Display stopped responding errors , and lots of info to consider.
StratDragon
Posts: 3,167
Over the weekend I noticed I had a handful of "Display stopped responding" errors which black out the screen and stop the DS application and I loose all my work.
I have a MacPro 5,1 which is the last of the legendary "cheese grater" models', but I'm running Win 7 native. I don't even have a macOS partition in this mac at this time
Im running the original ATI card that came with that model, the drivers are from Apple bootcamp I'm driving a single 27" Samsung display at 1920x1600
I'm running a 1660ti to drive a 32" Samsung display at 1600x1920 (portrait), the GPU is using single DP to HDMI cable, it's the only display hooked up to this card.
The logic board providing 75W each to both cards, and the PSU is supplying the other 75W - so 150W each card
The 1660 peaks at 130W but using GPUz to monitor I've never seen it go up that high. The ATI maxes at 103W
I'm not running much else other than DS that would stress this card, ocansionally Blender 2.8 but very basic modeling, no rendering.
I've noticed when it did happen in DS there were several variables:
1) The scenes were rather complex, like dozens, possibly 100 or more of individual textures but most of those textures are like game asset sized.
2) I added Gen8 to my library and it's still very new to me.I bought some more assets over the weekend.
3) I probably should have rebooted after each error but I did not every time, I did most of the time.
If I loaded one of these large scenes and the display state was in Iray before then the scene loaded a solid meshes and began to render in Iray I was getting that error; this was with Gen8/V8
If I did the same but the scene did not have Gen8, it had V4 it wouldnt give the error
If I was in the Iray preview mode in a simple scene, say Gen8 and some related assets it might throw a display stopped responding error when I deleted one asset and loaded another and it looked like it was about to calculate the shaders for that new asset
I did a clean reinstall of the Nvidia drivers a few weeks ago when I saw this erorr once, then not again until now all-of-the-sudden. I don't use GeForce experience
Reinstall of Studio 4.12, which still says it's Studio 4.11 on the installer.
Removal of PostGRE from Windows program install/uninstall wizard in the control panels. I don't know what Postgre is honestly but I don't think I use it. I think it adds a red banner that says "scene" on my scenes but not having isn't a show stopper I manage my own installs, and postgre had like a half a dozen instances of it running.
followed the directions here from Microsoft
https://support.microsoft.com/en-us/help/2665946/display-driver-stopped-responding-and-has-recovered-error-in-windows-7
If I switched the scenes that were crashing on load when they started to get their previews rendered in Iray to load and render in texture mode there was no problem. If I switched to Iray preview after that it was hit or miss.
my boot drive and DS are running off a Samsung SSD
I don't run 3rd party AV on my Windows box because defender is actually a pretty robust AV and I don't use this computer to go on line very much, most of the time its not even connected.
specs are below in my sig
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Which display driver is crashing? The Nvidia or the AMD? Getting bothd rivers to play well together can be tricky. If you haven't installed both drivers properly you should find a guide to doing so and follow the instructions.
Second the old cheese graters were not very well ventilated. If the crashing occurs after the render/preview has run fora while it could be an overheat problem.
Finally the old geezer could be giving up the ghost. 8 years is a long time for that HW.
Nvida driver is crashing, and while I don't dismiss the fact it is an old machine the cheese grater is very well ventilated, the fan on the cards are hardly engaging on either card. I've also tried with the side door open so to paraphrase spinal tap the only more ventilated it can get is none more ventilated. vent was never really the problem with the 5,1's, but it is a space heater.
The insides of that system are immaculate and I swapped the stock 5620 xeons for a matched pair 5770 in August so the fans were cleaned and the thermal paste is new and the SSD is a year old, the cmos battery is new and there are no other PCI cards installed. I spoke to an engineer at Apple who's on the support user forums and who I know fairly well and the nvidia/ati combo is not a unique combo to production content creators.
not ruling heating out, the display stop issue has happened a handful of times but strictly with Gen 8 in the mix and more so as I've added new content.
I suspect that it is the drivers. I'd use DDU to remove both completely and then install fresh drivers.
I'm suspect of DDU but I might try it.
the issue with the AMD is it's proprietary to Apple and the AMD drivers I found from AMD don't like that card.
there were some smoothing options for models I'm going to shut off in DS and see if it makes any difference to the strain of that card.
thanks for your help BTW
The 5770 and 5870 are not specific to Apple. They may no longer be supported by AMD though, they are 8 years old.
