Front page | perl.loop |
Postings from April 2008
Hi!
The next release of AnyEvent contains support for a few more "backends",
notably POE, so AnyEvent is now by definition compatible to POE (before it
was only compatible when using an even loop used by POE, such as Event or
EV that could be shared).
To get a feeling about the overheads imposed by the extra layer of
AnyEvent and especially of the performance of the POE backend (which is
severely limited by a number of design problems in POE), I made some
benchmarks.
I started with benchmarking watcher creation/destruction time and callback
invocation time.
The results were mostly as expected, with EV leading and POE being
abysmal.
The surprising one was the pure perl implementation, which was quite on
par with C-based event loops such as Event or Glib. I did expect the pure
perl implementatioon to be at least a factor of three slower than Event or
Glib.
As the pure perl loop wasn't written with high performance in mind, this
prompted me to optimise it for some important cases (mostly to get rid of
the O(n²) degenerate cases and improving the select bitmask parsing for
the sparse case).
I then made a second benchmark, designed not to measure anyevent overhead,
but to measure real-world performance of a socket server.
The result is that the pure perl event loop used as fallback in AnyEvent
single-handedly beats Glib by a large margin, and even event by a factor
of two.
For small servers, the overhead introduced by running a lot of perl
opcodes per iteration dominates, however, reflected in the last benchmark.
However, the net result is that the pure perl event loop performs
better than almost all other event loops (EV being the only exception)
ins erious/medium-sized cases, while I originally expected it to fail
completely w.r.t. performance and being only usable as a workaround when
no "better" event module is installed.
All the benchmark data and explanations can be found here:
http://pod.tst.eu/http://cvs.schmorp.de/AnyEvent/lib/AnyEvent.pm#BENCHMARKS
The code is not yet released and likely still buggy (the question is
whether any bugs affect the benchmark results). It is only available via
CVS: http://software.schmorp.de/pkg/AnyEvent
Of course, I am very pleased by the performance of perl 5.10 in particular
and perl in general. This case also shows that one can beat C code in
many important cases by using better algorithms (i.e. by writing it the
"perlish" way).
It also means one can write medium-sized servers without relying on any
XS modules for performance, and that AnyEvent adds no serious overhead to
programs.
Feedback would be appreciated, otherwise: enjoy :)
--
The choice of a Deliantra, the free code+content MORPG
-----==- _GNU_ http://www.deliantra.net
----==-- _ generation
---==---(_)__ __ ____ __ Marc Lehmann
--==---/ / _ \/ // /\ \/ / pcg@goof.com
-=====/_/_//_/\_,_/ /_/\_\
Thread Next