ZBrush CPU Performance, Is Higher CPU Frequency or more cores better and for what tasks

[Bevents]

Hi

I don’t know what’s more valuable for ZBrush performance in a CPU, and for what tasks. the ZBrush system spec page it’s states that “A fast processor with multiple cores is a key item because ZBrush is only CPU based. The more powerful, the better”, but I’m looking for a more in-depth answer to this.

As an example, let’s say I was looking at the Intel 9900k and the AMD Ryzen 3900x. The 9900k has 8 cores/16 threads and a top frequency of 5Ghz, the 3900x has 12 cores/24 threads at a lower 4.6 top frequency. Which is more important to have and for what tasks, Higher GHz and less cores or more cores at a lower frequency?

I know that in general for CPU renderers more cores/threads are more important and is likely the case for BPR and Keyshot for ZBrush.
But what about when your Sculpting in the viewport, or zoomed in, or working with a high poly/ high subdivision subtool/multiple subtools. I know for apps like 3Ds Max or Maya that higher frequency single core is more important for working in the viewport but is that the case for ZBrush too?
Are brushes in use benefitted by higher speed cores or more cores? What does Fibremesh or the new cloth simulation benefit from more? What about Dynamesh and Zremesher.

I’m hoping to get some answers about these tasks and more, so that I and others can better select the best CPU for our use cases.
For example if I concerned with the best performance within the viewport sculpting high poly assets for use with real time solutions (like Ue4 or marmoset). Or perhaps someone was going to model a shoe to then render and use as a product ad. Or An 3D illustrator who likes to make furry fantasy creatures with fibremesh.
I imagine what they would benefit and what cpu they should look to would change depending on the use case.

I’m hoping I can get answers about what’s more important to have, and for what.

Thanks

[tobor8man]

I had the same question over a year ago when looking at the 9900K and Threadripper 2950X. See the discussion on reddit here CPU discussion

FWIW - I ended up getting the 9900K with 32 GB ram. Have been very happy with it. (Of course I haven’t been able to do a personal direct comparison with the AMD chip). Note: make sure you get an NVE SSD as your main ZBRush drive. I have a 1 TB SSD as my C drive (Windows). Zbrush does a lot of file I/O and also saving/loading is so much faster. I rarely have to wait more than a second or two when saving tools with 50M polgygons.

[Bevents]

It was actually seeing your reddit post that made me want to ask this question.

On that post XYZ3D says
"Now if he were to render with Keyshot, you would see the cpu jump to 100% with all cores being used. That’s where the threadripper is worth it.

So again while he’s working in zbrush, cpu usage is 15% even though all cores are being used to render, zbrush itself isnt really benefiting from all of those cores like you would think because there are other functions in the program such as brushes, vertex translation etc that arent as threaded and the cpu has to wait for those threads to do their work. The more frequency you have, the less you have to wait. So yes cores do matter but frequency matters more. Of course a 2 core cpu would be a bad idea, but a 8 or 10 core at 4.5 to 5.0ghz would be far more efficient than the threadripper. To give you an idea, my 6 core i7-4960x which is from 2013 is using 80 to 100% cpu usage while working on 40 million polygon subtools. I have Ztools with 200 million polys and I work at 4K. Could I benefit from a 10 core cpu? Absolutely, but 32 core? It doesn’t scale as effeciently as we would hope. Not for that cost."

But later on in that post cglink posts a link with a quote from Pixologic support staff Aurick which states:
CPU influences how smoothly you interact with the mesh as you are sculpting. So the better your CPU, the better your sculpting experience will be as you increase the polygon count. Here, more cores is typically better than faster core speed. An easy way to compare processors is to take the speed and multiply it by the number of cores. Doing this for each CPU will give you a value that you can compare. That with the highest score, wins. This means that a 6 core CPU at 2.5 GHz would have a score of 15 while a 4 core CPU at 3 GHz would only have a score of 12. The former edges out the latter even though each core is slower."

These two quotes i’ve taken contrast one another a little, and whilst Aurick is Pixologic support which adds credibility to his statement, he does say "more cores is typically better than faster core speed" which leads me to believe that depending on task and use case that answer of what is a better CPU might change. Again leading to to the question of what Zbrush tasks value what more in CPU performance.

I agree that a large amount of ram, potentially at higher clock speeds is important as aurik does say "When it comes to ZBrush, RAM influences how many polygons a SubTool can have and speeds up switching between SubTools." I also concur that NVME is the best storage for Zbrush due to having the fastest read and write with large files, which 3D files typically are.