Renders suddenly failing with error

Riggswolfe

I was rendering fine and this newest render is throwing these errors. PC reboot and/or Daz restart did not fix the issue. Drivers are up to date as far as I know.

2021-03-25 12:24:06.872 WARNING: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(359): Iray [ERROR] - POST:RENDER ::   1.0   POST   rend error: A CUDA error occurred: out of memory
2021-03-25 12:24:06.893 WARNING: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(359): Iray [WARNING] - POST:RENDER ::   1.0   POST   rend warn : Failed to allocate device memory. Postprocessing will fall back to CPU.
2021-03-25 12:24:06.895 WARNING: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(359): Iray [FATAL] - API:MEMORY :: Memory allocation failed.
2021-03-25 12:24:06.899 WARNING: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(359): Iray [ERROR] - POST:RENDER ::   1.0   POST   rend error: Exception during postprocessing: bad allocation
2021-03-25 12:24:06.904 WARNING: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(359): Iray [ERROR] - IRAY:RENDER ::   1.0   IRAY   rend error: Internal error in postprocessing pipeline.
2021-03-25 12:24:06.905 Iray Render error: Internal rendering error.
2021-03-25 12:24:06.921 WARNING: ..\..\..\..\..\src\pluginsource\DzIrayRender\dzneuraymgr.cpp(359): Iray [WARNING] - API:DATABASE ::   1.0   API    db   warn : Transaction is released without being committed or aborted. Automatically aborting.

Riggswolfe

Update: When I close Daz I'm getting this error:

DAZStudio.exe caused ACCESS_VIOLATION in module "C:\DAZ 3D\DAZStudio4\libs\iray\libneuray.dll" at 0033:0000000026FB7D71, mi_neuray_factory_deprecated()+1327793 byte(s)

Riggswolfe

I tried reinstalling Daz and get this error in DIM. I'm utterly puzzled. Could the original error have been some kind of random permissions error?

Riggswolfe

I was able to manually install Daz inside DIM simply by clicking the install button. Renders are still not working though.. It randomly just started rendering again but it took a while. I'm wondering if maybe this is a preview of hard drve problems. And then threw an error again on the very next render.

kgrosser

I am so grateful as I was already doubting my mental integrity.

Exactly what I have been experiencing recently and out of the blue. I somehow noticed my renders and viewport were kind of slowed down and when I occasinally started DAZ with my Cubix GPU Xpander that adds another 3 cards to my 1080Ti the true nature becam evident (need to add I have disabled GPU fallback). Iray viewport doesn't load in timely manner, renders start but after loading the assets stops with a black render frame with memory allocation issues and "no device" in the log - note this is on A 11GB VRAM 1080Ti card that was previously rendering slow but fine on its own. Now it doesn't do anything without the  Cubix back on. This will at least let me work but performance has definitely slowed to a crawl.

There has been no change to my system, still, I downgraded nvidia drivers, changed the PCIe slot, un- amnd reinstalled DAZ with freshly downloaded package though DIM, no avail. It seems DAZ cannot access the CUDA cores or the VRAM on a single primary card as supposed to.  Can it be some Public Beta files corrupted both Beta and Stable?

 

Any clue on this would be more than helpful.

Jeremion

same thing here, scene with 3gb > and message "A CUDA error occurred: out of memory" wich is not the case.

Richard Haseltine

Jeremion said:

same thing here, scene with 3gb > and message "A CUDA error occurred: out of memory" wich is not the case.

How do you know it is not the case?

Jeremion

There is nothing in the scene, I monitorize my memory usage, verbose and logs.

It"s pretty frustrating and appeared like recently. Don't know why. Today, for example, scene of 800 mo > failed, black render (since I deactivated CPU fallback).

PerttiA

Jeremion said:

There is nothing in the scene, I monitorize my memory usage, verbose and logs.

It"s pretty frustrating and appeared like recently. Don't know why. Today, for example, scene of 800 mo > failed, black render (since I deactivated CPU fallback).

Which GPU, how much VRAM?

Assuming that you are using W10, which reserves 1GB's to start with, DS baseload is around 250MB's of VRAM, while rendering a "work space" of 1-1.7GB's is needed and if you are using a non-RTX GPU some additional memory is needed for emulating the missing RTX functions in software. 

