What is the best Graphic Card to maximise IRay rendering time?
Hello,
I'm willing to migrate to IRay but as for now it's still impossible to work with it since I only have a GeForce GTX 970 Turbo, 4Gb.
I want to reduce considerably my render times and I'd like advise to what GPU is best to acquire, like the Titan X, or the GTX 980 Ti.
My motherboard is Asus X-99A, CPU is i7 hexacore.
Thank you.
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
© 2026 Daz Productions Inc. All Rights Reserved.
Comments
I have a GeForce GTX 960 4GB and don't have any problems with Iray Renders.
I don't have any problem per se. I just want to speed up my rendering time, either with the actual rendering, or the DAZ3D 4.8 option to see a preview of the rendering while working on the scene.
Currently the GTX 980 Ti is the best bang for your buck.
Hi,
I am also using a 970. For your portrait renders with just a backdrop, even 20min can get up to 1-2k iterations . But with an enclosed scene, 20min is maybe 200-400 max. Maybe can be better with decent lighting, but i doubt it can improve much for enclosed. Go for the best if you got the budget.
Those that don't think an ATI card or even a lowly Intel HD Graphics 3000 integrated GPU don't contribute to an iRay render consider that after my Intel HD Graphics 3000 video driver crashed after about 30 minutes and wasn't recovered DAZ Studio continued rendering but really in CPU mode only and a render that usually takes only 1 hours was still running 10 hours later. That's not to say nVidia card doesn't make it even faster because the iRay SW was designed to use nVidia card driver extensions.
The thing is that I don't know if TitanX is better than 980Ti. It has more cuda cores and double the memory. But does it matter for IRay? Does it get me actual faster renders?
Yes but more for being able to render bigger scenes within the TitanX but it's actually faster too in scenes that both the TitanX and the 980TI can both fit in their RAM.
That would be more likely due to a slow-down of DS and the CPU render as an indirect result of crash and recovery, perhaps interfering with window draws if you were not rendering to file, rather than an indication that the non-nVidia GPU was actively controbuting to the rendering.
If you're planning on rendering large scenes with 6+ characters on a regular, than I would consider a Titan X, because of it's VRAM capacity. The amount of VRAM won't make a difference on render times. Otherwise, the difference in the number of CUDA Cores, which makes the biggest difference in Rendering speeds is about 9%. So, to pay almost twice as much to render 9% faster doesn't make sense to me, unless you plan on rendering very large scenes, like I previously stated. In my opinion, I would buy two GTX 980 Ti's before buying a Titan X. And I put my money where my mouth is, I just bought a 980 Ti a few weeks ago...
Other than if there is not enough, the card won't be used (which is a big impact on render times...CPU slow:GPU...fast).
According to the thread title, you want to maximize render time in Iray
So any non-nVidia card will do and you'll need a 1.6 Ghz processor and only 1GB RAM (which is the minimum requirement for Daz to work) ;)
I meant to imply that, but didn't really lol. Thanks for clarifying me.
Thank you all. So,if I got it right, GPU Ram is only to handle scene lag, more RAM more content on complex textures. Nothing to do with rendering speed.
To get a better rendering speed I have to invest in cuda cores. So, better pay just a little more and buy 2 GTX 980ti to have more cuda cores for the same ammount of GPU Ram as a Titan X.
Correct?
Yes, more GPU RAM means you can render larger scenes. Besides the fact mjc1016 pointed out, if your GPU doesn't have enough VRAM to fit the scene it will be excluded from the render job, otherwise it won't make a difference on render speeds. In my experience for example, 4GB VRAM should allow you to render an environment with 4 characters in it. 6GB VRAM should render an enviroment with 6 characters in it. I'd assume 12GB VRAM and you could probably render an enviroment with 12 characters. Those are just general guidelines and you might be able to get it to render with + or - one character. Also, VRAM isn't cummaltive with Iray, meaning two 6GB cards doesn't equal 12GB VRAM to work with, your still limited to 6GB.
I wouldn't buy a card with less than 4GB VRAM.
More cuda cores is the biggest contributor to faster render times. My test results showed clock speed makes very little difference.
I wrote about all this stuff on my blog, it may be explained more clearly on there. laylo3d.com
In conclusion for the fastest render speeds, if you will be rendering scenes with 6 characters or less, go with 2 GTX 980 Ti's. If you need to render more than 6 characters in one scene, go with the Titan X.
I hope that all makes sense....
That sums it up nicely...until Nvidia actually starts releasing the next-gen cards (then it will be all confuzzled again).
