GTX 1080 Ti Benchmark
Godless8
Posts: 25
Greetings!
Considering I know what it's like to look around, to see which hardware does what for us Daz users.
I thought I might aswell add my own findings of my brand new machine.
My Machine consists of the following specs:
GPU : 2x GTX 1080 Ti's Both from Inno3D (NO SLI) (one at 1950 Mhz second at 1800Mhz)
CPU : Ryzen 7 1700X @3,6Ghz boost
RAM : Corsair Vengeance3000 64Gb @2111Mhz
MOBO : Asrock X370 Taichi
Storage: Samsung 1 Tb 850 EVO SSD
All tests are Iray based, I Used items that daz provides for free.
I kept everything as is, used the camera's that came with the scene's. Anything I adjusted is in the details.
Test1:
Sickelyield's benchmark:
https://www.daz3d.com/forums/discussion/53771/iray-starter-scene-post-your-benchmarks
using 2 Ti's and optix acceleration. NO CPU
Completed in 1 min 20s
+- 5000 iterations
Test2:
Color bleeding Sample:
using 2 Ti's and optix acceleration. NO CPU
Image resolution set to 2000 x 2588
Completed in 7min 49s
1768 iterations
Test3:
Dragon slayer:
using 2 Ti's and optix acceleration. NO CPU
Image resolution set to 2000 x 1125
Completed in 1 min 44s
+- 1000 iterations
Test4:
Dragon slayer:
using 1 Ti and optix acceleration. NO CPU
Image resolution set to 2000 x 1125
Completed in 3 min 15s
+-1000 iterations
If you have any guestions just ask! I'll try and answer them as best I can!
Godless8
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
Are you testing by having your viewport set to iray first, before hitting render? ( I ask because it can shave 30 seconds or so off your test time, and your time right now of 1:20 doesn't seem right.)
if 2x 1080ti can only manage that, I'd be returning them. :)
1 minutes 40.86 seconds
a 980ti and a 970 with optix
I made sure the scene was loading onto the cards first. Yours should be a lot better. And if not, I am absolutely not upgrading the 970 to a 1080ti.
Just the 980ti. The 970 adds more than I remember, but the issue is noise; there is a hell of a lot more. I use strix on both cards, and even rendering they are quiet, when only one but two is horrible.
2 minutes 38.81 seconds
Would you mind just rendering with one card, for comparison.
Stop a render, keep it open and start another one, that way there is no delay on loading data to cards.
yeah, my dual 980tis does the benchmark in 1min 27sec.(optix enabled, no CPU, preloaded to the GPUs)
I would expect dual 1080tis to do it under 1 minute flat.
I have read that the change in the Nvidia drivers has slowed things down somewhat, but it shouldn't be that much slower. they changed the way iray converges a scene.
This was on the developers blog.
Have also just recently built a new PC for Iray rendering.
Sickelyield's benchmark -
GTX 1080 Ti only with optix acceleration - 2min 4secs
GTX 1080 Ti only with optix acceleration and viewport set to Iray - 1min 21secs
Wow, so a single card is just as fast as dual 980tis. Thats fantastic. The dual 1080ti might do it in 45 seconds.
Which 1080ti did you get? I've heard some concerns about overheating problems and throttling with the founders edition.
@DSDT No initially I did not have the viewport set to iray. I tried that now, and now the exact same scene from sickleyield takes 1 minute to render.
@nicstt one card with iray viewport on is 1min53s . Doing the same thing with the previous render still open gives the same result.
I would honestly take this render timing with a grain of salt. I would honestly prefer to compare a high resolution, memory heavy scene with these cards to other cards.
It's not impossible that at a certain point, the speed difference is unnoticable. certainly in smaller scenes. Sicklyields scene is not very big. Taking into account we don't know exactly how Iray operates, there might be a lot of stuff happening in the background that we can't simply 'speed up' the process off with faster cards or cards with more cores.
I hope this answers your questions :)
Sorry have just noticed the iray viewpoint was set to default camera. Have tried again and was 1min 50secs.
I went with the MSI Founders Edition, overheating doesnt seem to be a problem. When I render temps are usually 64C with fan speed circa 35%, highest I've seen was 68C.
Thanks for retesting. 1 minute is much better. Its not a perfect test but at least it lets me get a ballpark of what to expect and it sounds right based on the specs. Driver versions can greatly affect speed as well. I know for the older cards 362.00 is the best version. Not sure what the best driver version is for your card.
The best test in my opinion would be a short 30 frame animation. I'm probably going to upgrade to dual 1080tis as soon as the supply catches up with demand. Right now, they are difficult to get and I want eithe a water cooled or triple fan model, which I don't think have been released yet.
You're welcome. I know what you mean, Getting my hands on these two puppies was hard enough as it was.
Well ...I'm pretty happy about my own benchmarks now:
viewport on standard: 1:09
viewport on Iray: 59 sec.
