Performance advantages of having a CPU with more cores and threads while using Unity

Hello everyone!!

I´m in the middle of building a brand new PC and the time to choose the processor has come. I´m going with Ryzen because of its great cost to performance ratio and I´m having trobule deciding between Ryzen 5 3600 or Ryzen 7 3700X .

What I want to know is how much editor performance would I gain by going with the R7 3700X instead of the R5 3600. I´m one of the programmers in our project and I´m usually the one that makes the builds. Our game is 2D , hand drawn with lots of sprites , some hand-made shaders and we have tons of sprite atlases that need to constantly be repackaged.

Would build times , script compile times , sprite packing and asset import time decrease with a beefier CPU like the R7 3700X? Does having more cores and threads matter while working with Unity? If we make the jump to making a 3D game would my processor requirements change?

Hope you can help me.

 

Hey! We're currently in the same boat. I'm the main designer and am in Unity the most out of anyone on our team, building all the maps, events and sometimes making code edits. We work with large textures for our sprite atlases (8192x8192) and noticed that it can take ~50s for a single uncompressed asset atlas to import. Importing assets is easily our largest bottleneck and this infuriates our programmer, as he will often pull our graphics changes which need to be fully imported, locking him out from the project for many minutes at a time as the library builds. I started on a journey to figure out exactly what kind of hardware is most optimal for projects like ours: 2D, hand drawn with many sprites and large textures.

I'm also working with systems integrator Puget Systems to help determine what hardware would be best. They've identified Unity's general import processes to be highly single threaded with the current Unity import pipeline, but are interested in textures specifically. I've also reached out to Unity support to see if they have any additional info on the texture importing process and if they have any hardware recommendations to speed things up. I'll report back here if I hear anything

What I can say with certainty is that texture compression does seem to be highly parallelized, and might benefit from many cores. Importing an atlas at High Quality compression settings adds another 30s to the import process, pegging my CPU (6c/12t) at 100% during the additional time. Still, it seems working uncompressed for the majority of the project will be optimal, compressing your textures only for a release.

So, right now, it's looking like Intel CPUs might be the way to go, as they feature the highest single threaded performance. Specifically, the upcoming i9-10900K is looking to be my pick.

I believe Unity is currently in the process of introducing a new import pipeline - I am still investigating this though. Not sure if it benefits more from additional CPU cores vs frequency.

 

Hey there, sorry for not answer your question. Just to say that im looking the same answer. Currently I have FX8350 with only 8gb RAM on 990FXA-GD80. Im doing builds frecuently of my project and that cost to me 15 minutes each build. As far I can see the amount of RAM is my bottleneck. But I need to decide if I just buy more DD3 RAM or invest in a Ryzen 3600x or 3700x with 16GB (2x8) 3200mhz RAM.