Thank you Laylo3D. That was really enlightening. It seems that I should choose 2 GTX 980ti instead of a TitanX. My scenes don't use that much characters usually. 2 to 3 is the norm.
But that arises a new question. You say that 2 980ti are better than aTitanX in those conditions. But I got to make sure my system supports 2 aditional GPUs (I already have the GTX 970).
My motherboard is https://www.asus.com/Motherboards/X99A/
Most of the specs are chinese to me. Do you think it supports 2 aditional GPUs? If so, I'd have a system with 1 GTX970 and 2GTX980ti. Is that a good combination?
Thank you all!
With that motherboard and the fact you said you have a 6 core CPU I'm guessing your CPU is either an Intel® Core™ i7-5820K or an Intel® Core™ i7-5930K? Both of those CPUs and your motherbaord will support 3 GPUs *As long as you don't have an M.2 device!* If the M.2 socket is occupied it disables you PCIe x16_4 slot...
I would put the GTX 980 Tis in the upper 2 slots and the 970 in the bottom slot. If you have the i7-5820K CPU you PCie x16_4 slot will only get 4 PCI express lanes, but from what I've read that shouldn't be a problem for Iray. My GPU cards all get at least 8 PCI express lanes, so I personally haven't done any testing with the 4 PCI express lane scenario. What I read says a low number of PCIe lanes will only slow down the transfer of data to the card and then back from the card. Meaning it should render just as fast except it will take a second longer to start rendering, i.e. to load the scene to the card, and then take a second longer to send the image back when it's done rendering.
With two GTX 980 Tis and a GTX 970, you'll have one awesome rendering machine, IMHO...
Good god man!! Your computer can handle 2 GTX 980's AND a 970? Talk about a machine you don't want to run into in a dark alley!
Thanks again Laylo3D. That's very satisfying.
I have a i7-5820K CPU indeed. I don't think I have a M.2 device. I don't even know what that is. All I have instaled is the CPU, the CPU cooler, an SSD hard drive, another hard-drive, 4X8GB Ripjaw DDR4 RAM, the GTX 970 Turbo.
If I buy 2xGTX980 Ti, will DAZ run IRay using my 3 GPUs? I don't want to make my GTX970 obsolete.
I also noted that you use the GTX 980 Ti Strix. When I look around I see a lot of GTX 980 Tis, like the Strix, Gaming, Extreme, Golden, Platinum, with prices fluctuations up to 200$. Do the diferences between those products have any relevant impact in rendering in IRay? I'm talking about lots of minutes impacts, not seconds.
So glad you're asking these questions. (sorry to interupt) but I want to upgrade from my 960 too. This is all good to know! I'm curious too, aside from the price difference, what can a person expect from a 4GB card vs. a 6GB card? Obviously there is a 2GB difference but like baldorvadras, I don't want to make my other card pointless. I just want to have one tough render machine.
The most commonly used m.2 devices are those bare SSD cards.
There's an often overlooked trick that makes having a GPU with more ram, more than just about scene size & character limits. This is something I just recently learned, I think from Dreamlight tutorial. If you have a scene which say takes up 3GB of ram and you have a 6GB card, you can speed up effective iray render time, by rendering at much higher resolution than you actually want for final output, and then downscaling after rendering. You can get away with rendering half the iteration cycles this way. Its uses more of your GPU RAM to do this, but is a trick which greatly reduces render time. Which means, that if you have the money, for sure get 2 x Titan X 12GB vs 2 x 980ti 6GB, even if you never plan to have scenes that big, you can use this trick to speed up renders. It would be like having 4 cards in your system, when you only have 2.
Further explanation if you want to know..
For example:
Lets say you have a scene with 4 figures, and total scene is 3GB, and takes up half your cards ram, and you currently have 2x980ti, this is what I have. You want a final render of 1920 x 1080. Lets say it takes 500 iterations to get a 90% convergency ratio that you are happy with, and maybe it takes 2 minutes to render with your 2x980ti GPUs.
You can change your render to 3840 x 2160, it will just about fill your GPU memory, but now your whole card capacity is being used, it takes maybe 100 iterations to get same quality after you downscale the 3840x2160 down to 1920x1080, and you rendered it in a minute.
These aren't exact numbers, but just giving an example of how this works. I use this in animating and its saves many hours of render time. There are other factors involved, your initial file sizes will be much bigger, so there's that to deal with. But its very practical. Now imagine, you do the same thing with Titan's, you can really speed things up, even for large scenes that would normally overflow your cards.
