New Ryzen Processors and VRay

[mohammadreza_mohseni]

and this is the scene I am using for benchmarking for vray since 1.5 years ago (which include vray instancer, dof, motion blur, glossy material, etc... )




thanks Vlado for this benchmark it will be really a great help.

[vlado]

Vlaod, will the team release the benchmark scenes for both Max and Maya or?

The benchmark is a standalone application, it doesn't need Max or Maya.

Best regards,
Vlado

[Raven]

Hey Vlado

Will be great to have an official benchmark, but why different scenes for GPU and CPU.
Would be kind of interested in seeing the same test for both to gauge how certain cards compare to processors.

[vlado]

Would be kind of interested in seeing the same test for both to gauge how certain cards compare to processors.

This came up and we even had fights about it internally. Right now we don't want to get in the middle of the CPU vs GPU thing. Later on we might be in a better position to compare the two types of hardware directly.

Best regards,
Vlado

[Jiri.Matys]

Maybe some more complex scenes will be better. Maybe doesnt matter, can compare procesors anyway, but more complex scenes makes bigger differences.

[vlado]

Maybe some more complex scenes will be better. Maybe doesnt matter, can compare procesors anyway, but more complex scenes makes bigger differences.

We have more complex scenes, and we considered it, but then the benchmark would run for too long and people would lose patience. The target was about 4-5 mins on an average i7 with 8 logical cores for the CPU. As the average hardware gets better, we will update the scenes accordingly.

Best regards,
Vlado

[theedge]

4-5 MN for the whole benchmark or for each scene?

[cmjohncheng]

We have more complex scenes, and we considered it, but then the benchmark would run for too long and people would lose patience. The target was about 4-5 mins on an average i7 with 8 logical cores for the CPU. As the average hardware gets better, we will update the scenes accordingly.

Best regards,
Vlado

maybe no need to noise free render or smaller resolution?

[vlado]

Yeah, will see what can be done for future versions. But even now the scenes should be good enough to test hardware performance.

Best regards,
Vlado

[eligiusz]

Double CPUs perform worse on heavy scenes than single CPUs, but I'm not sure if benchmark should measure that or just pure CPU power.

[vlado]

Double CPUs perform worse on heavy scenes than single CPUs, but I'm not sure if benchmark should measure that or just pure CPU power.

Sometimes I've seen the opposite actually

Best regards,
Vlado

[Morne]

Maybe some more complex scenes will be better. Maybe doesnt matter, can compare procesors anyway, but more complex scenes makes bigger differences.

We have more complex scenes, and we considered it, but then the benchmark would run for too long and people would lose patience. The target was about 4-5 mins on an average i7 with 8 logical cores for the CPU. As the average hardware gets better, we will update the scenes accordingly.

Best regards,
Vlado

I agree with both comments. People will loose patience with waiting too long. But also, a benchmark that only takes 4-5 minutes doesn't tell me anything about the hardware, just tells me it's an "empty" scene that is nowhere near my usual scenes.

[cmjohncheng]

I agree with both comments. People will loose patience with waiting too long. But also, a benchmark that only takes 4-5 minutes doesn't tell me anything about the hardware, just tells me it's an "empty" scene that is nowhere near my usual scenes.

cinebenchR10-15 benchmark tool maybe just take 1-2 min, but it is quite popular benchmark tool too.
if just specific for vray benchmark, maybe more complex scenes will be better, like included displacement, fur, motion blur and etc(not so sure intel and amd render these feature is it same as normal render)

[vlado]

if just specific for vray benchmark, maybe more complex scenes will be better, like included displacement, fur, motion blur and etc(not so sure intel and amd render these feature is it same as normal render)

The scenes do include some of these features. But really beyond a certain point, it doesn't matter what's in the scene - the performance scale would be the same.

Best regards,
Vlado

[vlado]

The first version of the benchmark is now available in the download section of our website. More details here:
https://forums.chaosgroup.com/showth...rk-1-0-release

Best regards,
Vlado