Yes, context switching specifically. It's well known that if spawning thousands of child processes, or if walking a directory of 100k small files to get metadat

Yes, context switching specifically. It's well known that if spawning thousands of child processes, or if walking a directory of 100k small files to get metadata, linux can be much faster than windows. You might consider these just bad program design - on windows you'd strongly prefer to use many threads, and pack files of some kind (zip, cab, etc). There ends up being lots of different reasons why this kind of stuff can be rather slow on windows (but if you deal with fewer processes and fewer files you won't notice, it'll be fine). Some fascinating details: https://randomascii.wordpress.com/2019/04/21/on2-in-createprocess/ https://randomascii.wordpress.com/2019/04/21/on2-in-createpr... (and that blog has many similar fascinating posts)