Daz3D stops using GPU
something12weird12
Posts: 0
I'm having this problem for few months now, DAZ simpley stops using GPU and changes to CPU.
I tried many sollutions that are suggested:
1. Under rendering tab - Advanced, Hardware is set to 2080ti only
2. Under rendering tab - Advanced, OptiX Price Acc. i tried turn off and on
3. Changing PsyhX in Nvidia control panel to CPU only
4. Changing Instancing optimization to Memory and Speed
5. Turning off every aplication and letting only DAZ run
Maybe few more but i can't remeber now.
Problem happends both in the most simple scenes (only one genesis model in the scene) and more larger scenes. It doesn't matter.
Only sollution that kinda works, is tuning off DAZ or reseting PC. But then it falls back only to rendering with CPU after few renders and sometimes at like 91% during a render.
Batch rendering over night and animation are almost imsposisble to do this way. Sometimes it runs okay the whole night, sometimes it starts using CPU only at second render. It's random.
AMD 2600, 32GB DDR4, 2080ti, 1200W PSU, 500gb SSD (250gb free)
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
I had that issue a lot when I was running win10. It does something funky with the handling VRAM, when I switched to win7 the problem stopped. If the render is able to start, it will go until it's done or I tell it to stop. Back before I got a topend cooler for my CPU, that crap was downright dangerous, couldn't ever let it render while I couldn't keep an eye on it.
I actually came in here to post my findings about this, but I guess replying to this will be just as good. I've had this problem with my GTX 1070 since I started experimenting with DAZ3D last month; Renders would start on the GPU, then fail randomly and revert to CPU only. Sometimes they'd fail immediately, other times they'd work away a few seconds then fail. The logs reported every kind of isse from "failed to initialize" to invalid operations when "deallocating memory".
I've spent hours looking around the web for solutions, information or just someone who would know what the heck was going on. The issue doesn't seem to exclusively affect DAZ3D, but pretty much anything that uses Iray GPU rendering.
Like yourself, I kept banging my head into canned replies, blaming anything from the size of the scene to the drivers to the PSU being underpowered to WIndows 10 doing weird things with the VRAM. None of these are remotely applicable, so I decided to conduct some research on my own, taking advantage of the fact I also have a laptop with an nVidia GPU (1050 Ti).
Hold and behold, on the laptop GPU rendering worked flawlessly, using the very same version of the drivers, Windows and DAZ3D. The issue HAD to be hardware.
And it was - a stupendously sneaky and easy to fix one, actually. Basically, most nVidia cards tend to "boost" their GPU and VRAM clocks as much as they can within the thermal envelope; My card is a Gigabyte G1 and its Aorus software has various "modes" - "Silent", "Gaming", "OC" and a "User" mode. I've set the card up a long time ago and never gave it any thought - I had it in "Gaming" mode; Because I live in a relatively cool climate, the temperature of the card never exceeds 60c, and it was boosting all the way up to 2088 Mhz, from the 1785 the "Gaming" mode should have been at.
For some reasons only nVidia will know, the card works flawlessly in games at that boost, for hours with no crashes nor glitches whatsoever, but CUDA freaks out and starts incurring into errors. Switched the card to a smaller boost (something like 2013 Mhz maximum) and guess what? IRAY doesn't crash anymore. It literally took me almost a month of research to figure this out, and I guess it makes a great case for why Quadro GPUs actually exist :)
TL;DR - check your GPU and try switching it to a lower performance mode, as many manufacturers allow their cards to boost aggressively - which works great with games, but CUDA apparently hates, at least when dealing with IRAY!
No joke, after recent update to the BETA, my renders have been heavier AND sometimes they don't even render but a glitchy bottom of the image.
I apply the same small overclock in 7 that I did in 10. It seems dumb you have to clock down your cards to work properly, the reason people spend so much on these things is to use more power, not less.
Ohhh, you are talking about DAZ Studio. I was shocked at first.
"WHAAAT, DAZ STOPS GPU RENDERING????"
Puuuh, only your software stopped running it.
+1
Daz 3D have a few products, and I presumed they meant Studio, but it is always good to avoid confusion.
Yeah, @something12weird12 based on your descriptions of both what's wrong and what has failed to fix it, it sounds to me like this is either GPU core or VRAM instability - most likely stemming from excessive boosting with perhaps not the greatest cooling solution on the card.
The fix @h3llr4iser_e353fcd692 described wasn't a case of underclocking. It was still overclocking - just overclocknig less. Oftentimes you can get away with much edgier performance specs for gaming workloads than 3D rendering ones because the former tend to be intermittent in intensity. 3D rendering is a constant load.
!hank you to everyone for replys! But this suggestions worked (knock on wood)
I don't want to jinx it, but the animation is rendering now for 10 hours and using the GPU.
I "downclocked" the GPU from 1850 Mhz to 1600 and its working fine for now.
It's not clocking down, it's just limiting the automatic boost the cards apply - as I said, in my case the card goes pretty much as high as these 1070 GPU can go due to the fact the temperatures never climb particularly high, thanks to living in northern Europe where ambient temperature hardly ever climbs above 20c, even in the middle of the summer.
