When are lock free data structures less performant than mutual exclusion (mutexes)?

Question

I read somewhere (can\'t find the page anymore) that lock free data structures are more efficient \"for certain workloads\" which seems to imply that sometimes they\'re actually

Answer 1

I don't know about making it slower, but it certainly makes it harder to get right. In the many cases where the two approaches are virtually identical in performance (or when it simply doesn't matter if it takes 500 pico-seconds rather than 100 pico-seconds), then pick the simplest approach - generally lock.

There are very few cases when that extra bit of performance is key; and if it is, I suspect you'd do well to use the pre-rolled pattern implementations from established libraries. Getting lock-free code working properly (and proving that it works properly in all conditions) is often very hard.

Note also that some environments offer a level of locking above the OS-provided mutex; mutex behaviour, but without some of the overheads (for example, Monitor in .NET).