What's your KeyShot Benchmark score?
0.76 for i7-10750H and 25.4 for RTX 2060
Based on the average score posted here, I believe the numbers are close to this:
| 4090 | 200 |
|---|---|
| 4080 | 150 |
| 3090 | 100 |
| 3080 | 90 |
| 4070 | 90 |
| 3070 | 60 |
| 3060 | 40 |
| 3050 | 25 |
| 2060 | 25 |
| 4090 + 4090 | 370 |
| 4090 + 3090 | 245 |
| 3090 + 3090 | 125 |
Think thatâs valid, with multiple cards scores should end up like double with 2x same card. At least thatâs my experience with other benchmarks. Depending on CPU it could take a bit longer loading the data into the GPU but while rendering the speed is double.
With the KeyShot Benchmark the loading time exceeds the rendering time sometimes which give not really reliable results. Rendering normal scenes as a test and record the time to reach an x-amount of samples (high number) would give a better indication when using multiple cards.
an ongoing little project I have is this, 100% gleaned from this thread-
and for CPUs, its a little more complicated.
Thatâs a nice list! Makes clear why you shouldnât buy a âproâ'series card if your main usage is renders (or gaming ;))
CPUs are really a different story. Currently I donât see any real benefits rendering with CPU at all. That is, if youâve a decent GPU and your scene will fit in the VRAM but if you use textures a bit smart you can do a lot with 12-24GB.
With V-Ray there is also a hybrid option for CUDA so your CPU can help render but the difference with my CPU is so small I always had the feeling my CPU (i9-9900K) was more a delay factor than actually shorten the render time.
It would be nice to know if cards with RTX (Geforce 20xx series and up) would get also an optimized version of the GPU renderer instead CUDA to make optimal use of the hardware raytracing capabilities. In some other renderers you can select CPU/CUDA/RTX.
Not that important for stills but with animation 30 seconds a frame is already 12 minutes on every second rendered (24 fps).
anytime people add to this forum post I try to add it to the list if its not on there already. at some point when I get some free time I might head to a mac store and microcenter with a usb stick wth KS benchmark on it and see what trouble I can get myself into.
LOL that reminds me when I went into a Mac store actually buying a MacMini to be used as media-center. They just switched from Motorola processors to Intel where they always spit on and I told the young guy in the shop it was a smart move they switched to Intel since the Motorolas couldnât really keep up for a long time. He thought it was a stupid move and looked pretty angry at a customer who was about to buy an Intel MacMini 
I think he always remembered since every time I saw him in the streets of the city centre he looked angry at me. So be prepared 
Can anybody share benchmarks of 4060ti & 4070 Please? I want to go for any of them but am not sure about the benchmarks.
Hey @as.if,
These are scores from CineBench 24 but that wonât really make a difference if you want to compare the speed of the cards.
As you see they are quite close but since youâre planning to render with a GPU I would advice you to consider especially the VRAM since that will be the limiting factor of the size of the scenes you can render.
Iâm not sure about second hand prices for a 3090 where you live but that card is fastest of them (well, close) but it has 24GB of VRAM. Which is currently still most you will get on a consumer card.
In The Netherlands the price of a second hand 3090 is close to the 4070 and personally I would go for the 3090 since it has double the VRAM.
The only disadvantage is the fact that officially the 3090 doesnât support frame generation which can be useful in some games but Iâm not aware of renderers actually using that feature (could actually save time rendering animations, so who knows about the future).
Thank you very much @oscar.rottink
I think I might go for 4060ti 16gb version brand new. But your post helped me to decide.
I didnât know you had also a 4060Ti with 16GB, thought they were all with 8GB. Must say NVidia is creating a bit of a mess with the current generations of cards. The introduction of the âSUPERsâ didnât make it more logical
I do think the 4060 has a bit slower memory installed if Iâm correct but thatâs not really a big impact on rendering times, just takes a little bit longer getting the scene into the VRAM. I think with games where it constantly swaps things in and out memory it has a bigger impact.
KeyShot Viewer 2024.1 Benchmark Result
CPU result: (12th Gen IntelŽ Core⢠i7-12700K - Threads: 20) 2.81
GPU result: (NVIDIA RTX A4000 with Driver: 537.13) 61.30
KeyShot Viewer 2024.1 Benchmark Result
CPU result: (12th Gen IntelŽ Core⢠i7-11700K - Threads: 20) 2.81
GPU result: (NVIDIA RTX A3000 with Driver: 537.13) 61.30