Tweet
Hello! im trying to get a better understanding of render-speeds between GPU and CPU using the vray benchmark page. However, i dont understand how i would compare the results when they all use different score-naming. How do i compare vsamples VS vpaths for example? What even is a vsample, a vpath and finally a vray? What i really want is to see how our GPU rigs (using Vray-GPU) compare speed-wise to different CPU configs. How could i do that? Thanks!
-
-
I guess results between the different rendering methods cant really be compared anyways, because theyre rendering different scenes. Does anyone have any data on the differences?
- What CPU rig would match 8x 3090?
- What CPU rig would match 4x 3090?
(In terms of rendering speed) -
Hello tobias_eek
On the benchmark page you will find Cuda numbers for many different CPUs. It should could you a conclusion for V-Ray Cuda mode
For example an RTX 3090 scores around 1900 in Cuda, and Threadripper 3970X 32-core scores around 900
So 8x 3090s would match 16x Threadripper CPUs or 512 physical cores roughly
That being said, a machine with 8x 3090 will need a special setup for this linear scaling/cooling/bandwidth..etc
Best,
MuhammedComment
-
Where on the benchmark site can i see cuda scores for CPU? Sorry i just dont see it? I also see in the ''understanding vray benchmark''- post, it is recommended to compare CPU vs CPU and GPU vs GPU.
- Also, why does the different methods, rtx, cuda, cpu have different scoring conventions?
- Also, what do the different scoring conventions mean? What is a vsamle, what is a vpath, and what is a vray?
ThanksComment
-
tobias_eek, here is an example search query for CUDA results of AMD CPUs: https://benchmark.chaos.com/v5/vray-...=false&gpu=AMDComment
-
-CPU Benchmark represents our V-Ray Renderer productOriginally posted by tobias_eek View Post
-Cuda Benchmark represents our V-Ray GPU product, in Cuda mode
-RTX Benchmark represents our V-Ray GPU product, in RTX mode
V-Ray and V-Ray GPU are different render engines, the latter was wrote from scratch and optimized for speed to run on GPUs. All 3 are production proven and part of the different integrations, V-Ray for Maya, V-Ray for Houdini..etc
V-Ray GPU is able to run on CPU devices, which we call Hybrid rendering. It can produce identical results to rendering on GPUs, so you get a rough idea of how GPUs compare to CPUs in speed
Different ways of describing the amount of rays/samples reached per pixel in one minute of rendering, they are named differently as the benchmark uses different scenes for CPU, Cuda and RTXOriginally posted by tobias_eek View Post
Best,
Muhammed
Comment
Comment