Benchmarks  ⋮  Hardware

Octanebench Benchmark Results (Updated Scores)

CG Director Author Alex Glawion  by Alex Glawion   ⋮   ⋮   52 comments
CGDirector is Reader-supported. When you buy through our links, we may earn an affiliate commission.
Octanebench Benchmark Results (Updated Scores)

OctaneBench is the currently most popular GPU Rendering Benchmark. It scales beautifully across multiple GPUs and has a series of rendering tests built in.

The OctaneBench Benchmark Rendering Results are assigned linearly scaling OctaneBench Points.

It is the popular GPU pendant to the well-known CPU Benchmark Cinebench R15, for which you can check the Results here.

Both benchmarks are crucial in finding the best Hardware for GPU Rendering, and ultimately the best Computer for Rendering.

So let’s take a look at the current OctaneBench Benchmark Results and keep an eye on the performance/dollar column:

Octanebench V4.00 Benchmark Results

GPU NameVRAM (GB)OctaneBench ScorePrice $Performance/Dollar
8x RTX 2080 Ti1124419592
8x GTX 1080 Ti1116735600
8x RTX 2070816234000
4x RTX 2080 Ti1111764796
4x RTX 208088743196
4x GTX 1080 Ti118532800
4x RTX 207088032000
2x RTX 2080 Ti115882398
2x RTX 208084411598
2x GTX 1080 Ti114291400
2x RTX 207084091000
Titan V123963000
2x RTX 20606392720
Quadro GV100323599800
RTX Titan243262700
RTX 2080 Ti113041199
Quadro GP100162847000
Titan Xp122501300
RTX 2080 Super8233720
RTX 20808226620
GTX 1080 Ti11222700
RTX 2070 Super8220550
RTX 20708210500
RTX 2060 Super8203420
RTX 2060 (6GB)6170360
Titan X121601200
GTX 980 Ti6156300
GTX 1070 Ti8153450
Quadro M6000241523849
GTX 10808148550
Quadro P5000161451600
GTX 10708133400
GTX 1660 Ti6132280
GTX 780 Ti3117150
GTX 16606117230
Quadro P40008115750
GTX 9804112200
GTX Titan Black6108480
GTX 970495170
GTX 1060 (6GB)688300
GTX 1060 (3GB)387200
GTX 1650478150
GTX 1050 Ti454160
GTX 1050245125
Quadro RTX 8000483165670
Quadro RTX 6000243044400
Quadro RTX 40008188950
Quadro RTX 5000161862100
Titan RTX243222800
GPU NameVRAM (GB)Octanebench ScorePrice $Performance/Dollar

Some interesting Octanebench Scores findings

Let’s take a look at some of those Octanebench Scores:

What becomes obvious right away, is that the performance and price don’t really seem to correlate all that much.

If you look at some of the top Results, the GPUs: Titan V, RTX 2080Ti, Quadro GV100 and RTX Titan, they all have more or less the same Octanebench Score, but the Price difference is HUGE.

Why is this so?

The main reason is, that the different Graphics Cards are meant for different use cases. An RTX 2080Ti is clearly meant more for Gaming than a Titan V or Tesla P100, and Gaming Cards, for our benefit, just happen to be great for Octane Rendering and score well in the Octane Benchmark.

Also, The Titan V and Quadro Cards have more VRAM, optimized Drivers, 10bit Displaying and lots more.

These Features though do not improve the Octanebench Benchmark Results. The Scenes are fairly small, and the VRAM is not an important aspect in the Benchmark.

Another extremely important thing to note is the GTX 1070 Ti and the GTX 1080 scoring 153 vs. 148 Octane Benchmark Octanebench Points. And this even though the GTX 1080 is 100$ more expensive.

It almost seems as though NVIDIA has not differentiated enough between those two cards.

The new RTX Cards are still somewhat too expensive for the small performance increase they offer.

A 1080Ti you can get for 700$ and it has 11GB of VRAM compared to a similar scoring RTX 2080 with only 8GB of VRAM.

What is Octane Render

Octane is a currently extremely popular GPU Render Engine that is supported by most modern and popular 3D Softwares such as Cinema 4D, 3ds Max, Maya, Revit, Blender, Houdini, Modo and more are being added.

