[Haskell-cafe] Compiled program running (extremely) slower than interpreted equivalent (no results produced)

[CASANOVA Juan]

Well, thanks for the tips.

Just for the sake of acknowledgment... it turns out it was something even more stupid. The program I was running was not the one I had just compiled, but rather an old version. The way I was compiling it it was producing no executable, only the .hi and .o files, so the executable was an old version from a previous compile, that did not produce any results.

Once I re-compiled it properly, it worked fine, and is slightly faster than the interpreted version.

Your comments were still useful and will be taken into account in the following days, though. So thank you again.

Juan.
________________________________
From: 宮里洸司 <viercc at gmail.com>
Sent: 16 July 2021 11:30
To: Haskell Cafe <haskell-cafe at haskell.org>; CASANOVA Juan <Juan.Casanova at ed.ac.uk>
Subject: Re: [Haskell-cafe] Compiled program running (extremely) slower than interpreted equivalent (no results produced)

This email was sent to you by someone outside the University.
You should only click on links or attachments if you are certain that the email is genuine and the content is safe.

I can't be sure without looking at your program directly, but
deleting all build artifacts (*.hi, *.o, the executable file, and
*.hi-boot files if there's) before compiling might resolve the issue.

What "stack ghc" does is starting GHC with appropriate configuration
(mostly which packages are to be used.) It does not recompile modules
when compilation flags were changed, just only when the source file it
depends on is changed after last compilation.
This means GHC might have compiled module A with -O0, B with -O2,
C with profiling on, etc. This mix is known to make optimizations fail
often.

2021年7月16日(金) 18:06 Tom Ellis <tom-lists-haskell-cafe-2017 at jaguarpaw.co.uk>:
>
> On Fri, Jul 16, 2021 at 01:10:36AM +0000, CASANOVA Juan wrote:
> > Here is my typical way to run it interpreted:
> >
> >   *   stack ghci +RTS -M500m -RTS
> >   *   :load CESQResolverEval.hs
> >   *   main
> >
> > This produces 4-5 outputs within a few seconds.
> >
> > When I try to compile it:
> >
> >   *   stack ghc --rts-options "-M500m" CESQResolverEval.hs
> >   *   stack exec ./CESQResolverEval
> >
> > This dies. Moreover, if I run without the RTS options and/or let it
> > run for a while, it completely kills my computer and I have to
> > restart it.
>
> Sounds like you have a space leak.  I don't know why the space leak
> would be exhibited in 'stack ghci' but not 'stack ghc'.  The first
> thing I would try would be to compile with '-O0', '-O1' and '-O2' and
> see if any of those make a difference (I don't know what 'stack ghc'
> uses by default.).
>
> Tom
> _______________________________________________
> Haskell-Cafe mailing list
> To (un)subscribe, modify options or view archives go to:
> http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe
> Only members subscribed via the mailman list are allowed to post.



--
/* Koji Miyazato <viercc at gmail.com> */
The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336. Is e buidheann carthannais a th’ ann an Oilthigh Dhun Eideann, claraichte an Alba, aireamh claraidh SC005336.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/haskell-cafe/attachments/20210716/55f513d9/attachment.html>