If you are telling about 144: yes, original codes runs like: f[0] = f0; f[1] = f1; f[2] = f[2-2] = f[0] = f[0] f[3] = f[3-2] = f[1] = f[1] f[4] = f[4-2] = f[2] = f[0] f[5] = f[5-2] = f[3] = f[1]
the new codes runs like: f[0] = f0; f[1] = f1; f[2] = 2&1==0 ? f0 : f1 = f0 f[3] = 3&1==0 ? f0 : f1 = f1 f[4] = 4&1==0 ? f0 : f1 = f0 f[5] = 5&1==0 ? f0 : f1 = f1 thus it runs faster for around 7% faster on my performance tests. Gilles Sadowski <gillese...@gmail.com> 于2020年6月7日周日 上午5:53写道： > Hi. > > 2020-06-06 14:21 UTC+02:00, Xeno Amess <xenoam...@gmail.com>: > > Alright, done. > > This pr is now splitted into two prs. > > https://github.com/apache/commons-math/pull/144 > > https://github.com/apache/commons-math/pull/143 > > I think that there are issues with some of the changes (e.g. where > the previous code used an "index" variable but in the new version > there is a hard-coded number). > > Gilles > > > [...] > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > >
