VRAY Benchmark & CPU + GPU Scores (Updated Results)

CG Director Author Alex Glawionby Alex Glawion   /  Updated 

Chaos Group’s V-Ray Benchmark is a widely used tool for benchmarking your Processor and your Graphics Card.

If you are thinking of building a Computer for 3D Modeling, would like to see what GPU performs best in V-Ray, or what CPU is best for your rendering needs, this benchmark is extremely helpful.

V-Ray 5 CPU Benchmark Results

Apart from the XEON Dual CPUs, these are all single CPU Results.

= AMD   |    = Intel

CPU Name# CoresGhzVRAY CPU Benchmark vsamples
AMD Threadripper 3990X642.973025
AMD Threadripper 2990WX323.027760
AMD Ryzen 9 5950X163.424733
Intel i9 10980XE182.622840
AMD Ryzen 9 3950X163.519077
AMD Ryzen 9 5900X123.718659
Intel i9 10940X143.115213
AMD Threadripper 2950X163.515996
Intel i9 10920X122.915100
AMD Threadripper 1950X163.415397
AMD Ryzen 9 3900X123.814706
AMD Ryzen 7 5800X83.812275
Intel i9 10900X103.311386
Intel i9 9900K83.611068
AMD Threadripper 1920X123.511039
AMD Ryzen 7 3800X83.99888
AMD Ryzen 7 3700X83.69829
AMD Ryzen 5 5600X63.79506
AMD Ryzen 7 2700X83.78904
Intel i7 8700K63.77965
Intel i7 8086K63.77756
AMD Ryzen 7 1800X83.67706
AMD Ryzen 7 1700X83.47472
AMD Ryzen 7 270083.27342
AMD Ryzen 7 170083.07234
AMD Threadripper 1900X83.86760
AMD Ryzen 5 2600X63.66056
AMD Ryzen 5 160063.25672
Intel i7 7700K44.25232
Intel i5 9600K63.74784
Intel i5 840062.84223
AMD Ryzen 5 1500X43.53486
AMD Ryzen 5 140043.22675
CPU Name# CoresGHzVRAY CPU Benchmark vsamples

V-Ray 5 GPU CUDA Benchmark Results

As we can deduct from the GPU Scores, scaling with multiple GPUs is not linear. Benchmarking time with 2x, 4x or 8x RTX 2080 TIs does not scale perfectly.

These Scores have been taken in combination with an Intel i9 10900K, except for the Multi-GPU Benchmarks.

GPU NameVRAM (GByte)V-Ray GPU CUDA vsamples
4x NVIDIA RTX 2080 Ti113314
NVIDIA RTX 3090 Ti242153
NVIDIA RTX 3090242091
NVIDIA RTX 3080 Ti122054
2x NVIDIA RTX 2080 Ti111902
NVIDIA RTX 3080101760
NVIDIA RTX 3070 Ti81510
NVIDIA RTX 307081409
NVIDIA RTX 3060 Ti81238
NVIDIA RTX 2080 Ti111004
NVIDIA RTX 306012873
NVIDIA RTX 2080 Super8855
NVIDIA RTX 20808821
NVIDIA RTX 2070 Super8775
NVIDIA GTX 1080 Ti11689
NVIDIA RTX 20708650
NVIDIA RTX 2060 Super8534
NVIDIA RTX 20606498
NVIDIA GTX 10808452
NVIDIA GTX 1660 Super6436
NVIDIA GTX 1660 Ti6418
NVIDIA GTX 16606387
GPU NameVRAM (GByte)V-Ray GPU CUDA vsamples

Download V-Ray Benchmark here

The V-Ray Benchmark is free to use and will put your Graphics Card and your Processor through a series of Rendering Tests.

Get the V-Ray 5.00.01 Benchmark Version we use in this article here: Download V-Ray 5.00.01 Benchmark

How to read these Scores

Many other Rendering Benchmarks, such as Octane Bench or Cinebench, assign a Score to the Benchmarking Result.

V-Ray, though, tells you the number of “vsamples” it was able to take within one minute. Of course, the higher the vsamples, the better.

Because GPU Rendering Performance relies so much on various factors such as PCIe-Lanes, CPU Clock Speed, and Number of Cores, it is important to note that all GPU Benchmarking times listed above were taken with an Intel i9 10900K CPU (Single-GPU Scores).

If you have a different CPU, it might very well be that your Benchmark Score Results are off by a few hundred samples or even more, depending on your setup.

For GPU Rendering, it’s better to get a high clocking CPU that doesn’t necessarily have many Cores to make your GPUs render at the maximum possible performance. Check our GPU Render Guide for an in-depth explanation of this.

CPU vs. GPU Rendering in V-Ray Benchmark

This question comes up frequently: CPU or GPU rendering? What is better?

Well, the answer depends on what you want to render. Here’s why:

The great thing about GPU Render Engines is that you get responsive or even Real-Time Feedback of your Scenes and can iterate and tweak your Settings much quicker and more often than when rendering on the CPU.

