cameron <cdor...@gmail.com> writes:
> Interesting problem, the slowdown seems to being caused by the reverse
> call (actually the calls to conj with a list argument).
Excellent analysis, sir! I think this points things in the right
direction.
> fast-reverse : map-ms: 3.3, pmap-ms 0.7, speedup 4.97
> list-cons : map-ms: 4.0, pmap-ms 0.7, speedup 6.13
The difference between these two I believe is just jitter. Once `cons`
is called on either a list or a vector, the result is a
`clojure.lang.Cons` object.
> vec-conj : map-ms: 4.0, pmap-ms 1.3, speedup 3.10
For the sub-goal of optimizing `reverse`, I get better times even than
for the `cons`-based implementation by using transient vectors:
list-cons : map-ms: 4.0, pmap-ms 0.6, speedup 6.42
tvec-conj : map-ms: 0.9, pmap-ms 0.2, speedup 4.10
(defn tvec-conj [coll]
(persistent! (reduce conj! (transient []) coll)))
> list-conj : map-ms: 10.8, pmap-ms 21.2, speedup 0.51
> clojure-reverse : map-ms: 13.5, pmap-ms 26.8, speedup 0.50 (this is
> equivalent to the original code)
I add the following:
cons-conj : map-ms: 3.3, pmap-ms 16.8, speedup 0.19
(defn cons-conj [coll]
(reduce conj (clojure.lang.Cons. (first coll) nil) (rest coll)))
I think this is the key, but I don’t understand it.
The `cons` function just immediately creates a new `Cons` object. The
`conj` function calls the `.cons` method of the collection, and the
`.cons` implementation `Cons` inherits from `ASeq` just creates a new
`Cons` object!
It’s like there’s a lock of some sort sneaking in on the `conj` path.
Any thoughts on what that could be?
-Marshall
--
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
