Hi, On 2024-02-07 19:52:11 -0600, Nathan Bossart wrote: > On Wed, Feb 07, 2024 at 04:42:07PM -0800, Andres Freund wrote: > > On 2024-02-07 16:21:24 -0600, Nathan Bossart wrote: > >> The assembly for that looks encouraging, but I still need to actually test > >> it... > > > > Possible. For 16bit upcasting to 32bit is clearly the best way. For 32 bit > > that doesn't work, given the 32bit return, so we need something more. > > For the same compASC() test, I see an ~8.4% improvement with your int64 > code
Just to be clear, that code unfortuntely isn't correct, the return value is a 32 bit integer, so the 64bit difference doesn't help. In contrast to the 16bit case. > and a ~3.4% improvement with this: I guess that's still something. Another branchless variant is (a > b) - (a < b). It seems to get a similar improvement as the overflow-checking version. Greetings, Andres Freund