What is Octanebench

Octanebench is the GPU Render Benchmark based off of Octane Render.

It renders several pre-packed scenes on your hardware and assigns OctaneBench Points depending on your overall Rendering Performance.

100 Octanebench Points are equivalent to the speed of an Nvidia GTX 980.

Octanebench Points scale linearly. This means an Nvidia TitanXp with 200 Octanebench Points will render a given Scene twice as fast as an Nvidia GTX 980.

Octane Render supports CUDA GPUs only and no OpenCL meaning it supports NVIDIA GPUs only.

In Octanebench you can select if Octanebench should run the Benchmark on all of your built-in Graphics Cards or only on one or on a number of selected GPUs.

This is great for analyzing your Multi-GPU scaling which should be near 100% effective in Octane Render.

Download Octane Benchmark Octanebench

You can download OctaneBench for Windows, Mac OSX and Linux here.

Other Popular Benchmarks

For testing your GPU Rendering Performance, Octanebench is certainly one of the most popular Benchmarks out there. Another popular GPU Render Engine is Redshift and Vray-RT.

For benching your CPU, you will definitely want to go get Cinebench R15 (Up-to-date Scores here).

It is the most popular CPU Rendering Benchmark. It can bench your CPU on Multiple Cores or Single Cores.

This is great for finding CPUs that are snappy for actively working on the computer as well as finding the best CPUs for Rendering.

What GPU to get for Octane

Now, there are certainly lots of extremely fast GPUs that can get you a very high Octanebench Score. These GPUs though usually come at a huge performance premium, price-wise.

If you would like to find the best GPU for the money, I wrote an in-depth article on finding the best budget hardware for GPU Rendering here.

To summarize that article, the NVIDIA RTX 2070 Super with 8GB of VRAM has the lead in most GPU Render benchmarks when you are looking at best performance per dollar.

What Graphics Card do you want to buy?


Join the New CGDirector Forum! Expert Advice & PC-Build Planning with a warm and friendly Community! :)

Alex Glawion - post author

Hi, I’m Alex, a Freelance 3D Generalist, Motion Designer and Compositor.

I’ve built a multitude of Computers, Workstations and Renderfarms and love to optimize them as much as possible.

Feel free to comment and ask for suggestions on your PC-Build or 3D-related Problem, I’ll do my best to help out!

52
Comments

Fernando Cesar Costa Augusto

Alex, which setup would work best for Octane? I already have one 1080ti.

1- Ryzen 3800x
2x 1080ti
32gb 3000mhz

2- Ryzen 3900x
1x 1080ti
64gb 3000mhz

3- Ryzen 3800x
1x 2080 ti
32gb 3000mhz

Thanks for the help!

clarance dsilva

Hey Alex, Im in the process of building a Pc from used parts. Already got a 3800x , x570 gaming X mobo, hx1000 corsair psu, 3600 cl18 32gb ram, AORUS AIO Liquid Cooler 280 and some rgb fans and DB-77 evga case. I want to inves in a best bag for buck GPU, should i go for a new 2070 super (789 cad) or a 8 months used 1080ti (675cad) ? My basic applications will be – revit , rhino , photoshop, Vray, Lumion , D5 render (the only application which uses RTX).

Please help 🙂

Nate B

These octane scores were very useful. Thank you Alex! But I’m a bit confused and forgive me if this is a stupid question. How do I use 8 2080ti cards on a single board? All the motherboard articles here show boards with 4 long pcie slots at most.
Is there a post or feature I’m missing somewhere?

Alex Glawion
Alex GlawionCGDIRECTOR

Hey Nate,
You can get up to 7 GPUs working on consumer boards (such as the Asus ws x299) but the 8x GPU benches were taken on rack-mounted server boards such as this one here: https://lambdalabs.com/products/blade

The added price though is usually not worth it for single artists or small studios. Better to get consumer priced hardware.

If you are starting a Rack though and need lots of performance, something like that could be interesting.

