New RTX 2060 vs. Daz....everyone looses
So I upgraded to a GeForce RTX 2060 about three four months ago. These are the system stats:
Dell XPS Studio 9100
i7 x990 @3.47 - 22BG ram
GeForce RTX 2060 6gb
Win 10 Pro/64bit
Studio Drivers - version 442.92
Daz Studio version 4.12.0.86
When rendering a scene, ocassionally something will go horribly wrong and somehow cause a cascade failure of nearly every component in my system. I've test rednered simple scenes and complex scenes and there doesn't seem to by any rhyme or reason to the crashes.It starts at any point after a render begind when the clock in the task bar stops updating. Shortly thereafter, the Daz interface will freeze, but not fade as apps normally do. I run dual displays, and the screen that Daz is living on has no other symptoms than large area refreshes don't seem to be working. That is, I can't move a window on that display, but i can still click through the menus and other buttons, but to no effect.
On the opposite screen, I can open and select items in the start menu but nothing starts or stops. Selecting shut down does nothing. Starting the task manager lets me select and force stop Daz, but the force stop never goes through. Eventually the task mager also stops, but the mouse conitnues to work more or less for a few minutes until it too ceases. During these episodes I've tried to log in ot the computer remotely usin anydesk and although I can get past the password screen, the app hangs on 'waiting for display' Eventually I have to force restart.
Whis is when it gets intersting. The Daz log file has no errors at that time and the last entries are:
2020-06-19 15:09:32.503 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : Received update to 00368 iterations after 1305.893s.
2020-06-19 15:11:08.917 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : Received update to 00386 iterations after 1402.306s.
2020-06-19 15:11:45.071 Iray [VERBOSE] - IRAY:RENDER :: 1.0 IRAY rend progr: 37.93% of image converged
2020-06-19 15:11:45.681 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : Received update to 00401 iterations after 1439.070s.
2020-06-19 15:13:28.914 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : Received update to 00422 iterations after 1542.303s.
2020-06-19 15:14:04.421 Iray [VERBOSE] - IRAY:RENDER :: 1.0 IRAY rend progr: 39.12% of image converged
2020-06-19 15:14:05.018 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : Received update to 00438 iterations after 1578.407s.
2020-06-19 15:15:58.721 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : Received update to 00461 iterations after 1692.110s.
2020-06-19 15:16:43.097 Iray [VERBOSE] - IRAY:RENDER :: 1.0 IRAY rend progr: 41.33% of image converged
2020-06-19 15:16:43.716 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : Received update to 00479 iterations after 1737.105s.
2020-06-19 15:18:44.290 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : Received update to 00504 iterations after 1857.680s.
2020-06-19 15:19:29.980 Iray [VERBOSE] - IRAY:RENDER :: 1.0 IRAY rend progr: 43.90% of image converged
2020-06-19 15:19:30.604 Iray [INFO] - IRAY:RENDER :: 1.0 IRAY rend info : Received update to 00523 iterations after 1903.993s.
The last entry there is two minutes before the task bar clock stops.
There is a huge 50 line break in the log and it picks up where after I've reset everything and start Daz up again:
020-06-19 16:43:21.468 Platform bits: 64
2020-06-19 16:43:21.468 Qt Version: 4.8.7
2020-06-19 16:43:21.468 OpenSubdiv Version: 3.0.0
2020-06-19 16:43:21.468 Running on Windows 10, Build 9200, No Service Pack Installed
2020-06-19 16:43:21.468 CPU Information:
2020-06-19 16:43:21.468 CPU String: GenuineIntel
2020-06-19 16:43:21.468 CPU Brand String: Intel(R) Core(TM) i7 CPU X 990 @ 3.47GHz
2020-06-19 16:43:21.468 Cache Line Size = 64
2020-06-19 16:43:21.481 L2 Associativity = 6
2020-06-19 16:43:21.481 Cache Size = 256
2020-06-19 16:43:21.481 Stepping ID = 2
2020-06-19 16:43:21.481 Model = 12
2020-06-19 16:43:21.481 Family = 6
2020-06-19 16:43:21.481 Extended model = 2
2020-06-19 16:43:21.481 CLFLUSH cache line size = 64
2020-06-19 16:43:21.482 APIC Physical ID = 2
So I go into the System Eventviewer:
Error 6/19/2020 4:42:22 PM EventLog 6008 None
Information 6/19/2020 4:42:06 PM Kernel-General 20 (6)
Information 6/19/2020 4:42:06 PM Kernel-Boot 32 (58)
Information 6/19/2020 4:42:06 PM Kernel-Boot 18 (57)
Information 6/19/2020 4:42:06 PM Kernel-Boot 27 (33)
Information 6/19/2020 4:42:06 PM Kernel-Boot 25 (32)
Information 6/19/2020 4:42:06 PM Kernel-Boot 20 (31)
Information 6/19/2020 4:42:06 PM Kernel-Boot 153 (62)
Information 6/19/2020 4:42:06 PM Kernel-General 12 (1)
Information 6/19/2020 2:03:09 PM Service Control Manager 7040 None
Information 6/19/2020 2:00:45 PM Service Control Manager 7040 None
Warning 6/19/2020 1:28:04 PM DistributedCOM 10016 None
The eventviewer complely misses the entire issue. Task bar clock stopped at 3:21, and the event viewer skips from 2:03pm (approximately when I started the render) to 4:42 (when the system started the report.
