[Date Prev][Date Next][Thread Prev][Thread Next]
[Date Index]
[Thread Index]
- Subject: Re: Lua (small optimizations)
- From: Andrew Gierth <andrew@...>
- Date: Fri, 11 Dec 2020 14:11:38 +0000
>>>>> "Ranier" == Ranier Vilela <ranier.vf@gmail.com> writes:
Ranier> Best Lua result (5.4.2 last git):
Ranier> Averange: 0.054667
Ranier> Three 0.051000
Ranier> Best Lua result (patched):
Ranier> Averange: 0.053500
Ranier> Five 0.051000
So here's a thing about performance differences: you can get speedups or
slowdowns of well over 10% simply by changing the size of a function
that's never called within your benchmark.
Running a benchmark dozens of times using the same binary to take an
average doesn't help. To determine whether some specific change actually
speeds things up, you need to test whether that speedup is itself
preserved if you make irrelevant changes.
--
Andrew.