More info needed on RTX 2080 Ti performance
RobertDy
Posts: 270
Hi everyone,
I'm very late to the Iray party but late is better than never I guess. I'm considering an RTX 2080 Ti, but info is quite limited for now. I saw a few Iray benchmark threads but the test scenes in there were mainly just one character in an outdoor setting, without any background.
The scenes I render usually have two characters and an indoor setting, like in the attachment. I understand that indoor renders take up much more time in Iray, so I wonder, for a single RTX 2080 Ti card rendering 1920 x 1020 pixels for such a scene, what's your usual time? Knowing the performance of a single card is important so I can decide if I should build a rig with two or more RTX cards.
Thanks in advance for your inputs!
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 have a 1080ti and such a scene rendering on it, assuming the same lighting and quality settings, would not take very long at all. There is a benchmark thread in The Common forum that shows performance of various cards. You could render the test scene and then compare your time to the time reported for the 2080ti. That should give you some idea of the speed improvement you canexpect.
Thanks, how long is 'not very long at all'? And by 'a 1080 Ti', I guess this means only one card?
As I wrote the benchmark thread's test scenes don't have any 3D background and have exterior lighting so although I've seen render speeds posted there with the RTX 2080 Ti, I can't gauge how it'd fare with indoor lighting settings + 3D background.
Run the bench on your own HW. Note the speed. Look up some 2080ti's and note the average on the bench. Render your scene. Note how long it takes. The 2080ti should scale roughly the same to your scene until DS 4.12 comes out and you can use the RTX features and see a much larger reduction.
That makes great sense but I’m a 3delight user at this point. Do you think iray render times can be scaled to 3DL’s?
The 2080ti is roughly 2x faster than a 1080ti. I have a 2080ti running the 4.12 beta and using the AI denoiser. An indoor scene with 2 G8 chars and clothes + some props at 4k gets to an acceptable quality for me in 7-15 min depending on the depth of shadows and lighting.
I will note that I dont go for hyper realism and my renders are almost always finished in 100-300 iterations.
Not really as they are doing very different things - Iray does by default a lot of the things that have to be explicitly set up in 3Delight (usually with a concomitant loss of speed) while being unable to do some of the things Iray can do (e.g. unphysical fall-off for light, objects or lights that don't cast shadows)
@RobertDy "That makes great sense but I’m a 3delight user at this point. Do you think iray render times can be scaled to 3DL’s?"
If you are a 3DL user, there's no point in getting an RTX card.
His first post indicates that he wants to begin using iray.
My advice would be to go with a single 2080ti for now, you can always add a second one later if a single isn't as fast as you want.
Thanks Richard, Paradigm and Petercat for the useful info.
Pardon my limited knowledge, isn’t 4000 iterations the default limit that Iray stops, so an ‘acceptable’ number of iterations for an audience (I render and sell comics at 1920 x 1080) should be maybe at least 3000 iterations? 100-300 iterations seems to be on the low side but you said you’re using a denoiser so maybe that helps I guess.
I was told that to truly optimise 2 GPUs or more, you’d need a strong motherboard too. I was recommended a x99 Sage kind of workstation which costs a bomb, what sort of motherboard do most iray renderers use? I’m on a i7-7700 currently, 4 cores 8 threads
I just purchased an AsRock Extreme7+
https://www.asrock.com/MB/Intel/Z170 Extreme7+/index.asp#Download
I like the options for added storage, including 2 m.2 SSD slots and all those SATA connections, and it has room for 2 video cards.
Didn't cost a fortune on Amazon either, although perceptions may vary.
Right now I'm running a 1080ti for rendering and driving a 4K monitor with the onboard graphics to free up the GPU memory.
I have a full tower case, but I really need a larger one.
You're not freeing up any VRAM by running the monitor off the IGP. Windows reserves a video buffer on every video card except pro type cards set to disallow that.
It depends entirely on the scene. I've had scenes that looked horrid after 5000 iterations and some that were finished after 500. That's what makes the denoiser so amazing. You let it do its thing, it doesn't effect the convergence calculation so it won't stop the render but you can stop after a reasonable number of iterations and usually have a great scene.
To run multiple GPU's you do not need a HEDT system or a workstation. I run a 1080ti and 2070 on a X370 I bought when first gen Ryzen came out. Any board with multiple full length slots can handle multiple GPU's fior rendering.
I'm running Windows 7. I checked with the monitor hooked up both ways, and there's more free memory on the 1080ti with the monitor connected to the onboard graphics. Just another reason for me to avoid Windows 10.
The AI denoiser is pretty crazy. If you enable it at the start all grain goes away at 10 iterations. It uses AI to guess based on real samples. At 10 frames it's really blurry, but it slowly gets sharper and sharper as the real scene converges. At 4000 iterations both will look almost identical, but at even 300 iterations the difference between that and a truly converged image is impressively minor.
Again, I'm not going for hyper realism and I'm not tweaking my surfaces too much so there may be details that others expect that are lost. If I remember tonight I can do a demo render as an example.