I've never seen GO before this thread, but after a glance at the GO web site, I suspect the difference may have more to do with concurrency than compiling. Perl is written in C and natively pretty good at file and string manipulation. My gut says a 2x win for this problem just by using GO compiled to a "static binary" is much too big a difference.
Is the code for the GO program being compared to the posted perl script also available? I'm curious, -Andrew > On Mar 14, 2014, at 12:42 PM, [email protected] wrote: > > From what I can see GO is purely compiled down to object code and linked into > static binaries. IMHO Perl, as an interpreted language, is doing *super* to > be *only* twice the runtime of Go! Maybe others have a better handle on > this. > >> On March 8, 2014 3:26:31 PM EST, Gyepi SAM <[email protected]> wrote: >>> On Sat, Mar 08, 2014 at 02:12:50PM -0500, David Larochelle wrote: >>>> On Sat, Mar 8, 2014 at 1:40 PM, Gyepi SAM <[email protected]> >>> wrote: >>> >>>> For fun, I wrote a version in Go and it's twice as fast as the perl >>>> version. I imagine a C version would be faster yet, but I get paid >> for that >>>> kind of fun. I'd be happy to send you the Go version if you're >> interested. >>> >>> >>> >>> I'm curious how you account for disk caches when your bench marking. >> I >>> fully expect that the CPU version of the code will be faster in GO >> than >>> Perl. But I wonder how much this matters if the file isn't already in >> the >>> disk cache. >>> >>> An interesting test would be to run the GO version first on a file >> that's >>> you can be sure is not in the disk cache, then to run the Perl >> version on >>> that file. I.e. let Perl benefit from the disk cache and see if GO is >> still >>> faster. >> >> I have an SSD drive on my linux laptop so disk caching plays less of a >> role in >> my tests. I clear the cache with: >> >> echo 3 | sudo tee /proc/sys/vm/drop_caches >> >> I repeated the tests after clearing the disk cache each time and also >> ran each >> test multiple times without clearing cache. The cached versions are a >> few >> seconds faster, but the overall results are the same; the Go version is >> about twice as fast. >> >> -Gyepi >> >> _______________________________________________ >> Boston-pm mailing list >> [email protected] >> http://mail.pm.org/mailman/listinfo/boston-pm > > _______________________________________________ > Boston-pm mailing list > [email protected] > http://mail.pm.org/mailman/listinfo/boston-pm _______________________________________________ Boston-pm mailing list [email protected] http://mail.pm.org/mailman/listinfo/boston-pm