Cheers,
Alex

Nate

Makes sense! Thanks for the link. I’ve been playin around with the idea of a Ryzen 3950x PC for my work (Nuke and ZBrush). Any plans to maintain benchmarks for these apps? I have no clue if this new build is going to be an upgrade or downgrade from my Mac. I’ve spent almost a month researching parts and this is what I came up with:
———————————————-
CPU: Ryzen 3950x
GPU: Quadro RTX 6000
RAM: G.SKILL Trident Z Neo (For AMD Ryzen) Series 64GB (4 x 16GB) 3600M DDR4
Motherboard: MSI X570 Tomahawk or X570 Unify
Cooler: Dark Rock Pro 4
———————————————–
Oh by the way – thank you for the benchmark lists and memory and motherboard articles. Even though I was familiar with PC hardware, the information helped fill in the gaps nicely making it easier to pick stuff.

Alex Glawion
Alex GlawionCGDIRECTOR

Hey Nate,
Both zbrush and nuke are on my list of articles that have to be written, and I’ll also see if there are any official benchmarks available, though if I recall correctly, although zbrush has an inbuilt benchmarking feature, this benchmark is not very demanding and tells you nothing much about the performance of your pc.

Might have to create some new scenes for those two apps?

You’ve picked some great parts! You probably have your reasons for going wiht a quadro rtx 6000 too? Need lots of vram?

Cheers,
Alex

Nate

That sounds great! Please do add them, I’m sure many people dabbling in VFX would find data for Nuke especially helpful. Yes, I can suggest some scenes if you’d like to get started with these apps. Been working with them a long time so I’m very familiar with what sort of stuff turns out to be the most taxing for a computer.
also, yes, Quadro is mandatory for me as I work for a pretty well-known movie studio and they mandate only PCs with Quadro or only Mac devices for some reason. Not sure why. The extra VRAM is helpful so I’m not one to complain!

GAO

I added two 2070 supers. When I render 4K images, the rendering time of one 2080ti is faster than the rendering time of two 2070ti. Two 2070 super renders scored 440, and 2080ti only had 308. Why is the 2070 Super rendering slow?

Alex Glawion
Alex GlawionCGDIRECTOR

Hey GAO,
That’s not slow at all, that is exactly what we measured too. 2x 2070 Supers also just cost half as much (each) as a 2080ti so this can be expected. In the end it’s the CUDA cores that matter, and those are pretty on par with what the scores for 2080ti and 2x 2070 supers are.

Cheers,
Alex

Eric

I added a 4th card and I cant get the Bench Mark to run. Each time is CUDA 999 error -> failed to link file, interestingly each time I run it, its a different card that failed to load the module. From a single one to all 4 at times.

I can render in C4D Octane without issue

Alex Glawion
Alex GlawionCGDIRECTOR

Interesting! What GPUs do you have and on what motherboard? What version of octane bench and what version of octane render are you using?

Thanks!
Alex

Eric

Just ran the benchmark and happy to report that despite my 3970x having a relatively disappointing viewport score, my 2x RTX 2080Tis performed well at a score of 623.24.

Theyre 2x Gigabyte Aorus Xtreme Waterforces, but itching to add a 3rd and 4th soon since rendering animations is still an overnight ordeal typically :/ Im running out of room for more radiators (CPU is also AIO cooled) so might have to mix in a blower GPU as my 3rd and 4th. As long as theyre RTX 2080Tis theres no issues with mixng manufacturers right?

Eric

And with that, if I were to get a faster card like a titan or quadro, will different arrangements of PCI slots matter, just the same if I throw it in 3rd?

I have the Aorus TRX40 Xtreme MB, I think it has same lanes across them but will need to double check.

Alex Glawion
Alex GlawionCGDIRECTOR

Hey Eric,
Yes you can mix manufacturers and even different tiers of gpus, but will have best performance with gpus that have the same amount of vram. Blower gpus perform well too, though will be somewhat more noisy than aio cooled gpus.

The #of pcielanes of the Aorus TRX40 Xtreme from top to bottom is 16, 8, 16, 8 as in most boards with 4 slots so put your fastest gpus in the x16 slots.