What are you even using it for?
these were distrib to Apple, the were customized for mac because they had the chipset necesary for EFI boot bios
without EFI boot you wont get a startup screen, you see what you're doing if you try and run internal diagnostics, you cant choose a boot disk when you start up and in some cases you won't get an aftermarket card to work by itself without a very costly flashed card or rolling the dice and possibly bricking your own.
If you won't use DDU I'm at a loss. Your best option maybe to find a cheap used/refurb PC and put the 1660ti in it. You'll almost certainly get better performance than that dinosaur.
A cheap PC is not going to work at all, so at the very least a motherboard with an intel northbridge, that supports 64GB of DDR4, a quad core hyperthread cpu somewhere in the neighborhood of 750W PSU, and a case, with better than average ventilation.
I've built scenes that surpass 32GB RAM and the GPU is long out of the picture. I'd have to do a comparison of whats that PC's trade off is at the CPU against a pair of 6 core Xeons that peak above 3GHz and run 28 threads.
At some point I'll build a dedicated PC again but until then I don't think the cheap or referb PC that will do what I need to do is going to be cost effective or that much better. I may find a buyer for the 5,1. They are still going for well over $1K with the specs I have minus the SSD and 1660ti.
Those old slow threads are pretty meaningless You're talking more than 8 years of ipc gains and 3Ghz is slower than molasses. You can get a R5 2600 with 6c/12t and a base clock 3.4 Gnz for $120 new and if you spend more you can easily get 8c/16t on the same platform or if you really spend 16c/32t. A B450 for it will cost less than $100. a $30 case, 64Gb of RAM would cost $275 nut why bother. 32 will be fine at 4150. a 750W 80+ Gold is $80. Put in the 1660ti and that will be an ok render box, far better than you have now.
All good points, but this was about the GPU freaking out, not a question about investing in a temporary upgrade. The 5,1 12/24 core actually renders very fast. large scenes at 2400x3840 about an hour or two off the CPU(s) and the GPU is long out of the picture. At some point I'll build another PC from the ground up but I'm not putting a i9 in a $30 case when I'm ready to do that.
the question goes back to what is going on with the card: is the size of the projects overwhelming the card or is there a bug in the Studio software, is there a setting I can try in the parameters I can try to remedy this. It doesn't happen all the time which would be indicative of a driver, but it seems to be when I really push this card or I'm undoing something. I haven't see this in Blender personally but I have seen threads with other users of Studio getting this error.
Running out of VRAM does not crash the driver. It's either overheating or its a driver conflict or some piece of HW is failing.
You ruled out heat and a driver conflict. That means something, likely one of the CPU's, is on its way out.
it may be content related. I ran a project that was re-rendered several times at 3822 x 2800 with some high sub-d assets I built in Bender imported in, all cpu and didn't come across any hiccups. In retrospect I used content that I had purchased before I started seeing this issue outside of some HD displacements on the figure that kicked up sub-D as well my system ran for 72 hours with no reboots, monitoring the CPU's during rendering albiet with the limited tools MS provides there were no drops or stagnation. I'm going to run HWiNFO and see if something red flags.
as I write this I may have removed CPU from the viewport devices for Iray as well.
my next step would be try this in windows 10 with the Nvidia studio drivers which are not deved for win7 along with the game ready ones.
if one of the CPU's was going out I would expect it would be affecting a more things not one specific.
Only if it was actually being used. If, say, a PCIE lane connection was going bad then the only issue would be data transmission along that single PCIE lane.
a good reason to stress test.
there was a post on the Nvidia site regarding issue with Iray preview handing off to CPU
https://devtalk.nvidia.com/default/topic/1067115/optix/attn-david-hart-all-drivers-after-430-86-are-breaking-iray-for-daz-studio-4-12/1
which opened a rabbit hole that I have not gone through entirely.
https://www.daz3d.com/forums/discussion/341006/daz-studio-pro-beta-version-4-12-1-16-updated
at some point I will drop $2K on a new build and stay home on a friday night, but used computers are a crapshoot and looking at the real time specs of the gen8/9 i5 and the xeon 5670 are very dramatic for some things but no as much as I expected for others and not enough to justify upgrade right now. CPU tech advanced in the last decade but nothing compared to GPU in the same time frame I think that's about to change. I think we are going to start seeing much more aggressive steps by the red and blue now that 'core' is the consumer trigger word.