4.11 Iray Render Times
bryanll1
Posts: 141
Has anyone else noticed their Iray render times doubling after upgrading to 4.11?
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
Yes.
No. Mine are a bit faster.
I'll be very interested to see what's causing this. And speaking of benchmarks, what would be a REALLY useful attempt at benchmarking right now is to find a scene that renders faster for some, and have the slower folks like myself give it a try, and try to figure out what's the difference.
Unless DAZ already knows what's going on...I think someone mentioned there might have been a 4.10 bug where not all rays were calculated, but that changed in 4.11. But that doesn't make much sense IMO since some are getting faster times with 4.11.
To be fair I believe there's no point comparing speeds. I mean, iray is evolving and we have anyway to upgrade to get the new features. Unless someone wants to stay with the old content that's perfectly doable anyway. I even see of people still using daz 3 with V4.
Overall if you use the denoiser 4.11 is faster for sure.
While I agree it's somewhat ridiculous to even consider benchmarking RTX performance at this point, that's not the issue. The issue is whether there is some unresolved issue with 4.11 which causes some to have much longer render times than previously. If it's a 4.11 issue then a bit of comparing of render times between users might help to troubleshoot. Or it may be related to scene composition, or some differences in hardware. But until/unless we do a bit of investigating, we may never know. Unless DAZ already knows.
Personally, I've found that the Sickleyield scene used to render in about 1 minute on my system. I tried with 4.11 and (forgot to write down the answer), but I think it was under 1.5 minutes. That's a big difference. That kind of difference needs to be figured out, independent of there being de-noising available. People spend a lot of money on GPU's to get those kinds of performance differences.
BTW, it just occurred to me...I'm wondering if the 4.11 automatically turns off Optix Prime (no longer supported in favor of Optix??), and maybe 4.10 didn't? Might explain some of this.
Is there any basis for the speculation that Iray in any version of DS uses Optix? I was under the impression that Optix was GPU only, and only Optix Prime could fall back to CPU.
- Greg
ETA: Was on my phone earlier, but dug this up now that I'm on a computer:
https://devtalk.nvidia.com/default/topic/1039223/-optix-optix-prime-compatibility-with-cpu-and-rtx/
Can't find the OptiX_Programming_Guide_5.1.0 that was referenced, though, and it's not the current version anyway.
Definitly. And I have a trouble viewing my projects in IRAY. Program keeps blowing up.
Filed a ticket, but so far no answer fro the Daz teams.
Only based on the knowledge that Iray is in the process of being RTX-enabled, and Optix is the NVIDIA renderer associated with RTX, and RTX and Iray and Optix are all NVIDIA, and Studio now uses Optix Prime (kind of a subset of Optix AFAIK).
So yeah the "...in favor of Optix" part could be wrong, but the rest of the question is still a question.
BTW, since you were wondering Optix Prime definitely was not not turned off in 4.11.0.383. Check your log for the following if you have it enabled:
2019-07-17 16:37:54.528 Iray INFO - module:category(IRAY:RENDER): 1.0 IRAY rend info : Using OptiX Prime ray tracing (5.0.1).
- Greg
There was a bug in 4.10, that was fixed. Iray bug, not Daz implementation. Although it rendered faster, it could be inacurate.
Yeah, I heard that, but I'm very skeptical. If 4.10 was faster due to a bug, why are some reporting that 4.11 is faster?
And I find it hard to believe that it was off by so much. I'm seeing MUCH longer render times in 4.11, like 30-50% longer. That's a big bug.
What is really interesting is that now both IRay- and 3DL users are experiencing longer rendertimes.
What is this and were and how to use it?
Hi greg must it be checked or unckecked?
I just meant it will only show up in the log if you have it enabled (checked). Whether or not you want to enable it is another question.
- Greg
ETA: Check this thread for more info:
https://www.daz3d.com/forums/discussion/284786/optix-use-it-or-not
Thanks a lot
Beta Studio 11 was slower for me on my 980 Ti, haven't figured out how to download the latest release onto a USB device so that I can try it with my new 1080 Ti.
Anybody know how I can do that?
To avoid cnfusion, you mean one of the 4.11.x.x builds rather than Studio 11.
Have you tried clicking the Download and Install button? I can't test, but it's possible that on a system without DIM installed that may pop up a what do you want to do with this prompt and you can then save it to the USB, then when you get hom pop the files in the Downloads folder for DIM.
Yes, I have noticed this too. It was somehow faster in the beta version, and could handle heavy graphic scenes better.
Now I have crashes in both when trying to render or switch to Iray display, even after switching to interactive mode.
Assuming you are comparing with the 4.1.0.383 beta they are identical, and I don't think iray chnaged over the course of the betas. Differences between beta and general release are likely to be down to some settings.
I never used the beta, and went from 4.10 to 4.11, so my comparison is between those two.
Unforunately the challenge here is I expect few people have saved render time numbers from previous Studio versions to compare with recent versions. And since 4.10 isn't easily available, and too much of a pain to go back and forth between versions, I think we'll have to leave it to DAZ to figure this one out. I've got 4.9 on an older backup computer, but with a different GPU, so I can't do apples-to-apples comparisons with my 4.11 machine with a dual GPU configuration.
I'm wondering if some default render setting got changed with the new 4.11. I thought maybe something like Optix getting enabled/disabled between versions might account for a big difference, and maybe pass unnoticed for users. Who knows...
Doesn't make it a bug, just annoying.
If you can use the denoiser in a scene, it can be incredibly fast; I find it useless when there are characters in it, especially if they are the main part or focus. The only time characters can be acceptable is if they are heavily stylised.
... Having said that though, I saw more details when comparing the two versions.
Well for an "issue" that results in a 50% increase in render times, in a world where users who will gladly spend around $1,000 to get that kind of decrease in render times, I think this is a bit more than annoying. It sure sounds like a big bug to me, and presumably, I'd think, to the vast majority of users. I'm really surprised folks aren't bouncing up and down on this issue.
And the fact that denoising might help for some is irrelevant. It's a bit like saying, "hey, suddenly my car is running real slow. But if I put up a sail on a windy day it will go fast again".
Y'know I'm wondering if it has to do with problems dealing with multiple GPU's. When I do the Sickleyield with both my GPU's the render takes about 50% longer with 4.11 than it did with 4.10.
But when I render with just the single 1080ti it renders in the same time as it did in 4.10 with just the 1080ti.
Anyone else notice anything like that?
I'm not using IRay but figure I am still allowed to have an opinion:) Yeah having to use a denoiser to produce a render in a reasonable amount of time seems to me like resorting to old technology.
Well, the point was not about old technology but the fact that using the denoiser, while it might give faster renders, doesn't fix the problem of slow renders in 4.11. The car is still broke.
The post I was addressing was comparing beta and general release.
Yes, 4.11 is slower than 4.10 (bug fix relating to normal handling) though 50% is not the norm, I think.
It would depend on the content of the scene.