Re: Performance issue with guile-2.2

[Stefan Israelsson Tampe]

Hi Thomas!

Have the lilipond team retried benchmarks with the newly released guile
2.2? One of the most resent improvements was fixing a gc issue that caused
performance issues with files of large scale. Could be, with some luck,
that that fix helped the lilypond benchmark that underperforms as well.

On Tue, Apr 11, 2017 at 4:22 PM, Thomas Morley <address@hidden>
wrote:

> 2017-04-11 12:28 GMT+02:00 Vijay Pratap Chaurasia <address@hidden>:
> > Hi,
> >      I have a tool which compares live data between two sources. I found
> > that the program runs more than 3 times slower than the same with
> > guile-2.0.11 . I have done profiling but there was nothing special to
> point
> > out. All most all calls were taking more than double the time compare to
> > guile-2.0.11. It is contrary to the claim of 30% performance boost with
> > guile-2.2 release. Can some one point the possible reason for slowness?
> >
> > I have created a simple test program which reports the diff of two alist.
> >
> > *time  /home/guile-2.2/bin/guile -s ./performanceTest.scm*
> > real    0m34.375s
> > user    0m37.616s
> > sys     0m0.361s
> >
> > *time /home/guile-2.0/bin/guile -s ./performanceTest.scm*
> > real    0m18.939s
> > user    0m18.829s
> > sys     0m0.104s
> >
>
> Hi vijay,
>
> I'm from LilyPond, where we have our own problems with every
> guile-2.x, we still use guile-1.8.
> So I'm very interested in performance problems.
>
> Though, I can't confirm your observations.
> Admittedly I used 2.0.14, with this I get:
>
> 2.2.0.17-685ca (built from recent master)
>
> real    0m31.642s
> user    0m33.272s
> sys     0m0.184s
>
> 2.0.14 (built from the released tarball)
>
> real    1m39.361s
> user    1m41.832s
> sys     0m0.224s
>
> Pity it doesn't work for guile-1.8...
>
>
> One other thing, please attach your files, or at least teach your
> e-mail-client not to insert *-signs at every line-break etc.
> Regardless whether there should be a line-break or not.
>
> Reformated versions are attached, if someone else want to check, too.
>
> Cheers,
>   Harm
>