I’m on WM2.3 standard version, on my home computer which is an i5 2500K with 4 cores. I’m strongly considering upgrading to an i7 with 6 cores and hyper threading, and upgrade to WM Pro in order to utilize all the cores. I would also upgrade to 16GB RAM, leaving four RAM slots open to upgrade to 32GB in the long run.
Just wondering how much of an effect I can expect in terms of work with World Machine.
The first terrain I made took just under 9 hours to build in 8K on 2 cores, as the standard edition is locked to.
Just thinking if I could shave a few hours off of that, it would be worth it for me.
It’s highly dependent upon the world file you’re building, as each device has different efficiency in scaling. But in general, expect less than linear speedup but usually more than 50% per core – so going from 2 cores to 6 cores will probably not be 3x faster; but could be twice as fast. As a rough way to extrapolate, you could time a lower resolution build of your world file with 1 vs 2 threads enabled.
Many devices do in fact get nearly linear scaling; but as you increase the number of cores the amount of time spend on poorly scaling devices grows in proportion until performance plateaus.
I can confirm that, if you limit yourself to the devices that scale well / are heavily multithreaded, you actually get pretty close to linear scaling, i’ve been using WM with 32+ threads & scaling was near linear except on a few nodes (there are a few nodes that are single threaded too, so only 1 thread will be used unless other parts of the graph can run in parallel)