View Full Version : LightWave 2018 VPR

05-09-2018, 10:29 AM
I'm looking into purchasing a new workstation I was wondering does the VPR Update faster with higher core count or with faster cores

05-09-2018, 07:09 PM
I'm looking too and I was considering the Ryzen 1700X with 8cores 16 threads. So this is a good question that I'm very interested hearing responses. My gut says go with more cores when it comes to anything involving CPU rendering, but let's see what the experts say. :)

05-09-2018, 08:59 PM
Hardly an expert but been researching for awhile now. Take a look at the vid in this collection titled Professional Overclocking, ~ 2/3 down the page.


05-10-2018, 08:08 AM
I'm no expert, but I'd say "both". :)

Here's a screen shot of the Performance screen in Windows Task Manager when I'm holding down ALT and rapidly rotating the VPR Perspective view with the mouse. While doing so, it utilizes all eight cores, so the VPR rendering correctly uses multi-cored processors. And, of course, depending on how many objects / surfaces it has to render, the faster the cores the better.



05-10-2018, 12:45 PM
Yep - that makes sense because they say VPR "is" the renderer in 2018. I still wonder if say a 3.4 ghz with 4 cores outperforms say a 2.4 ghz with 8 cores.

05-10-2018, 01:11 PM
it's the same rule as in LightWave 2015 really.

so as long as it is "fast" it really doesn't matter

higher core count is often a good thing, though not always.

check the web to see what CPUs are the best ones versus price.

examples >

05-10-2018, 02:20 PM
just bought a used i7 3930k
overclocked it is close to the speed of a Ryzen 1800X

only way cheaper

05-10-2018, 04:04 PM
Yep - that makes sense because they say VPR "is" the renderer in 2018. I still wonder if say a 3.4 ghz with 4 cores outperforms say a 2.4 ghz with 8 cores.

Well, if the code being run is 100% "parallel" (i.e. it can run simultaneously on multiple processors resolving different portions of a computational problem without synchronization issues), and all other parameters are kept equal (e.g. processor architecture and instruction set, amount of memory and memory access time, cache size and speed, etc.) then obviously 3.4 * 4 = 13.6 aggregate Ghz, while 2.4 * 8 = 19.2 aggregrate Ghz, so the 2.4x8 wins.

If the code to be executed must be executed sequentially due to the nature of the "problem space", then the 3.4Ghz would win, since only one processor (core) could be used.

Without knowing how "parallel" the VPR rendering engine is, it's not really feasible to say how it would perform on these two systems.

Usually it isn't that simple. Usually, standardized benchmarks concentrating on particular performance requirements (e.g. 3D rendering) are used to determine the relative performance of competing systems within a market.


05-10-2018, 09:14 PM
Thanks - sounds complicated. I don't build my boxes I buy ready made off the shelf and just look at the basic performance score on Passmark as my guide for CPU/GPU. Motherboards don't mean anything to me. https://www.passmark.com/index.html