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 know when I used it in CPU rendering, renders seemed to take longer.
Yes but as I reread it just now I saw that OptiX was a key factor especially for the AI denoising. That's why I'm curious about the adoption of the OptiX into the current state of affairs ahead of RTX specific rewrites
Yes, I believe that is part of it. Optix is and has been part of Iray. Some people have had issues with it, but I'm not sure why and if they reported everything correctly. Some have seen speed increased and render times drop with Optix clicked on. https://developer.nvidia.com/optix
Maybe I'm naive but I don't see what devs of Arnold,Octane and others would get out of spreading lies about what RTX can do for GPU rendering. I do think it's a given that the software needs to first support this stuff though. That's my biggest worry for Iray as NVIDIA so far cannot even be bothered to mention it all. That's not a good sign, they could be out there hyping performance of RTX in Iray, and probably would be if they had anything to show yet? Not just for these new game cards but for the RTX Quadros that are already out too.
OptiX is just an API that probably makes it somewhat more accessible for third parties to implement this NVIDIA-specific functionality. Iray is an NVIDIA product and so naturally it already supports OptiX. What's in Daz Studio as an option is 'OptiX Prime Acceleration'. Here's a wiki entry describing that part, I'm assuming OptiX Prime:
https://en.wikipedia.org/wiki/OptiX
Here's what NVIDIA says about Optix:
"The NVIDIA® OptiX™ ray tracing engine is a programmable system designed for NVIDIA GPUs and other highly parallel architectures. The OptiX engine builds on the key observation that most ray tracing algorithms can be implemented using a small set of programmable operations. Consequently, the core of OptiX is a domain-specific just-in-time compiler that generates custom ray tracing kernels by combining user-supplied programs for ray generation, material shading, object intersection, and scene traversal. This enables the implementation of a highly diverse set of ray tracing-based algorithms and applications, including interactive rendering, offline rendering, collision detection systems, artificial intelligence queries, and scientific simulations such as sound propagation. OptiX achieves high performance through a compact object model and application of several ray tracing-specific compiler optimizations. For ease of use it exposes a single-ray programming model with full support for recursion and a dynamic dispatch mechanism similar to virtual function calls."
All of this stuff is about optimizing the hardware and software to do the things users need to do most. It's less and less about generatlized hardware like CPU's which can do zillions of very different and very complex types of operations for all kinds of software tasks, to software and hardware working together to do simple and specific stuff very quickly and efficiently. It's about focusing on specific tasks and doing them well. You design the software and hardware to work together.
AI Denoise is already in DAZ's beta. >>
NVIDIA didn't promote the changes with Iray on the Pascal cards. So I wouldn't worry. Just hide and watch! :)
In some of the videos from NVIDIA, they say Tensor cores help out RTX... so that could be a performance improvement there that we are neglecting.
Yup, sure is.
By the way, in my thread explaining my recent efforts to write a very simple ray tracing software, I showed some of the same things that NIVDIA is saying. For example, when you do ray tracing you do the same, fairly simple operations millions and millions of times to render a 1920x1080 image with over 2 million pixels. To determine whether a ray has hit a sphere you need to solve a simple quadratic equation, which means you need to solve the following "determinant":
So if you can set up your hardware with the necessary registers and memory locations and processors to solve that EXACT equation as efficiently and quickly as possible, simultaneously for millions of pixels, you've got a very fast ray tracer. So that's what they do, they design their hardware and software to extract those key equations used in ray tracing and de-noising, etc., and design hardware and software to solve those exact things. That's what NIVDIA is doing with the latest generations of hardware and software. Their RT cores presumably are excellent at solving these ray tracing equations, and their Tensor cores are presumably excellent at solving de-noising, and the SM's are good at working with Physx software to solve physics problems, and so on. Do programmers always use those exact resources the same way for the same task? Probably not. Who knows? Maybe they use RT cores for de-noising sometimes.
It's incredibly complex, and I'd caution folks not to think they really understand it. The more you know, the more you realize you don't know nuthin'. I sure don't.
A really interesting demo off that page, Control.
https://www.youtube.com/watch?v=PCD_7e0cMv0&feature=youtu.be
Don't know the breakdown of gamers in the genres, but it appears that the majority following the tech are in the twitch fps types. Watching son play Battlefront 2, RTX technology really falls into the "who cares" category. It's just global illumination, but has reflections. It's offering all those little details you just don't have time to sit around and stare at the environment. And arguably may be better for fast action stuff - it's easy to light without shadows being everywhere. Nvidia really used the wrong game to showcase the tech - when they zoom in to show reflections off of a characters eyes in Battlefront V. Like you have time to look at those sort of things in a fast paced shooter.
But for slower paced games, exploring places or with creepy things... RTX will be VERY cool. I've never played it, but a suspenseful game like Alien Isolation will be so much better. A deviation from Iray, but the thread title is OT and about gaming cards anyways.
Some fairly indpeth info:
https://www.chaosgroup.com/blog/what-does-the-new-nvidia-rtx-hardware-mean-for-ray-tracing-gpu-rendering-v-ray
I like one bit especially though:
Doesn't that sound very much like the gamingf cards pool just fine?
