CommonMessaging - Improved inlining and branch predicition
All threads resolved!
All threads resolved!
Compare changes
This MR includes a minor change to CommonMessaging.h
designed to improve the runtime performance of CommonMessaging::setUpMessaging()
, which is quite heavily called, by making a few trivial changes.
CommonMessaging::setUpMessaging()
is split in two, moving the part that is infrequently called to a new private method. This makes the footprint of CommonMessaging::setUpMessaging()
much smaller and consequently improves the likelihood of it being inlined.Together these MRs address a small performance regression I recently noticed, most likely due to !566 (merged). See the callgrind heat maps below for the cost of a given algorithm call, before and after the change.