That's my take as well - actual graphics load in games is not constant, although a good case can be made by benchmarks and stress test tools, like 3DMark or Heaven for attempting to generate a more extreme load.
As I said, I had this card for about two years, it's been used extensively both in actual games and in benchmarks, and it never showed any issues, except for GPU rendering under DAZ Studio's IRAY.
My pure speculation is that IRAY must also be doing something "under the hood" that is way more susceptible to small processing / memory errors than gaming use is. The central point is that these cards weren't really designed as studio quality rendering rigs, their main job spec is to render realtime animation of an acceptable quality at 60+ fps. They can do it decently enough for the hobbyist user like ourselves, with the occasional headache.
That might be a bit extreme in terms of lowering, although I don't have experience with the 2080 (yet, might get one soonish). Glad it worked for you, and hope this thread can be usefult for others as well - from what I can tell reading the various posts around the 'net, I'd say 75%-80% of the people having this "rendering drops to CPU only" are experiencing the boost-related issue.
An important thing to know is that at NO POINT the temparature of the card, at least in my case, climbed above 60c, so it's not like there's a clear "ah, I see the problem!" situation.
I noticed this for the first time (at least the first time in a long time) yesterday. Small scene, one character, clothes & hair, I realised that it was taking a long time to do a test render (normally takes just a couple of minutes on my 8GB GTX 1070). I couldn't see any problem in the error log but it was rendering in CPU mode. I tried hiding the hair and the dress but it didn't help. Eventually, I shut down DAZ Studio and then loaded the scene again - all fine. I suspected it was the old problem of not releasing VRAM between renders (I had been trying different IRay shaders). The fresh scene rendered in GPU mode in 2.5 minutes.
By the way, this was 4.11 general release. I did update the NVidia drivers recently when I downloaded the 4.12 Beta but I wasn't running the beta yesterday. I did run GPU-Z while after re-loading the scene and my GPU temperature was hovering around 55 deg C. I was not a warm day here.
Is there any other sollutions for this problem?
2080TI fails for me and drops to CPU. I tried all different studio and game drivers (from 430.86 to 442 the newest)
Updated Daz3d to 4.12
Changed the GPU and MEM clock in MSI Afterburner to the lowest, and the power unit to 80% (and many between)
PhysX to CPU
Virtual Memory to 4, 8, 12, 24, 36, 40, 48 gb....
Daz3d defulat settings (no denoiser etc) CPU is unchecked and optix, tried with checked as well and many other settings
Dunno what to do anymore. I even tryed to get a replaced GPU but the store says its not faulty. All the test are running fine (max3, furmark etc)
Problem seem to be only with Daz3d and IRAY and this is the ONLY thing i need from this super expensive hardware
Please help!
Good God! Thank you man! This was driving me insane. I was two weeks into trying to figure this out and that finally worked!
You are a Demigod amongst mortals, man. :D This is actually works! I was messing around this issue since the 4.11 build of daz came out and spent many nights trying to find a solution why i cant do more than 3 renders in row, as the live renderer just stopped working and fell back to the CPU. Altho it boggles the mind that the solution was under my nose all along, i would honestly never suspected that tuning this thing down to 1500 Mhz would actually make such a huge difference. It seems the trick is that the card shouldnt go higher than 2000 so getting it down just makes the difference. So thank you.
Would like to try this with MSI Afterburner but I'm pretty new to the whole underclocking/overclocking business, which settings to toggle here?
tried the under-clocking solution with MSI afterburner, but render after render, can't manage to use GPU instead of ram and cpu.
GeForce GTX 1060 6gh is iddle while RAM usage boosts to 100% and the 8 CPU's cores go to 100%, letting the pc unusable for about nine hours for a sinple render.
Total desperation...
I am also having this issue trying to render an extremely basic scene with a single Genesis 8 model and simple interior. GPU craps out around 200 frames in, and can't be used again until I restart Daz, even if I clear the entire scene. Afterburner said my GPU clock speed was ~1900 Mhz during the render when it stopped working. So I decreased both the clock speed and memory speed in afterburner by the maximum allowable amount (502 Mhz), and afterburner said my clock speed was now sub 1400 Mhz during render time. Still no dice, GPU stopped working.
I have the RTX 2060 S with 8gb of Vram, and I've been able to render far more detailed scenes with multiple characters in the past, so I assume this is not an issue of lack of memory. I'm hoping this is not a sign of a dying GPU, as I purchased it just under a year ago, and my temps rarely exceed 65°C.
If anyone has alternative suggestions other than underclocking my GPU, which didn't seem to work for me, I'd greatly appreciate it
EDIT: I discovered that my issue is in fact not related to GPU clock speed and is actually an issue with Iray section planes causing the GPU to fail. Specifically, I was trying to use the X-ray camera included in this HDRI pack, though I suspect it is not a problem with the product (which I have nothing but praise for), but rather a deeper issue with Daz and Iray itself. If I don't use the x-ray camera, my scene renders perfectly fine. I don't know what specifically the section planes are doing that cause my GPU to fail, but I intend to investigate further and find out.