Other important points to take away from the article:
SLI is dead. It was a flawed way to combine cards that resulted in micro stutters in many games, or sometimes wasn't supported at all. This statement makes no sense to me, NVLINK will be NVLINK no matter what. Worst case it won't pool memory but it has nothing to do with SLI. That's the whole point.
Hmm ok, so the SLI term is here to stay? Guess I had the wrong idea. SLI has such negative connotation, would have thought they would replace it. But maybe the underlying tech is still SLI, just much faster with NVLINK.
SLI has been good here except for Maxwell... I figure the link was too slow for the cards & output. I had the micro stutters with 4K and read about other users with micro stutter problems. Once I upgraded to Pascal based cards, SLI was super smooth again.
NVlink is a new tech speeding up transfers greatly from system memory to the GPU's memory.
Any render we do will have a good portion of ray-casting though. I think it's safe to say that in every single render there will be performance enhancements beyond just raw CUDA, if/when Iray gets such support. Note also:
So, are you stoked yet about it all, and about perhaps a renewed hope that NVLINK might indeed pool for game cards too? Or should we still be "cautious"
My take on those 6-8x faster render times is, that those are probably achieved with ai denoising or something similar. If you render cars or something else with easy metallic surfaces, then I'm sure that works like a charm. For characters with hair, sss and all those little skin details we are so obsessed about, I doubt it works that well. Probably those that have pascal card and have tried new DS beta can tell better. My current maxwell card doesn't support iray ai denoising ( I think ), so only guessing here.
Still, that new 2080ti looks like an upgrade to me. Price is high, but over 4000 cuda cores, higher clock speeds and new faster memory is nothing to sneeze at. Despite my scepticism, I do believe that at least I'll get nice increase for rendering performance. Who knows, maybe those RT and Tensor cores really are some magic ingredients, and we really see some huge improvements here, but I'm still little sceptical about that. Haha, it's so good to be a pessimist, because then you only get nice surprises
Like I said earlier, if I had Pascal card, I probably wouldn't be upgrading and definitely not preordering, but for older model owners I think it'll be a considerable performance boost no matter what.....when turing architecture is supported. Meanwhile, I thought I could try Octane. Never tried it before, but apparently they have a DS and Blender plugins, so worth a try at some point.
sigh i think the 2070 is in price range for me.
My old GTX 970 needs a change :)
But i wait for the first benchmarks to decide
...still have my doubts they will actually offer the full NVLink on GeForce cards. Full memory stacking is a huge selling point for their professional Quadro series. GeForce is aimed mainly towards the gaming community and while GPU core pooling for improved frame rate is important, memory stacking is not. The former can be achieved through SLI or a more "limited" form of NVLink as they are offering. Go to the site, the GeForce NVLink bridges have he same configurations like the old SLI ones: 2, 3, 4 slot. Also the price for a pair of bridges for the Quadro line is currently 599$ while the bridges for the GeForce ones are 79$.
it also sounds like the software needs to support the process of memory stacking as well since only one render engine apparently does so.
...only doable if someone was magnanimous enough to send me a certified bank draught to cover the price of two 2080Ti's + the NVLink bridges, I'd be happy to to give it a go.
edited post to be more precise
- - -
@ proof for possible RTX speed increase
The 8x speed increase was tested with an RTX Quadro and showcased in videos on the Otoy twitter channel with a "new (and unfinished) vulkan kernel".
https://twitter.com/otoy/status/1030269165318287360?s=21
https://twitter.com/otoy/status/1030626195669311488?s=21
And yes, I find it somewhat strange that all the new information available about RTX performance is from other render engines and not from Nvidia Iray.
- - -
@ customer expectations for new card releases
To be clear I do not expect people who use DAZ Studio as a hobby to invest large amounts of money on new graphic cards.
But keep in mind that some customers who frequent the DAZ3D forum also own other software and use other render engines that will make use of the RTX technology of the cards.
It will not be possible to install the Turing cards for use with other software and then "quickly" swap them out again for Pascal cards for a render with DAZ Studio Iray.
Once the new cards are installed they stay in there. Either DAZ Studio and Iray will be ready this time or customers are left with the option not to use DAZ Studio and Iray.
DAZ Studio and Iray may not yet support RTX features at launch of the new cards at September 20th. Still, it should be a reasonable request that DAZ Studio Iray is updated in time so that the Turing cards can be used for normal CUDA based calculations as the previous Pascal cards.
- - -
They were silent last time when Pascal cards were released. I think it's their way, or they aren't ready yet.
By all means discuss the information we have, and try to draw inferences from it, but please avoid discussing others' pre-ordering or not.
I don't think it's dismissed.
The issue is that there were no indications; the presentation was completely missing any comparrisons with previous generation cards. None at all: why was that?
So whilst I am prepared to admit, indeed I'm expecting the new features to aide rendering, but that aide (that extra performance) is added to what exactly?
Will the new cards be better? Yes.
How much? Only Nvidia know for sure; the rest of us can (at best) speculate - and largely it's guesswork.
I was excited about the 8 times faster.
I was dubious - very dubious - when I saw the 'up to'.
I can certainly see 8 times faster renders when we're talking about objects such as cars and architectural scenes; what many of the folks do in Studio involves people, and denoising (as an example) does poorly in comparrison there; it basically does not look as good in denoising.
Now if the speed increases are due to other functionality then it should hopefully be much better.
... All this speculation is fun, but ultimately pointless.