Re: When Doing More Is Faster: A Curious Xojo Performance Test – Xojo Programming Blog
It would be neat if Xojo Inc could provide a sample project to replicate this.
One thing I’m noticing is the dramatic speedup of each run (for both Test1 and Test2):
Test1 takes 16348 on the first run but finishes iteration 10 at 5783 (over 3x speedup)
Test2 takes 9198 on the first run but finishes iteration 10 at 5488 (nearly 2x speedup)
The first run of a benchmark is often slower, since data hasn’t been cached yet, but it’s weird to see a consistent speedup with each run.
I feel like we may be seeing something else at play - perhaps the OS is noticing a high CPU process and giving it more CPU time?