@jeroenwarnaar

Hi Jeroen,

There are processes, or processtructures that benefit from multiprocessing.

As you probably know, to effectively use it and have performance increase depends on a bunch of parameters.

First you need to set a workbench up to use it, massive GroupBy use.

Secondly it is usually more effective with large datasets and sufficient relations between those.Fme must set up all its workers so for small datasets there is no gain (or virtually none).

Nut spatial operations, like areoverlyer with lots of overlapping objects, benefit a lot. I ave some running that perform 2x to 4x from tiling and parallel processing.

Or creating accurate hulls form large datasets, benefits a lot.

Standard single stream processes mostly do not benefit.

Also buying a decacore hyper treading processor wont give any benefit if the licence level is standard base edition. You need high licensing to use 32 threads.

According to this document 16 processes is max.

see https://docs.safe.com/fme/html/FME_Desktop_Documentation/FME_Transformers/parallel_processing/parallel_processing.htm

Parallel processing in FME desktop is finicky, and only available in limited circumstances. Fewer, faster cores are usually better than more, slower cores. As a general rule of thumb, there is not much benefit to having more cores than processes available to the license - which is 8 for most licenses.

FME only uses a single core per running workspace, unless you specifically use parallel processing in your workspace:

https://docs.safe.com/fme/html/FME_Desktop_Documentation/FME_Transformers/parallel_processing/parallel_processing.htm

thanks for the answers, maybe i should have mentioned in the question that i'm configuring a desktop for someone who works with the software in our company so i don't know much about how FME works specifically, i can work with these answers though so the question is answered for now :)