Display/memory issues with s8 & nvidia?

[wkearney99]

Anyone else had trouble with the desktop display going crazy when S8 has been left running for several hours? I've got an i7 machine running 2008R2 with 16gb of RAM. If I leave S8 running for more than a few hours, with a model loaded, the display starts acting up. Most noticeably if I have windows mail running the text windows and toolbar in it will start redrawing wrong, eventually going black. Left alone long enough, like overnight, the whole desktop display becomes unstable and unresponsive. I can usually bring it back to life by making a remote desktop connection in the machine from another workstation and killing either the s8 or the winmail process. This returns the machine to normal. I can reload both s8 and winmail and the problem isn't there. Leave it all running for a few hours and it crops back up again.

When it happens the task manager still shows better than 6gb free and neither s8 nor winmail appear to be consuming an excessive amount of memory (both can be pigs). I'm wondering if something is running amok with the video card driver and corrupting how all of the windows are drawn.

I can definitely say it's related to s8 because I've not seen the problem on this machine for quite a while since I've not been using s8 on it. It's hard to reproduce because it definitely seems to be time-related, not activity.

Has anyone else seen this sort of thing? Fixes?

[thomthom]

Never seen that. I've used nVidia cards for many years now with SketchUp. (Tried ATI once, but had too many issues.) I run multiple windows for the entire day.

Have you ensured your drivers are up to date?
Is your computer getting enough cooling? It's not overheating?

[wkearney99]

@thomthom said:

Never seen that. I've used nVidia cards for many years now with SketchUp. (Tried ATI once, but had too many issues.) I run multiple windows for the entire day.

Have you ensured your drivers are up to date?
Is your computer getting enough cooling? It's not overheating?

It's a rack-mounted machine in a chilled room, and it's fans are working properly, as is the one on the video card itself. The machines I use from my desk are all in a rack that's in the room next door (to eliminate their noise and heat). I've got a long DVI cable bringing the video to my desk.

That and the problem doesn't come back for quite a while if both mail and su8 are quit and reloaded. Were it heat-related it would come back much faster, if not immediately.

It is hard to replicate in that it involves letting the machine 'sit there'. The machine is not configured to use any sleep or power-reduction modes, so that's not it either. The drivers are current nvidia-supplied drivers (not MS nor card vendor). I've had the same problem crop up on a different machine, also using an nvidia card, su8 and mail.

It's not like I'm losing data, it's just a visual disruption, one that requires stopping work and reloading. Not a terrible burden on today's machinery, just inconvenient.

I'm sure there's a combination of factors coming into play here. I'm looking to find if others have seen the problem and possibly found solutions.

[mac1]

Shorten up your dvi cable length to <= 16 ft and see if that helps or better yet move your monitor over close.

[utiler]

Actually, the other guy in the office has a GTX 570 and has been having issues for the last few days.... I updated the driver and seems to be going ok since,
seems its paramount to keep them up to date.

[mac1]

Drivers up to date YES;
Cable length <= spec should be

[wkearney99]

@utiler said:

Actually, the other guy in the office has a GTX 570 and has been having issues for the last few days.... I updated the driver and seems to be going ok since,
seems its paramount to keep them up to date.

I was already running the latest release version (306.97) of the driver. Did he upgrade to the 310.54 beta version?

[mod=:vwh9xj5p]Moved from SketchUp Bugs as the issue apparently does not come from SketchUp itself.[/mod:vwh9xj5p]

[wkearney99]

@mac1 said:

Shorten up your dvi cable length to <= 16 ft and see if that helps or better yet move your monitor over close.

And given that it's not a display artifact issue, just exactly how do you think that would help?

Or is that just another example of the 'have you rebooted?' approach to customer service we all know and loathe? Don't get me wrong, it's good to have advice, but it does help more if it's actually relevant to the problem at hand.

Were I seeing pixelization or noise, sure. But I'm not and didn't describe it as such. Had I described it as 'snow' or other sorts of noise then cabling could have been an issue. But then how a digital HDMI/DVI signal reacts to bad cabling is ENTIRELY different than how VGA would.

It's a matter of the windows and their contents being drawn incorrectly. It appears to be a memory issue, in that the OS/driver is not correctly drawing everything. Clearly either mail or su8 is pushing past some limit of what the OS/driver is prepared to handle properly and the end result is wildly mis-drawn elements.

The solution will probably just be a 'Henny Youngman'. As in, "Doc, it hurts when I do this." and the doctor responds with "Well, don't do that."

But if someone else that's had similar problems with su8 on an nvidia card cares to share their actual experiences I'd be glad to hear of it.

[wkearney99]

Moved? Why? It is a bug, one that only comes up when su8 is running. Now, it may be that su8 is either defective in how it allocates display resources, or it might be the underlying driver. But without better evidence regarding the driver, and seeing su8 as the only repeatable cause, how is this not suitable as a post in the su bugs forum?