Edit; Using my previous tests, I would say a 8GB RTX GPU on W10 can render a scene in IRAY if the VRAM usage for geometry and textures does not exceed 3.5GiB (less with non-RTX GPU) and a 6GB RTX GPU does the same if the VRAM usage for geometry and textures does not exceed 1.5GiB.

Jeremion

Here I have a black screen with a 800 mo texture & geometry (it's little bit less that 1Gb when I start DS actually, around 1,2 Gb taken with a file opened).

RTX 2060 / 6Go. I first thought it was due to some hairs, I changed, same thing. But once again, it's new, I don't know why now :) And sometimes, for any reason, after opening and closing the app, it works. I think there is an issue with the memory management, files remains somewhere, I don't know.

ebergerly

How are you measuring VRAM usage? The Iray log won't show you the right answer. I would go into Windows 10 Task Manager, and under Performance select your RTX 2060 and look at the graph vs. time. 

When I add a base G8 character with skin textures but no hair into a blank scene and render, the VRAM usage goes up past 3GB when rendering. Note that's a total VRAM usage across all processes that are using the VRAM. Depending on what other apps you're running at the same time they could be taking some of that, including the Windows Desktop Windows Manager (dwm.exe) and other apps. Right now my RTX 2070 has about 10 apps that are all using its VRAM.

So I would watch that graph as you load your scene and do a render to see if the total gets up near 6GB. You need to be sure that some other processes aren't using a bunch of VRAM behind your back.  

Aside from the memory question, the error you noticed regarding "mi_neuray_factory_deprecated" seems strange, and the NVIDIA docs say there might be a libray/software/driver problem (although all of that could be totally irrelevant too, and just a side-effect): 

"Returns:

A pointer to an instance of the main mi::neuraylib::INeuray interface, or NULL in case of failures. Possible reasons for failures are

• allocator is not NULL,
• the valued passed for version is not supported by this library, or
• the function is called a second time. A typical cause for the second reason is that the library used at runtime does not match the headers used at compile time."

If Iray really is claiming to run out of VRAM even though Windows says there's VRAM available, it starts to point to maybe some driver/Iray problems or maybe even VRAM hardware problems, though I'm not sure how you'd even check for that (reliably). 

PerttiA

Attach your log to your post with the "Attach a file" under the "Leave a Comment" window, that way we can have a look at it.

Heckbarth

Same problem here.

128 GByte RAM
2x Titan RTX
Studio Driver v. 462.59
W10 Pro latest

 

After restarting DAZ Studio, the problem disappears for a while and at some point, this error occurs again. Happens also for smaller scenes (e.g. 5 characters).

 

Best wishes,

H.

Leonides02

1 year later, I'm also having this problem. When I try to render, I get this in the log:

"Transaction is released without being committed or aborted. Automatically aborting."

Did anyone ever solve?

lana_lass

Leonides02 said:

1 year later, I'm also having this problem. When I try to render, I get this in the log:

"Transaction is released without being committed or aborted. Automatically aborting."

Did anyone ever solve?

 

Hey!

 

I also had this issue, and discovered that it was down to the fact I was using detail morphs that were bumping my render subD minumum level up. So, I had my subD level set at 4 but a morph was then subsequently increasing the render subD to 5. This render SubD level was too high and thus causing the bork.

 

I reset the SubD to 1 and then turned it back up to 4, making sure both the subD and the render subD minimum were at 4... and hey presto! Rendered in 15 minutes.

 

Might not be the solve for everyone, but just in case it helps someone. 

PrefoX

lana_lass said:

Leonides02 said:

1 year later, I'm also having this problem. When I try to render, I get this in the log:

"Transaction is released without being committed or aborted. Automatically aborting."

Did anyone ever solve?

 

Hey!

 

I also had this issue, and discovered that it was down to the fact I was using detail morphs that were bumping my render subD minumum level up. So, I had my subD level set at 4 but a morph was then subsequently increasing the render subD to 5. This render SubD level was too high and thus causing the bork.

 

I reset the SubD to 1 and then turned it back up to 4, making sure both the subD and the render subD minimum were at 4... and hey presto! Rendered in 15 minutes.

 

Might not be the solve for everyone, but just in case it helps someone. 

thanks for that intel,

I have that problem popping up just now and reducing the subd level to 0 fixed the issue. stll its crashing with lvl 1 (on some products) but it must be a DAZ issue since it started with 4.21