I've tried various configurations of drivers, Studio, Gaming, older, newer, and the problem only happens on random iRay renders. Games, video editing, streaming video, or anything else I've put the card it doesn't miss a beat.
Anyone heard of anything like this?? I'm close to returning it to the vendor and requesting a replacement.
Comments
Have you tried rendering it via CPU mode only to eliminate the nVidia GPU iRay rendering as the trigger to the problem? (which is probably triggered by DAZ Studio iRay rendering triggering a bug in the nVdia drivers or even the GPU card internally).
6 GB of video ram is probably the problem.
DAZ will crash if it loses headroom.
If the scene is too big even though it says it is only a couple gb of data it will freeze the mouse. If you wait a long time it will usually unfreeze. But the scene will still be to slow to edit.
This is a sign the scene is too big.
If you return the graphics card, get one with 11 gb of graphics ram. Get it refurbished for half price on Amazon.
I got a 1080ti w/11gb for $500.
I've got a GTX 1060 with 6 GB and haven't had Daz Studio crash because of too large of a scene. In fact, it has only crashed a couple of times on me and that was usually when I tried to run a dforce simulation.
The first thing that I think of when I hear something like this is a faulting memory stick. There are many programs out there - some commercial and some free - that will exhaustively test your memory for any problems. It typically requires setting up a USB boot drive that it can boot off of so that it has access to all physical memory in the machine. Even if you're only seeing this when you're using Daz Studio because Daz Studio consumes a lot more memory than you're typical applications. It's likely a spot in the higher address range which is what happened to me. I replaced the bad stick and have been fine since.
The second thing that comes to mind is that it could be a timing issue if you've tweaked any of the BIOS settings or you have your machine overclocked. This is another scenario where the computer will exhibit strange behaviors and hang ups.
The third thing that comes to mind is that something could be overheating. Daz Studio does tend to stress the PC a lot more than other applications.
Based on the symptoms the render is dropping to CPU and that ancient CPU can't take the strain.
Start the render with tack manager already open andwatch the CPU usage. If that pegs when the system becomes unresponsive that is the problem.
With 10 year old HW though almost anything involved could be physically failing. I would also make sure the system isn't overheating. Are all the fans turning? Is the box dustfree?
Given the symptoms mostly involve the displlay, is it possible that something is causing the PCIE interface to clog or jam in some way?
It's been almost 10 years since I touched an X58 mobo but if memory serves they had a mix of PCIE gen 1 and 2 slots. Gen 1 slots are pretty slow but that shouldn't matter when the render is loaded on the GPU the data getting passed back to the CPU is pretty minimal.
What ever it is is taking Windows out, my first thought would be the Nvidia drivers but you've already got the latest studio driver as far as I can see. First step would be to look at the motherboard bios version (Dell?) and update it if required followed by updating the WIndows drivers.
Good luck
Steve.
Since I CPU render all the time I agree it's a combo of RexRed & Rishard Haseltine's comments and that it's because the CPU becomes I/O bound on complex DAZ iRay CPU renders. For me, it used to be a huge problem with only an intel i7 with 4 cores & 8 threads but now with an AMD Ryzen 7 2700 with 8 cores and 16 threads it very rarely is a problem for me anymore.
That DAZ crashes is likely because once your scene falls out of the GPU due to lack of GPU RAM to be CPU rendered apparently you lack enough CPU RAM too and the Windows 10 system is swapping and growing system swap maddly trying to get the requests from DAZ Studio taken care of. Windows 10 used to purposely abort DAZ Studio once the available system RAM sunk below a certain point toi avoid crashing all of Windows 10 but it seems they had stopped that behaviour mostly. Windows 10 will though premptively abort processes that behave like trojans in requesting excessive amount of RAM or other system resources. I used to have when CPU rendering on Windows 10 on my i7 laptop with 16GB system RAM grow the system swap on my 223GB SSD system disk from less than a gig to over 40GB sometimes on complex scenes: basically growing swap until I would run out of diskspace. Now that I have a new desktop with 32GB RAM and a 2TB SSD that is not a problem anymore.
Wow! I step away for the night and get a bunch of answers! Thanks for the responses everyone!
nonesuch00s idea seems to most closely decribe what I think is going on, in that it might be dropping to CPU. I'll keep an eye on my card's RAM usage, but what bakes my noodle is that it'll sometimes crash when I have a very cimple scene, say a single character with a single poly BG and HDRI light dome. But let's assume that it is dumping to CPU for some reason...how can I find out if it is?
Open task manager before you start a render. Watch the CPU usage. If it suddenlyhits 99% and stays there then the render is on CPU.
This may not be your problem, but it is worth ruling out to be sure. You upgraded your card ... can your power supply handle what the computer and the card require together? An on-edge or under-level power supply can cause many "mysterious" things to happen.
Still trying to figure out what's being loosed...
This is a really good point. If the power supply was not upgraded, then that is most likely the culprit, especially if everything works fine untill a load is placed on the system.