Page 1 of 1

Is OctaneBench predictive of LightWave plugin performance?

Posted: Mon Jul 15, 2019 4:19 pm
by scooternva
Hi,

I just installed a new RTX 2080Ti card and benchmarked it with OctaneBench 4.0. I scored 295, which is pretty good for an unoptimized (no custom overclocks) card, and a healthy 159% bump in performance over my Quadro P4000’s score of 114. But when I render a 1M polygon test scene at 4K resolution with the LightWave 2019.0.3 plugin (OctaneRender version 2018.1.3), the RTX 2080Ti takes 75 seconds vs. 156 seconds for the P4000–a much more modest increase in performance of 108%.

The 4K scene test runs are identical in all respects: same model, same materials, same Octane render target settings. So I’m left to wonder, how should I interpret these results? Is the LW plugin slower than the standalone OctaneRender? Are the benchmark scenes using render target settings different from my 4K test scene that don’t scale linearly? I am thrilled with OctaneRender’s performance; I am just trying to understand why Im getting a 108% bump in performance with my new GPU in LightWave versus the expected 159% bump in render speed. Thanks!

Re: Is OctaneBench predictive of LightWave plugin performance?

Posted: Mon Jul 15, 2019 5:04 pm
by juanjgon
Hi,

The Octane render engine is the same in all the plugins, the Standalone and the OctaneBench app. Only be aware that for example, the RTX acceleration is not available in the plugins yet, and the speed difference between two GPUs can also depend on the scene complexity or characteristics.

Thanks,
-Juanjo

Re: Is OctaneBench predictive of LightWave plugin performance?

Posted: Mon Jul 15, 2019 7:48 pm
by scooternva
RTX acceleration is not available in the plugins yet
Thanks; that makes perfect sense and explains why the RTX 2080Ti GPU is only twice as fast as the P4000 with my “real world” scene, and not 2-1/2 times as fast (the delta between the GPU cards on OctaneBench). I also suspected what you confirmed regarding scene characteristics (that my “real world” scene is going to have different content and settings that won’t necessarily scale from one card to the other at the same linear rate as the OctaneBench scenes).

Apologies if this has been asked before, but when will RTX acceleration debut in the plugins?

Re: Is OctaneBench predictive of LightWave plugin performance?

Posted: Tue Jul 16, 2019 10:47 am
by juanjgon
In the 2019.2 plugin builds, later this summer, as far as I know.

Thanks,
-Juanjo

Re: Is OctaneBench predictive of LightWave plugin performance?

Posted: Wed Jul 17, 2019 5:29 pm
by scooternva
Thanks. Since I am on the Studio subscription for the LightWave plugin (and currently running OR 2018.1.3), will that update be available to me on my current plan or will I have to upgrade to a different level to get this “2019.2” version?

Re: Is OctaneBench predictive of LightWave plugin performance?

Posted: Wed Jul 17, 2019 5:51 pm
by juanjgon
As far as I know, if you have any subscription plan active, you can use any of the new Octane versions.

Thanks,
-Juanjo

Re: Is OctaneBench predictive of LightWave plugin performance?

Posted: Wed Jul 17, 2019 9:47 pm
by scooternva
Great! OctaneRender is already ridiculously fast (a 4K scene in 50 seconds? no problem!) so it’s nice to know there might be even more acceleration around the corner. Thanks!