Iray Preview Mode and GPU
ericred
Posts: 13
Greetings fellow Daz users,
Can someone answer this question for me? The thing I spend the most time tinkering with in DS is the lighting, sometimes spending many hours to get it right. I use the Iray preveiw mode during this process so that I can experiment with subtle changes in the settings and see what the final render will look like. Recently I have been thinking about upgrading from my current GPU (GTX 970) to a more powerful 1080ti. But I keep reading tidbits of information in the forums which lead me to believe that upgrading my GPU will not improve the speed of the Iray preview mode, it will only speed up the actual render. Is this accruate, or am I mistaken?
In addition, if the GPU can improve the performance of the Iray preview mode, what would be the best configration? Should I keep the 970 as my primary card and add the 1080ti as a secondary card only for rendering (as many people recommend)? Or would that only help with rendering but not Iray preview? Can the Iray preview benefit from multiple cards/cores if I decide to add more down the line?
Sorry for the long rambling question and thanks in advance for any insights into this matter!
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
All GPUs help with the preview. Now, if one card doesn't have enough memory for the scene, it won't get used. There's a big performance difference between the 970 and the 1080Ti. I'm not sure what benefit the 970 will give (not being dismissing, I really don't know). You can certainly use both cards for rendering. But then you can't do much else. One thing I like about having two video cards is that I can keep playing games or use VR while the other card renders. Or if I'm not really using the machine, I can use both cards to render.
My guess is that if you get a 1080Ti, you'll soon ditch your 970 and get another 1080Ti. You're gonna want the 1080Ti as your primary card. Doesn't really make sense to have the 1080Ti as your secondary card even if you don't do much else besides render.
+1
Use 1080TI for rendering and preview (both of which use GPU). Use 970 for your monitor since its contribution to iray performance would be negligible. The 1080ti dwarfs it.
Before you go buying anything, first make sure you have your settings optimized to speed up the iray preview:
Thanks everyone for your input, that clears things up! And I will make sure to optimize my settings as per ebergerly's instructions!
Some really great info here, I;m considering swapping out 2x 980ti for 2x 1080ti - benchmarks seem to say this is a good idea but just curious if anyone who did this noticed any marked improvement in render times?
(There are a couple of benchmarks for 2x 980ti and 2x 1080ti but they are on different machines so its always a bit subjective.)
Can't find these two in DS 4.9.4.117, are they for a later version?
They date from 4.8, so they are not new. I suspect that you do not have Nvidia Iray selected as your DrawStyle. Do this and the options Ebergerly highlighted should be visible.
That was it. Thanks!
...this is why I don't use Iray View Mode or HDRIs as I only have an old 1 GB 460.
oh sweet thx for sharing. That makes a huge difference with running a 1080ti
And if you create a small render of the scene in a new window, 100x100 px is enough, before using the preview, and keep it open in the background, the preview will often update much faster because the render will keep some data in memory as long as it's open. At least with some scenes, it may depend on the size of the scene (RAM/VRAM used) and/or the content used.
Thank you for the info:) Would those settings impact the render quality. I am asking because I use render settings that came with an HDRI set and I wonder if those settings might interfere with it?
As far as benchmarks for those cards, the 980ti did the Sickleyield benchmark in around 3 minutes, and the 1080ti in around 2 minutes, which is only about 33% faster. Not sure if that qualifies as "smoking", but anyway.
For dual cards, 2 x 1080ti did it in around 1.3 minutes, which is consistent with the general findings of dual cards not quite cutting render times in half compared to the single cards.
Also, I think any attempt to assume the benchmark times are accurate to 1 second is a huge mistake. I tend to assume they're, at best, +/- 15 seconds or so since there are so many variables. IMO, benchmarks for iray renders are ballpark figures to give you a general idea of relative performance, nothing more.
As to how all of this affects preview times (assuming the settings are optimized as I suggested) is unknown. Maybe they'll be proportionally faster compared to render times, maybe not. I'm not sure anyone has ever come up with a way to measure that. And if the settings ARE optimized, does a 30% faster (or whatever) preview really matter, and is it that noticeable? Dunno. If someone wants to come up with a benchmark for that, be my guest.
The question I responded to was specifically about render times, not the viewport. As to what is smoking or not, that is your opinion. My 1080tis don't do 1.3 minutes to render. I've done the test MANY times. Enough times to be able to say that my time is not some sort of error, because every run is only 2 seconds apart at the very most. NOT 15. The only time it was greater was from comparing 4.10 to 4.11, where my fastest time in the bench is 58 seconds in 4.10. I have not matched that in 4.11.
I just ran the bench yet again to be sure. Here it is straight from my log
2019-07-19 21:36:34.513 Total Rendering Time: 1 minutes 2.79 seconds
That time is nearly identical to what I posted before, which was 1 minute, 2 seconds..., oh yeah, I just ran it in 1 minute, 2 seconds, how about that. Just like it was several months ago, just like it was several months before that. Do you know what science likes? Consistency! I think I have proven that here. I have never once ran that bench in 1.3 minutes. Where are you getting this number from, anyway? Who benched that?
And once again, if anything, my machine should have worse variables than most others who also have 2x 1080tis, given that my motherboard and CPU are from 2014, and they were not even top of the line from that year to begin with. I'm also sporting DDR3 RAM, rather than DDR4, because again, my PC predates it. But yet none of that matters. My 2x 1080tis match up against anybody's. The only possible variable is just how much of a pimp I am. Perhaps you need to add a pimp factor adjustment calculation to your cost analysis table.
