http://dev.clojure.org/jira This file is an XML representation of an issue en-us 4.4 649 25-07-2011 http://dev.clojure.org/jira/browse/CLJS-326 ClojureScript <p>As discussed with mmarczyk in irc, the attached patch implements Clojure's hash-set function and modifies the compiler to create sets using a new fromArray static method. To test these, I created three new benchmarks. Here's before and after benchmark results:</p> <p>$ time ./script/benchmark | grep '#{'<br/> [], #{}, 100000 runs, 77 msecs<br/> [], #{1 2 3}, 100000 runs, 582 msecs<br/> [coll #{1 2 3}], (conj coll 4), 100000 runs, 104 msecs<br/> [], #{}, 100000 runs, 740 msecs<br/> [], #{1 2 3}, 100000 runs, 1809 msecs<br/> [coll #{1 2 3}], (conj coll 4), 100000 runs, 221 msecs<br/> [], #{}, 100000 runs, 281 msecs<br/> [], #{1 2 3}, 100000 runs, 1310 msecs<br/> [coll #{1 2 3}], (conj coll 4), 100000 runs, 288 msecs<br/> ./script/benchmark 109.41s user 3.82s system 127% cpu 1:28.96 total</p> <p>$ time ./script/benchmark | grep '#{'<br/> [], #{}, 100000 runs, 1 msecs<br/> [], #{1 2 3}, 100000 runs, 333 msecs<br/> [coll #{1 2 3}], (conj coll 4), 100000 runs, 100 msecs<br/> [], #{}, 100000 runs, 1 msecs<br/> [], #{1 2 3}, 100000 runs, 1670 msecs<br/> [coll #{1 2 3}], (conj coll 4), 100000 runs, 325 msecs<br/> [], #{}, 100000 runs, 1 msecs<br/> [], #{1 2 3}, 100000 runs, 689 msecs<br/> [coll #{1 2 3}], (conj coll 4), 100000 runs, 321 msecs<br/> ./script/benchmark 103.94s user 4.89s system 108% cpu 1:39.88 total<br/> grep '#{' 0.00s user 0.00s system 0% cpu 1:39.85 total</p> CLJS-326

hash-set and PersistentHashSet/fromArray == faster set construction

Enhancement Minor Resolved Completed Unassigned Brandon Bloom patch patch, Sun, 24 Jun 2012 20:01:16 -0500 Mon, 25 Jun 2012 06:33:47 -0500 Mon, 25 Jun 2012 06:33:47 -0500 0 0 <p>Slightly better patch with the EMPTY case</p> <p>Getting a weird error when I try to run the tests. Also can we squash the commits? Thanks!</p> <p>Squashed patch.</p> <p>Squashed and rebased on latest master. Is the rule no-multi-commit patches? Or simply just not for such small patches? In this case, I made and tested each each change in sequence. I also wanted the benchmarks to have a different sha1, so that the perf improvement would show up in the graphs. I'll try to provide squashed patches in the future.</p> <p>What weird error are you getting? Works for me :-P</p> <p>Updating 2nd benchmark in description to include the EMPTY optimization</p> <p>fixed, <a href="http://github.com/clojure/clojurescript/commit/c60df78e2fd318109a9338a87277c11565b506ae">http://github.com/clojure/clojurescript/commit/c60df78e2fd318109a9338a87277c11565b506ae</a></p> Global Rank Patch Code and Test