Okay, after several hours of more testing & profiling it seems the culprit
is the implementation of (nth) - and IMHO this is quite a biggie:
(use 'macrochrono.core) ; only used for bench macro below
(def t [[0 0] [100 0] [100 100] [[0 0] 10000 100]])
(defn foo [[a b c [[dx dy] r x2]] [px py]]
(let [xx (- dx px) yy (- dy py)]
(< (+ (* xx xx) (* yy yy)) r)))
1.3.0
user=> (bench 100 (dotimes [i 100000] (foo t [i i])))
{:median 12.647, :max 31.255, :min 11.365, :avg 12.73133999999999}
1.4.0:
(bench 100 (dotimes [i 100000] (foo t [i i])))
{:median 21.574, :max 32.243, :min 18.031, :avg 22.32677999999999}
1.5.0-master-SNAPSHOT:
(bench 100 (dotimes [i 100000] (foo t [i i])))
{:median 21.429, :max 106.704, :min 17.776999999999997, :avg 22.94385}
VisualVM shows most of the time is spent in calls to nth, which makes sense
since the function makes heavy use of destructuring.
However, could someone please explain why its implementation is now almost
half the speed as compared to 1.3.0? nth is such a fundamental function
that it should not show such degradation in performance. I'm about to dig
through
the commit logs to find out more, but that's gonna take a while and I hope
someone from Clojure/Core can shed some light instead.
Proposals for workarounds are appreciated too (apart from avoiding
destructuring)... Moving & reducing destructuring in the inner (let)
improves it somewhat, but not much and makes the code less legible:
(defn foo2 [t p]
(let [[d r x2] (t 3)
xx (- (d 0) (p 0)) yy (- (d 0) (p 0))]
(< (+ (* xx xx) (* yy yy)) r)))
1.4.0
user=> (bench 100 (dotimes [i 100000] (foo2 t [i i])))
{:median 17.122, :max 108.78, :min 14.293, :avg 19.723170000000007}
1.5.0-master-SNAPSHOT
user=> (bench 100 (dotimes [i 100000] (foo2 t [i i])))
{:median 17.146, :max 93.47699999999999, :min 13.777, :avg
18.988480000000003}
Many thanks!
K.
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
