Beefy Boxes and Bandwidth Generously Provided by pair Networks
more useful options
 
PerlMonks  

Re^9: Hockey Sticks

by BrowserUk (Patriarch)
on Jul 11, 2012 at 05:13 UTC ( [id://981005]=note: print w/replies, xml ) Need Help??


in reply to Re^8: Hockey Sticks
in thread Hockey Sticks

routinely attracts trolls

I won't ask you to define troll in this instance, but I do have a question.

Has anyone performed even rudimentary benchmarking of the various implementations of P6?

I'm still attracted to (what I remember of) the P6 syntax, but for the type of applications I mostly write -- generally cpu-bound -- cool syntax isn't sufficient justification for taking a 10x or even 5x performance hit over P5.

A simple benchmark -- as a minimalistic start I'd suggest: a) a non-trick, non-caching implementation of Ackermann; b) Stirling's series; c) the string handling benchmark from here; -- would give a somewhat real-world measurement of how the implementations are progressing from that point of view against P5, each other, and against themselves over time.


With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
"Science is about questioning the status quo. Questioning authority".
In the absence of evidence, opinion is indistinguishable from prejudice.

The start of some sanity?

Replies are listed 'Best First'.
P6 benchmarks (was Re^10: Hockey Sticks)
by raiph (Deacon) on Jul 12, 2012 at 21:04 UTC

    > Has anyone performed even rudimentary benchmarking of the various implementations of P6?

    To give a sense of #perl6 awareness and use of benchmarking, you could browse this search for "benchmark" mentions on #perl6. Nearly 300 days in the last 7 years include at least a mention. From the most recent page of results, 7 of the 13 matches involving 8 folk relate to benchmarking improvements or benchmark results rather than mere discussion. Imo the mandelbrot script is particularly interesting.

    From another angle, here's a slideshow about the performance of Rakudo (a P6 compiler).

    > I'm still attracted to (what I remember of) the P6 syntax, but for the type of applications I mostly write -- generally cpu-bound -- cool syntax isn't sufficient justification for taking a 10x or even 5x performance hit over P5.

    Right. Speedwise, in some ways current P6 implementations are in a position analogous to Python or Ruby, especially earlier versions. It's entirely possible that no current P6 implementation is yet in the same planetary system, let alone ballpark, for what it sounds like you need.

    That said, in theory, @Larry have tackled the issue of speed in P6, including at the cpu level, from the start. For example, in theory, P6 autothreading is supposed to automatically distribute a lot of computation over the multiple CPUs in multicore hardware. However, following good dev practice, a lot of P6 optimization has been left till "the last mile" -- ie hasn't happened yet.

    as a minimalistic start I'd suggest...

    Perl6 needs contributors much more than it needs ideas. That's why I posted this comment. It would be awesome if you visited #perl6 on freenode with an open mind and volunteered to try to set up one of the benchmarks you mention for one compiler and then post about your experience here. Maybe using japhb++'s perl6-bench project, a P6 workbench for doing comparative benchmarking?

      For example, in theory, P6 autothreading is supposed to automatically distribute a lot of computation over the multiple CPUs in multicore hardware. However, following good dev practice, a lot of P6 optimization has been left till "the last mile" -- which is where I think things are at now.

      Okay, thanks. The above, combined with slideshow, told me -- albeit rather verbosely in the latter case -- everythng I needed to know.


      With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
      Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
      "Science is about questioning the status quo. Questioning authority".
      In the absence of evidence, opinion is indistinguishable from prejudice.

      The start of some sanity?

        A follow up note to Patrick's Rakudo performance slideshow. The Rakudo team has only recently begun optimizing the compiler. So, although loops are something like 30x faster than they were 18 months ago, and Perl 6 is now considered fast enough that some folk have chosen to use it in commercial production settings, loops are still about 30x slower than Perl 5. Fortunately, it's been speeding up a lot for many months now and several big jumps are expected over the next year.

Log In?
Username:
Password:

What's my password?
Create A New User
Domain Nodelet?
Node Status?
node history
Node Type: note [id://981005]
help
Chatterbox?
and the web crawler heard nothing...

How do I use this?Last hourOther CB clients
Other Users?
Others sharing their wisdom with the Monastery: (6)
As of 2024-03-28 15:32 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    No recent polls found