So again, two 1080tis are not 1.3 minutes, you need to revise your performance gain over the 980tis, as it is much higher than just 33%.
The 1.3 minutes (aka, 1 minute 20 seconds) I posted for 2 x 1080ti was from what I saw some other users post in the benchmark thread. So it appears that your results differ from some of those users by about 15-20 seconds. Which is why I tend to assume that render times between users in those benchmarks are +/- 15 seconds or so.
Different people have different hardware, do things differently, probably have different software and/or drivers, probably have different stuff running on their GPU, and so on. So while an individual user might have highly repeatable render times, the general population of users is more concerned with what they might achieve, which is probably better reflected by an average across users. Especially since it's highly unlikely that users will be rendering the benchmark in their normal usage, and instead will be rendering their own scenes whose render times can vary due to scene composition, etc.
Ballpark numbers are the best you can hope for. And two GPU's tend to fall short of cutting render times in half (50%). Closer to a ballpark of around 40%, give or take.
BTW, here's a thread discussing some results from multiple users for 2 x 1080ti showing 1 minute 20 seconds. For those who want to go thru and check all the nitty gritty details of all the benchmark results and the search for a reason why they might be in error, feel free. I didn't (nor would I) spend any time researching every single benchmark result to make sure it was performed perfectly when I made my summary spreadsheet. I'll leave that to others.
I did see others getting closer to 1 minute, but I took the longest render times as worst case, since it shows that different, well-intentioned users can get significantly different results.
https://www.daz3d.com/forums/discussion/162446/gtx-1080-ti-benchmark
Could also be a difference caused by people starting a render from scratch vs textures already loaded in Iray Preview? Obviously with such a short total render time, this would be more dramatic.
Sure, it could be a bunch of things. And we can (and I'm sure we will) speculate for the rest of the year as to exactly why the differences. But at the end of the day most users just want to get a ballpark idea of "if I spend $X how much faster will my renders be?". And since everyone renders different types of scenes, and they'll encounter tons of other factors down the road (eg, new versions of Studio that slow renders by 30% for some due to a bugfix, or they're doodling with Gimp during a render, or playing a game during a long render, etc.) there's no perfect answer down to the second that always applies now and in the future. Hence the need to use ballparks.
"If you spend $800 your renders will be something like 30% faster, give or take. Your results may vary". That's about it.
I agree with this. You take the average of benchmarks, you get a fairly good idea. That's why I said in the other thread, there is no need to be completely anal with one result vs the other. Look at the big picture and that is plenty of information.
But we digress and should probably not pollute this thread with even more of that debate.
Fyi you can skirt the issue of render time variances between different machines almost entirely by looking for the following line (or lines, in the case of multi GPU) in your Daz Studio log file:
The last bit,
Is the actual bare metal amount of time (in seconds) it took the GPU to render the scene regardless of issues like loading textures from memory, scene content being "pre-loaded" via use of Iray liveview, etc. Total Rendering Time is an inherently confounded variable because it includes the amount of time taken by Daz Studio to initialize the rendering engine (either IRay or 3Delight), load textures into the rendering engine's active memory, and receive/save rendered results from the rendering engine during/after the render is finished in addition to the actual amount of time rendering a scene.
Really? What about differences due to some folks having other processes running that are using the GPU and causing slowdowns? What about differences in software and versions and drivers? What about those who are unknowingly encountering throttling of the GPU due to some thermal issues?
The list goes on...
Yes, really.
Simple - don't do that (purposefully run other things and cripple your apparent system performance) while benchmarking.
Check them first, and then state which versions were used to get your benchmark result. Problem solved.
Download/open GPU-Z. Click the Sensors tab and watch the PerfCap Reason field for anomalies during the rendering process (something you should be doing periodically anyway if you care about the longevity of your computer hardware.) Disaster averted.
All of these "issues" are non-issues if you know what you're doing.
Thank you very much, ebergerly
I,m enjoying Daz Studio again
By the way, maybe you know how to force DS to use only 98% of my CPU ?
And if all those who were submitting benchmark render times were trained professionals, following a clearly defined set of procedures to make sure everything was perfect, then yeah, you could get more accuracy.
In the real world, that's not the case.
Hence why people write guides for this sort of stuff (since anybody can follow a guide...)
There may not have been decent guidance for benchmarking in Daz Studio previously, but the future is a virtue. Once Iray with RTX support fully makes its way into Daz Studio, things will start to work themselves out much more nicely, I promise. Till then - I suggest patience.
Unless your CPU(s) give 50 threads you can't do that literally, but you can exclude one or more virtual cores (in Windows, I assuem Mac has an equivalent) by finding the application in Task Manager's details tab, right-clicking on it, and selecting set Affinity then unchecking one or more cores. On an 8 thread system each represents 12.5% of the total possible load
How do you interprete that field?
I suspect that when some of us are presented with a long, 26 step procedure which includes scrolling thru long log files and monitoring PerfCaps and checking GPU processes and writing down all our software and driver versions, as well as having to download special monitoring software and a special benchmark scene, for the sole purpose of improving reporting accuracy from 15 seconds to 1 second (which most of us don't really care about, for the reasons I mentioned), we might have a lot of non-participants in the new benchmarking thread
Personally, I just post my render times to brag about how fast my 2 GPU's are
I don't care about all that other stuff 