You're welcome! Like nonesuch00 said, most likely if your M.2 slot was occupied, it would be occupied by a SSD. The way to know if you have a M.2 SSD is it plugs directly into the motherboard, while the normal route is using a SATA cable. Like he said, they also just look pretty much like a bare circuit board, not housed in it's own little case.
Daz Studio will use as many GPUs as you can throw at it as long as each individual card can fit the scene in it's VRAM like has been mentioned in this thread. It doesn't matter that they run at different clock speeds, have different numbers of CUDA cores, etc. Using Iray Server you could create your own render farm, but that's a whole nother can of worms...
One thing I forgot to mention is, you need to make sure your power supply can handle the additional GPUs. If it's around 800 watts or more I would think you should be fine unless you're going to run overclocked GPUs, they can draw exponentially more power. Which brings me to your next question...
All those different variations of the 980 Ti differ the most in clock speed and the parts they use in the construction of the card that can make them more "overclockable." Like higher quality transistors and fun stuff like that. For Iray, the spec clock speed should be fine. I only shaved off a couple seconds on my render time when I overclocked my cards briefly... The different cards can also vary on their ability to dissapate heat. I persoanlly really like the EVGA 980 Ti Hybrid, and the ASUS 980 TI STRIX. Although in my next computer I'll probably go with 4 of something like the ASUS POSEIDON-GTX980TI and build a custom water loop. But, to answer the question, you probably won't notice a significant difference in render speeds between the Strix, Gaming, Extreme, Golden, Platinum, etc. There would bigger differences in gaming FPS...
I have an EVGA 980 Ti Hybrid and two ASUS GTX 980 STRIX in my computer. The hybrid is definitely the shiznit, you just need to be able to mount the water cooler somewhere in your case.
I've heard of this as a thing to do, but for whatever reason it's never appealed to me, so I don't have experience with it. I would question how well it works with low light scenes though? I seem to do a lot of those. Then you could get into the debate of just making bright scenes look like night scenes using shutter speed, or whatever, but I personally, again for no real good reason, don't do that either LOL!!
You would gain the abiliy to render more characters, props, or whatever in a scene using the specific GPU. If you look more closely at some of the other posts in this thread, I believe it is discussed thoroughly.
Hi Laylo,
Yes I'm sure it is but most of it is going over my head. I did an Iray render last night that took WELL over an hour and had 3 characters in it. I figured a 960 could handle that but since I added HDR light to my scenes, it seem I have very long render times. I only have 8GB RAM at the moment and that may have something to do with it. I consulted my manufacturer and they confirmed that my system an handle 2xGPU. So I guess my question was if a 6GB would overpower my 4GB but if i am understanding some other things you said correctly, I would be working with 10GB total and things would be better. I doubt I my system can handle much more as it is a 600w system.
My SSD hard drive is plugged with a SATA cable.
My power supply can support up to 850W. I'll see how much money I can save until the summer.
Thank you for your feedback!
For Iray, having a 6GB VRAM card and a 4GB VRAM card would be fine, but together Iray wouldn't consider it 10GB VRAM to work with. The entire scene has to be loaded to each individual card. Meaning, if your scene required more than 4GB VRAM Iray would exclude the 4GB card from the render job. If the scene required less than 4GB VRAM, Iray would use both cards.
As far as your long render time, there are other factors at play. Sometimes doing renders one after another your VRAM will get full and to render a scene that would normally fit in your card's VRAM Iray will switch to CPU rendering and you have to restart Daz Studio to get the VRAM to dump it's contents. I wrote about this on my blog as well, and possibly more thoroughly... If you're a Windows PC user, and you render with only your GPUs checked, you can hit CTRL + ALT + DELETE, open up Task Manager, and see what percentage your CPU is being used. If Iray is using your GPUs exclusively, it shouldn't be at 100% in normal circumstances. If it is at 100% or close to it, then Iray is probably using your CPU to render and you might save your scene, restart Daz Studio and start the render over again, unless it's the first render you've done since you started Daz Studio. If it's your first render, it probably means your scene is too large to fit in the card's VRAM...
I hope that helps and answers your questions.
you all are the bomb.com. i have a 2gb gtx 670. my render times aren't really long (2 figures about 20 minutes in iRay with NO background) but i do want to not have to wait around for 30 minutes. im running an 08 octocore mac pro with 16 gb ram and i'll upgrade to 32 soon. so should i get the Rx480 or a 980ti? i LOVE the 480 price point and the 8gb ram too....
Laylo, is there any advantage to having identical cards? I have a MSI 980ti Gold, and want to add another soon, but they're getting a bit scarce.