Please, moderation works best when it's sensible, not when it's trying to hide something.

[Gaieus]

No-one is trying to "hide" anything. Clearly it seems that not SU itself is the culprit but something in your system. Most of the folks are trying to be helpful when suggesting things but it seems nothing from "inside SketchUp" helps here so you need to discover the environment it is running in better.

If it turns out to really be a SketchUp bug, I will be more than happy to move the topic back and even report the issue through other channels.

[mac1]

wkearney99
One of the first things I look for when evolved in trouble shooting of spacecraft or hardware failures is non standard configuration;
What jumped out at me in your posting was what I commented on.
Your comments show a complete lack of under standing of high speed interfaces, common mode rejection and EMC problems one can run into with widely separated interfaces. I have seen cases like this where switch to ground of even short cable lenghts with a low voltage causes the circuit to act like a pulse forming network( because of reflections) and the signal will propagate through several circuits until the impedance loading slows it down enough that it will then trigger circuits down stream form the actual cause. You obviously don't know Teflon cables go through a state change at room temp and the phase velocity can change drastically and is a function of the processes they are built to; balanced circuits can operate properly even with one side open and will change performance with small changes in temp.
I am not saying these are the root cause of your problem but it just a simple suggestion to eliminate a variable.

I would do it ,but unlike you I am not omniscient and don't know all the possible causes of problems so will defer to your sophomoric posting and wish you luck.

BTW you cannot assume because SU8 shows the problem that it is the cause. I just spent a week trouble shooting a problem when I switched from FireFox 3.2.6.8 to the new 16.02. When using the vertical scroll bar it would cause a OS kernel blue screen for some web sites but IE would work ok. It was traced to a Nvidia driver I had not updated.

[wkearney99]

@mac1 said:

wkearney99
One of the first things I look for when evolved in trouble shooting of spacecraft or hardware failures is non standard configuration;
What jumped out at me in your posting was what I commented on.
Your comments show a complete lack of under standing of high speed interfaces, common mode rejection and EMC problems one can run into with widely separated interfaces. I have seen cases like this where switch to ground of even short cable lenghts with a low voltage causes the circuit to act like a pulse forming network( because of reflections) and the signal will propagate through several circuits until the impedance loading slows it down enough that it will then trigger circuits down stream form the actual cause. You obviously don't know Teflon cables go through a state change at room temp and the phase velocity can change drastically and is a function of the processes they are built to; balanced circuits can operate properly even with one side open and will change performance with small changes in temp.
I am not saying these are the root cause of your problem but it just a simple suggestion to eliminate a variable.

I would do it ,but unlike you I am not omniscient and don't know all the possible causes of problems so will defer to your sophomoric posting and wish you luck.

BTW you cannot assume because SU8 shows the problem that it is the cause. I just spent a week trouble shooting a problem when I switched from FireFox 3.2.6.8 to the new 16.02. When using the vertical scroll bar it would cause a OS kernel blue screen for some web sites but IE would work ok. It was traced to a Nvidia driver I had not updated.

I've over 30 years experience in the IT field, you'd do well to avoid assumptions, or insults. And re-read what you wrote, I'm sure there's something sensible you're trying to include, but it doesn't read very clearly.

I'm well acquainted with how high-speed interfaces function. I've even seen how cabling can affect HDMI/DVI signalling. This is not one of those situations.

[TIG]

So... at the risk of injecting some sense into this extended diatribe...
It's almost certainly not a 'SketchUp bug' - that is, something that is inherently wrong with SketchUp...
Many users of v8, high-end PCs and NVIDIA cards etc have no comparable issues.

So what is specific to your setup that might be implicated in the issues?
Cable length could be.
Unless using a much shorter cable has been shown to still produce the issues, then the current cable length is still a suspect.
The card's drivers might also be suspect. The latest driver might have some unexpected side effects; have you tried rolling back to a previous one? If so are these issues still the same ?
If this might be an NVIDIA card/driver bug then what have they said on their forums?

As the issue seems to occur when v8 and another app are loaded in tandem does the unloading of one of them relieve the issue ?
What has been said in the other app's forum ?

Does this issue occur on all PCs in your organization that use the same card/drivers/v8 etc. If not, then perhaps it's a specific System / configuration issue ?

Issues like this are difficult to track down, and are often the combination of several overlapping minor problems, seemingly unrelated at first sight.
They can also manifest themselves in other totally innocent applications.
This doesn't mean that SketchUp v8 is innocent, but it certainly doesn't demonstrate that it's a 'SketchUp bug' either.

The guys here have no reason to 'hide' anything. SCF is independent of SketchUp, those offering advice do so freely and in good heart. Falling into side arguments about this being a bug or not, or trading insults about whose got the bigger dick better experience, doesn't advance a resolution of this issue on iota.
Get back on track please...