More iterations mean higher quality end-results.

Of course, this only works if you don’t need any kind of features that either don’t work well on the GPU or take a long to time to prepare, like huge displacements, Motion Blur, or Meshes that are so big, they or the Scene don’t fit into your VRAM.

If you have to swap to System RAM, things get slow.

There were times when GPU Render Engines were just not mature enough and lacked many of the features that CPU Render Engines offered.

But nowadays, this argument is fully valid anymore. GPU Render Engines such as V-Ray, Octane, Redshift, Furryball, and many others have reached the same level as CPU Render Engines and have even overtaken them in some cases.

V-Ray Benchmark Results: Interesting Findings

Multi GPU Setups don’t scale well in V-Ray. With 2 or 4 GPUs you’ll see lower performance per GPU than when running just a single GPU.

In V-Ray’s CPU benchmark you’ll have to make sure to use Benchmarking Scores and not Cores for selecting your new CPU for Rendering. Because individual Core Performance depends a lot on the CPU’s architecture and IPC, you’ll find CPUs that are much faster than others even though they have the same number of cores. There are even CPUs that score higher even though they have fewer cores.


What GPU & Computer are you buying and building?

CGDirector is Reader-supported. When you buy through our links, we may earn an affiliate commission.

Alex Glawion

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!


Also check out our Forum for feedback from our Expert Community.

Leave a Reply

  • vex_cro

    Did someone compared V-Ray and Redshift with displacement maps and vram usage? V-Ray sucks vram almost as Arnold gpu. Is Redshift more optimized with this?

    • vex_cro

      Yes, V-Ray deffenetly sucks vram as Arnold. That’s a bad thing. And it has strange hang before render starts, also as Arnold. Arnold now have denoiser for cpu rendering…

      • vex_cro

        Hello, V-Ray has a hang in the beginning of rendering but after that is quite fast. Now i want to buy rtx 3090 for my 4790k cpu. Will there be bottlneck in rendering? I don’t mind bottleneck in gaming…

        • Alex Glawion

          Should be fine. PCIe 3.0 x16 shouldn’t bottleneck the 3090.


          • vex_cro

            Hello, Sorry for bothering but i have one more question. Would scene preparation be much faster on more powerful processor?

            Thank You.

        • Alex Glawion

          Hey Vex_Vro,
          Scene Preparation is single-core performance dependent. So if by “powerful CPU” you mean a CPU with high single-core performance, then yes.

          More cores will not benefit preparation time.


  • ammar

    hi alex..
    i am trying to build my workstation with a tight budget.. around 1000$
    and i am wondering if you have any suggestions for me..
    i work with 3D Max _ v-ray next and photoshop ..

  • Lenny

    I’m really confused, why is the Threadripper 1950x ranking so high? this can’t be…

    • Alex Glawion

      That really seems off. I’ll do another benchmark of the 1950X to confirm. Thanks for the heads-up!

  • Victor Gama Sabbithi

    Hi Alex, I am curious if there are any bench marks for AMD GPU’s for 3ds max and Vray. With most of AMD cards being more affordable- however I am not seeing much on bench marks – is there a reason for this?
    Also, I really want to upgrade from my RTX 4000 – which tends to crash quiet a bit and has a hard time handling large scenes. Do you have any suggestions?

    Thank you

    • Alex Glawion

      Hey Victor,
      Most Vray Versions do not support AMD GPUs. You can check our compatibility post here: https://www.cgdirector.com/render-engine-hardware-compatibility-cpu-gpu-hybrid/

      The best value GPUs for CUDA rendering in vray are currently the recently released RTX 3070 and RTX 3080 GPUs. They are quite hard to come by though as they are so popular and stock is low. See if you can get your hands on one of those.


      • Victor Gama Sabbithi

        Hi Alex,

        Thank you for you reply. 🙂 I will try to get my hands on one of the new RTX 3000 series.


  • sadegh

    sorry alex…
    for my last question , i wanna make it simple..
    Do PCI-E Lanes Matter For GPU Rendering?
    if u can help me , tell me abaut pcie x4 and x1(wifi port)
    because i must choose between these 2. i dont ve thunderbolt port

    • Alex Glawion

      Hey Sadegh,
      PCIE-Lanes matter in most cases. It depends on the GPU you are using. If the GPU isn’t fast enough to saturate a certain amount of pcie-lanes’ bandwidth, then having more pcielanes will do nothing for performance.

      For all RTX / GTX 20xx gpus or below, x8 pcielanes is enough (2080ti barely saturates x8)
      For RTX 30xx Series GPUs you should make sure to have x16 pcie3.0 lanes

      If all you have is pciex4 and x1, obviously you should chose the x4 slot. Then again,the x1 slot will probably not be a mechanical x16-slot and won’t even fit your gpu at all. But a x4 slot will heavily throttle a 3080. Not recommended.


      • sadegh

        thank you alex
        i got it

  • Load More Comments