io_rewiring branch needs testing
Patrick R. Michaud
pmichaud at pobox.com
Mon Jun 8 02:45:12 UTC 2009
On Sat, Jun 06, 2009 at 07:51:16PM -0700, Mark Glines wrote:
> Andrew Whitworth wrote:
> > Infinoid++ have been working on the io_rewiring branch and are at a
> > point where we would like to get some extra testing. Not only do we
> > need testing to prove that the branch isn't broken and stupid, but we
> > also want verification that the work in this branch does represent an
> > actual performance improvement. Infinoid has done one test that
> > suggests a 4x speedup for IO-intensive programs, which would be very
> > nice indeed.
>
> We're looking for better ways to test and benchmark it, and we'd also
> love some assurance that things work (or reports that they don't work)
> for lots of HLLs.
In early February I posted a message [1] to parrot-dev that
demonstrated how Parrot's (current) IO system is very slow.
In the new branch the test program I ran is indeed about 4 times
faster:
trunk : 48.394s
io_rewiring: 11.151s
So, good work thus far!
By way of comparison, the 0.8.0 version of Parrot (pre-I/O-refactor)
runs the benchmark in 10.437s, which means the branch is getting
close to what Parrot once had. It's entirely possible that the
difference here is due to startup or other non-I/O items, too.
Another comparison: Perl 5 is able to perform the equivalent
set of operations in 4.554s. It seems to me that for
straightforward I/O such as this Parrot really ought to be a
bit closer to the Perl 5 speed.
Pm
[1] http://lists.parrot.org/pipermail/parrot-dev/2009-February/001356.html
More information about the parrot-dev
mailing list