(Just a little testing of the exciting new math lib.) Is it currently normal that computing the mean of a big list is 10x slower compared to the (/ (apply + l) (length l)) method?
The math lib is very recent so I'm not entirely surprised, anyway. Most of the time seems to be due to GC. E.g. (on command line Racket) : #lang typed/racket (require math) (: n Integer) (define n 10000000) (: l (Listof Integer)) (define l (build-list n (λ(x)(random 100)))) (: my-mean ((Listof Integer) -> Number)) (define (my-mean l) (/ (apply + l) (length l))) (time (mean l)) (time (my-mean l)) --> cpu time: 3932 real time: 3931 gc time: 2848 495033609/10000000 cpu time: 408 real time: 406 gc time: 4 495033609/10000000 Cheers, Laurent
____________________ Racket Users list: http://lists.racket-lang.org/users