So I can see that adding a 1080 Ti won't do much to speed up[ small scenes like the benchmark. That being said, I have my eye on the EVGA GTX 1080 Ti Black SC gaming. The issue with me is speed on large scenes. My 780 Ti kicks off if I have let's say 3 characters in a simple setting, but many of my works use more VRam than that. The 6 gig of the 980's can usually handle the load ...but sometimes I have to really simplify the textures or lower the resolution for bigger scenes. It's likely that driver updates will speed up the 1080 Ti's from what I've read.
Well that sucks! I would try it again, and if it's unchanged roll back the driver.
I have to keep this version because the latest Windows 10 upgrade kept giving me BSOD with the older driver ...but that driver was a real beaut!
Isn't putting the viewport on Iray prior to doing a render test a bit like saying "y'know, if you first pre-render your scene to 90% convergence, the final 10% render only takes 5 seconds !!"
Of course it's faster because you're taking the time to wait for it to pre-render enough to put an image in the viewport. When in fact the render time is pretty much the viewport time plus the "render" time, or something like that...
The GPU kicks in after the pre-render. So if you are testing only the speed of your GPU you should put viewport in IRAY mode. If you went to test your CPU and GPU combination, you should not set viewport to IRAY.
I do the test with the viewport without IRAY because that how the original test was done. When this post started DAZ program would do a full render even if veiwport was set to IRAY.
You sure about that? I have my viewport set to Iray and it's using 100% GPU, and I never asked for a formal render. I'm pretty sure that's why it's called an Iray viewport render because it uses the GPU using NIVDIA Iray.
For that reason I assumed it uses the present results of the viewport Iray render as a starting point for the formal render
Hmmm....I guess then that I'm not understanding what you mean by pre-render. I thought the CPU loads the scene into the GPU memory, then hands it off to the GPU to do the rendering. Maybe you mean the loading part?
Anyway, not sure it's that big a deal in relative terms, it just seemed like we might be getting a false sense of total render time.
Yeah, I just checked again. I think what happens when you select Iray viewport is this:
Sounds like you're calling step 1 as "pre-render". In my experience, that takes some time depending on the scene.
And the same steps happen for a formal render. So as long as you're comparing apples to apples, it's fine. But kinda by definition, if you're doing a formal render with the viewport already set to Iray, you've already bypassed the loading/pre-render step because presumably it's already happened. And if you're comparing that render time to a render which INCLUDES the time to load the scene and do a formal render, then it's apples to oranges I think.
But my point is, the elapsed render times that are given by the Iray renderer INCLUDE what you call the "pre-render" (aka loading) time. And people are posting those times, and comparing to the non-load-inclusive time you get if the viewport is set to Iray and the load has already been done.
Apples to oranges....
Got it, my boy?
2x 1080Ti Viewport set to iRay first.
Total Rendering Time: 56.68 seconds
No iray in viewport: Total Rendering Time: 1 minutes 8.85 seconds
@Godless8: Please try to increase the speed on your RAM. I know it's more difficult with 64GB, but you should be able to get better than that. The RAM speed affects your whole system since that is also the speed that infinity fabric runs at that interconnects the two CCX modules (each CCX contains 4 cores).
Bumping an old thread I know but does anyone have any experience with 4 GTX 1080 Ti ‘s? and performance? I was going to go for 2 RTX 2080s but it seems to make more sense going with 4 1080 Ti’s.
Thanks!
I wouldn't until there are benchmarks to make an informed judgement from.
"Makes sense" when it is based on rumour is not sense.
Ignoreing the performance differences, which are currently unknown.
4 1080ti is as good as it gets.
2 2080ti may allow for 2 more to be added later, and of course may not.
Here's the benchmarks for a bunch of cards, including 1, 2, and 3 1080ti's. I would take them with a grain of salt (maybe +/- 15 seconds or so). I don't recall seeing a benchmark for 4 though...
The 2080 Ti's have more CUDA cores, but what turns me off about them is that they have the same VRAM as a 1080 ti. For me, the VRAM is at least as much of an issue as the cores, because it determines the size of the scene you can render. From what I've seen, the biggest advance of the 2080 is real time raytracing, which is maybe an issue for gaming but not so much for IRAY. My next rig was going to have 3, maybe 4 1080 ti's, so I can see where you're coming from. Actually there isn't a ton of difference in rendering speed between a single 980 Ti and a single 1080 Ti, as long as the scene is small enough. So don't get too caught up in the latest card. I would have a different opinion if the VRAM was higher, but it isn't. The main advantage of the 2080 Ti release as I see it is that 1080 Ti prices will drop a bit.
With only both of my 1080 Ti's running on the benchmark scene with Optix on, no CPU, my benchmark comes in at 1:14. When I add the 980 ti, it comes in at 46 seconds.