other times when not rendering with HyperLight.

4. What version of Windows?

5. Are the scenes in Lumion using about 4GB or more of VRAM?  Run GPU-Z or HWInfo to see how much is being used.

6. What is your PC's Power supply wattage and Make and Model?

7. Attach to a post the Lumion DXDiag Report.

More questions    later on as needed, that's enough for the moment  .

[johnboy]

Hey Peterm,
thanks for the information and questions. There is a bit of a list so I will go through systematically and knock up a few things. Look at some of the diagnostic and error reports it may be a case of the battle of video ramm when flicking between applications, especially when Lumion is running a very heavy scene

First Up I will ensure I update to Lumion 5.7.2 and to the latest Geforce driver. I had tried a couple of times with constant crashes through the "geforce experience interface"

1. Just watching monitor using GPUZ and HWmonitor. I have reviewed and opened the details inside the windows event viewer and you can see the number off errors and occurrences which is definitely a step in the right direction of problem solving. I have attached the image


2. Using Nitrous (direct3D 11.0 Feature 11 - Nvidia Geforce GTX

3. It has happened across different scenes but not really isolated to lumion specifically from what I can tell it is more the switching between applications that leads to the error. So it is not frequent enough to work out the exact pattern of switching between A and B, but I will start to note if and when it happens to try and isolate it

4. Windows 7 64 bit pro (attached an image with more details
5. Video Ram usage - Yes it looks like this current scene is (image attached) and in the internal Lumion Monitor it shows 1926 of 1979 mb free which I know it probably problematic in itself. yes this is a heavy scene at 91 million polys. So that could be a contributing factor right off with this particular scene. I have run a few animation passes at full Full res 5 start video and it renders well and doesn't crash at that point

6. HP Genuine 1150W PSU that comes standards in the HPZ820 workstations

7. Image attached

8. One last issue I noticed it every time i restart lumion it displays my system specs as -1 which is not correct, is there a reason why it does this?

thanks for the help
John

[peterm]

Hi johnboy

Thanks for answering all those. 

..5. Video Ram usage - Yes it looks like this current scene is (image attached)...

??Do you have that to post.

..5. ... and in the internal Lumion Monitor it shows 1926 of 1979 mb free which I know it probably problematic in itself. ...

For Windows 7 there is a known issue with the Lumion Resource Monitor in that the total VRAM is incorrect.  It's the same as is reported in the DirectX Report (interesting).


The DXDiag Report indicates DirectX is only seeing 2GB of VRAM    That's of major concern (initially), it may be DirectX in Win7 that's the problem, but the 64 bit report should report up to 4GB.

What do the GPU-Z and HWInfo reports show in terms of VRAM?

If you run HWInfo using Sensors it will show two pages of data, (or if 1 click on button to expand).  On the right page scroll down to the GPU data.  Can you do a screenshot of that.  Alternatively, run HWInfo and save the complete report and attach, thanks.

Good to see a suitable Power supply .

[johnboy]

Hey Peter,
3 images attached with the screen grabs.

thanks
Johnboy

[peterm]

Thanks.

All those look good.

I've just gone back to your original comment about the driver.  Usually it's a combination or some other specific issue given the update frequency to fix bugs is good with NVIDIA.

However, in this case, and it only affects some, there's a known issue for drivers 352.86 or possibly newer.  This goes back a few months as well.  It centers around Google Chrome though.  It may be related to some Chrome Extension or Chrome or other. 

NVIDIA forum link: gpu driver crash.
The suggestion is to revert to driver 350.12 with a full uninstall.

I would assume that by now though the issue has been identified and fixed, but it's a difficult one apparently.  I have not done any more research to see if there's some other related topic that indicates a more recent fix.

This is not really the same event record as yours in switching.

I would first keep with most recent drivers, if you have any components overclocked (CPU, GPU etc) then reset to factory.  Work with for a period, log any events, and if still happening after 3 or 4 days, then look to the option of the older driver, and turning of auto-update until ? that's the big question - knowing when things will work with some new driver( again if it's the driver and not some other mix)       .

Keep us informed.

[johnboy]

Hi Peter,
here is the screen that I always see on startup, and also after the re-run.

Does this affect how Lumion is seeing my comp specs and available resources or is this just displaying incorrectly

[peterm]

Hi johnboy

I have to say I'm not sure about the Benchmark auto-displaying if the Benchmark indicates a warning, or the -1.  The -1 just indicates that card is not recognized by the system.

Overwrite these two attached files in "C:\Users\username\Documents\Lumion 5" and run Lumion again, and then run Benchmark for your system, close Lumion and re-open.  Does it still auto-display?

The Benchmark in the second report is on the very low side for a Titan, even the earlier Titans, should be around 18000.

Can you try the burnin in tests just to see if they show up anything unusual:
 GPU Shark and stress test FurMark.

And run this system report check: PassMark PC Performance tool.

[johnboy]

Hey Peter,
I have updated to 5.7.2 and the latest geforce drivers. So I will see if the stability increases from here.
Here is the latest bechmark, I had a 16500 and two at teh 15660. Is there anything else that could be holding this back?

I think I had one of the first Titan revisions out

thanks