Cheers,
Alex

Joey Curry

Hey Alex,
I am in the process of adding some more render power to my computer and have a couple of questions. Can you mix up the kinds of gpu’s or do they all need to be the same kind? Like two 2080ti’s and two external 1080ti’s on one computer? My current configuration is using two 1080ti’s. I was considering replacing them with two 2080ti’s and placing my existing two 1080ti’s in an external housing connected via USB 3. Do you know if these will all work together and would I still be able to take advantage of the RTX acceleration? Or will the mixture of different gpu’s disable this feature?

Alex Glawion
Alex GlawionCGDIRECTOR

Hey Joey,
Good question! Normally you cann mix and match similar enough gpus like the 1080ti and 2080ti without any trouble, as they have the same amount of vram and will just pick up a render bucket to process. They don’t necessarily have to be the same type / performance.

External gpus too should be no problem, apart from the fact that the bandwidth is greatly decreased when you hook them up over usb. So you might see the external gpus perform considerably slower than when they are mounted internally.

Octane certainly has a fallback when raytracing cores are not present on a gpu, though i am not entirely sure if this is per gpu or per scene / per frame rendered. So this might not work when rendering one frame acrosse 4 gpus. You might have to render 2 frames at once in a sequence for example and have the 1080tis render one frame and the rtx capable gpus render the other frame.

Cheers,
Alex

Sean Sullivan

Hey Alex!! I’ve already been posting on another page on your great site here, figured I’d come join another thread. oops, maybe it’s all the same one. lol
Anyways, a couple of questions here for you again…. on these octane bench marks…
Seems like even though the GTX series came out more recently than the RTX ones, the RTX ones out perform the newer GTX ones? is that so? Strange if thats the case. Also, I saw a video on yt about how the GTX series was kinda not that big of a difference and even made the case that it was kinda a rip off when comparing them to the RTX’s. Seems like the octane benchmarks support this case.

Also, what kind of case and mother board supports 8 cards and can you do that? Or, is it just a simulation for the benchmarks.. as if you had 8x RTX 2080 Ti??
I’m hoping that’s a possibility, I’d love to see how fast a stack of 8 GPU’s would perform!

Next build I’m going for the best price per dollar performing card i can get!

Alex Glawion
Alex GlawionCGDIRECTOR

Hey Sean,
Nvidia likes to release their strongest cards of a generation first and after that scale down their chips for the mid-range and lower-end market. The RTX cards are the mid to high-end, and the gtx 166xx(ti) cards that came out later, are targeted at mid-range and lower end use. That is why you see “newer” cards (of the same scaled down architecture and generation) perform worse here.

There is no mainstream board that you can get off a shelf supporting 8 gpus. Those benchmarks with 8 gpus have been run on servers optimized for a mindboggling number of supported gpus.

Cheers,
Alex

Brandon

I am running octane with one Nvidia Gtx 1070, I’m wondering if i can add another 1070? New to computers/hardware compatibility. Please let me know!

Alex Glawion
Alex GlawionCGDIRECTOR

Hey Brandon,
Usually you can, though it depends on what kind of motherboard, case and PSU you have. What are your system specs?

Cheers,
Alex

Brandon

Intel Core i7-8700K 3.7GHz, 16GB RAM, 3TB HDD + 240GB SSD, NVIDIA GeForce GTX 1070 Ti 8GB, Liquid Cool,

Thank you!

Scott Gammans

Hi Alex,

Any experience with EVGA cards? GTX 1080 Ti’s are rarer than hen’s teeth but I’ve got a new EVGA GeForce GTX 1080 Ti on order and was wondering if you had thoughts…

Thanks!

Alex Glawion
Alex GlawionCGDIRECTOR

Hey Scott,
Yes as far as I know there are not many of the 10xx generation that are still being manufactured massively. Nvidia wants us to buy either the RTX or the GTX 16xx series of GPUs.

The 1080Ti is a great GPU IF you can find it at a reasonable price nowadays. I’d say anything around 700$ is a good deal on these GPUs. EVGA GPUs are usually top of the Line. They tend to be a bit more expensive, but they are built well and the support and RMA is among the best in the industry.

Cheers,
Alex

Scott Gammans

Hi again,

The 1080Ti card turned out to be a reconditioned unit. I canceled the order and bought an EVGA RTX 2080 Ti FTW3 Gaming instead, which arrived today and has been installed (along with an uprated Corsair AX1200i power supply).

