FreeBSD replaces bubblesort with mergesort on SYSINTs, results in 100x speed

Re:FreeBSD replaces bubblesort with mergesort on SYSINTs, re... by masklinn on Monday August 21, 2023 @05:30AM (#37207161)

Here it took almost 30 years for it to stop being fine, which is a pretty good run.

And it was a blatant O(n) algorithm, easily swapped out, it’s not like they were calling sscanf in a loop and whoopsie turns out C strings are shit so now parsing a 4 characters decimal number is O(buflen^2).

[ Parent | on HN ]

Re:FreeBSD replaces bubblesort with mergesort on SYSINTs, re... by alkonaut on Monday August 21, 2023 @05:35AM (#37207200)

One should indeed be forgiven when getting 30 years out of a code base, any code base. It's hard to predict that you'd need to shave milliseconds of cloud/VM startups or from some embedded device, if you can't imagine Clouds, VMs or computers the size of watches.

[ Parent | on HN ]