After running EVGA’s boost tuning, I have to say, I am not seeing the OctaneRender performance boost over my Quadro P4000 that I was expecting. A 4K test scene that I render with just the P4000 finishes on average in about 156 seconds (this is also the card that is performing Windows 10 display duty). The RTX 2080 Ti, which is solely being used for GPU rendering, does the same scene in around 75 seconds. That’s only a little over twice as fast, but the 2080 Ti’s OctaneBench of 304 compared to the Quadro’s 115 led me to expect that I’d see more like 2-1/2 times the improvement in speed (304 / 115 = 2.64).

I’m running the OctaneRender 2018.1.3 plugin for LightWave and not the standalone OctaneRender… is the performance of the plugin not comparable to the standalone version? Or do I maybe have a bad card? (It overclocked to 1920 MHz without any problems, and while monitoring the card while the test scene was running I didn’t see any throttling.)

Any idears?

Alex Glawion
Alex GlawionCGDIRECTOR

Hey Scott,
Do a benchmark run and see what the score on your 2080ti is. It might very well be that the Scene you are rendering has some (cpu-bound) preparation time involved, which lengthens the render duration before the bucket rendering phase starts?

Cheers,
Alex

Scott Gammans

Hmm, yeah something is definitely off here. I ran OctaneBench and only scored a 260 (vs. the expected 304). I also tried MSI’s Afterburner for tuning and it came up with the same results (1920 MHz OC).

Alex Glawion
Alex GlawionCGDIRECTOR

Hey Scott,
Some things from the top of my head: Are you running Octanebench 4? Do you have both power cables plugged into the GPU, and the PSU is strong enough? What kind of Drivers are you using? The Game Ready Drivers for the RTX and other drivers for the Quadro? Try taking the quadro out and see if only the 2080Ti scores better.

Cheers,
Alex

Scott Gammmans

Hi Alex… thanks for your help.

* I am running OctaneBench 4.
* Both 8-pin power cables are plugged into the GPU.
* Brand-new 1200 watt Corsair 1200AX PSU,
* Running the latest NVIDIA “Studio” driver 430.86, and NOT the latest “Gaming” driver.

I am at my wits’ end here. The card is actually running SLOWER than last night now… render times are now between 87-91 seconds. I’ve uninstalled all of the overclocking software that had been installed, but now the GPU is acting like it’s throttling itself… the fans hardly speed up at all when I hit F9 (and why should they; the dang GPU is hardly exerting itself now that it’s running slower than before). Out of the box with no “optimization” the thing was rendering in 76-78 seconds. Do those overclocking programs flash the BIOS in the GPU?

I’m going to turn my computer off and go have some lunch to let the GPUs cool off and I’ll try your suggestion to remove the P4000. But one last item: Could the M.2 SSD be an issue? I seem to recall reading somewhere you cautioned against putting the system drive in the same PCIe lane as the GPU. How would I check for that? In the Gigabyte X399 Designaire’s boot menu?

Thanks!!!

Scott Gammans

OK, this is downright confounding. After removing the Quadro P4000, my OctaneBench score jumped to 290… still not near the global average, but better. But my test scene render is completely unaffected–it still renders in 83-85 seconds.

One thing I noticed is that the fans didn’t run hardly at all during the OctaneBench run this time around–I’m guessing that’s because the heat of the P4000 is no longer contributing to the overall thermal load in the environment. (The fans also hardly blip when rendering with the OctaneRender LightWave plugin.)

After I took out the Quadro P4000 I checked the Gigabyte motherboard setup utility and didn’t see anything related to where the M.2 SSD is PCIe lane-wise; the system BIOS is set to assign PCIe assignments automatically. Physically speaking, I installed the Sammy SSD in the M2Q_32G slot and the RTX 2080 Ti (which is now the only system GPU) in the prime PCIEX16_1 slot.

thanks…

Scott Gammans

OK Alex, never mind. I think this card is a dud. Now it won’t even run stably at its default settings. Returning it to Amazon and gonna try something else. Thanks for your help anyways!

Alex Glawion
Alex GlawionCGDIRECTOR

Hey Scott,
That’s really weird. Yes best to get a replacement to be sure its not the gpu.

Cheers,
Alex