On Mon 23 Feb 2004 15:47, Tim Bunce <[EMAIL PROTECTED]> wrote:
> On Mon, Feb 23, 2004 at 10:48:45AM +0100, H.Merijn Brand wrote:
> > On Sun 22 Feb 2004 20:29, Tim Bunce <[EMAIL PROTECTED]> wrote:
> > > A second release candidate is available for testing at:
> > > 
> > >   http://homepage.eircom.net/~timbunce/DBI-1.41-rc2-20040222.tar.gz
> > > 
> > > =head1 CHANGES in DBI 1.41 (svn rev 130),    22nd February 2004
> > 
> > Blead fails (5.9.1)
> 
> Thanks.
> 
> >  1136 NullP sth/sec on this perl 5.008    PA-RISC2.0 (10000 in 8.8 cpu+sys secs)
> >  1131 NullP sth/sec on this perl 5.008003 PA-RISC2.0 (10000 in 8.8 cpu+sys secs)
> >   989 NullP sth/sec on this perl 5.008003 PA-RISC2.0-LP64 (10000 in 10.1 cpu+sys 
> > secs)
> > 10869 NullP sth/sec on this perl 5.006001 PA-RISC2.0 (10000 in 0.9 cpu+sys secs)
> >  7633 NullP sth/sec on this perl 5.008    PA-RISC2.0 (10000 in 1.3 cpu+sys secs)
> >  7936 NullP sth/sec on this perl 5.008003 PA-RISC2.0 (10000 in 1.3 cpu+sys secs)
> >  7874 NullP sth/sec on this perl 5.009    PA-RISC2.0 (10000 in 1.3 cpu+sys secs)
> >  6944 NullP sth/sec on this perl 5.008003 PA-RISC2.0 (10000 in 1.4 cpu+sys secs)
> >  6993 NullP sth/sec on this perl 5.008003 PA-RISC2.0-LP64 (10000 in 1.4 cpu+sys 
> > secs)
> >  2304 NullP sth/sec on this perl 5.006001 PA-RISC2.0 (10000 in 4.3 cpu+sys secs)
> >  2136 NullP sth/sec on this perl 5.008    PA-RISC2.0 (10000 in 4.7 cpu+sys secs)
> >  2304 NullP sth/sec on this perl 5.009    PA-RISC2.0 (10000 in 4.3 cpu+sys secs)
> >  2105 NullP sth/sec on this perl 5.008003 PA-RISC2.0 (10000 in 4.7 cpu+sys secs)
> >  2197 NullP sth/sec on this perl 5.006001 aix (10000 in 4.5 cpu+sys secs)
> >  2288 NullP sth/sec on this perl 5.008    aix (10000 in 4.4 cpu+sys secs)
> >  2369 NullP sth/sec on this perl 5.008003 aix (10000 in 4.2 cpu+sys secs)
> > 17857 NullP sth/sec on this perl 5.008001 i586-linux-thread-multi (10000 in 0.6 
> > cpu+sys secs)
> > 24390 NullP sth/sec on this perl 5.008003 i686-linux-64int (10000 in 0.4 cpu+sys 
> > secs)
> >  5282 NullP sth/sec on this perl 5.008002 cygwin-thread-multi-64int (10000 in 1.9 
> > cpu+sys secs)
> 
> That's a big range of speeds! Are all the big differences explainable?
> (different hardware, compiler options etc)

All compiler options are comparable. Hardware is not :)

HP-UX 11.11/64 U      9000/861/D370/32    PA8000/160    pa-2.0   256 Mb
HP-UX 11.00/64 U      9000/800/A500-7X/64 PA8700/750(2) pa-2.0  2048 Mb
HP-UX 10.20/32 U      9000/800/D390       PA8000/240    pa-2.0   512 Mb

AIX   4.3.3.0/ML11/32 IBM,7043-150        PowerPC_604e/32        640 Mb

Linux 2.4.21-144      HP nx7000           i686/M 1400MHz         756 Mb

CYGWIN_NT-5.0 1.5.7(0.109/3/2)/2004-01-30
                      HP Vectra XU        i686/4 1.70GHz         511 Mb

As you can see, 11.11 has one 160 MHz CPU and only 256 Mb phys mem. 11.00 has
two 750 MHz CPU's of a much more recent architecture and 2 Gb phys mem. AIX
has never been fast. You conclude that 11.00 is probably about 10 times faster
than 11.11 in this configuration? You're most probably right :)

FWIW this 11.11 machine is ONLY smoking perl, and has no other function at
the moment.

SuSE clearly shows that threaded is slower than non-threaded

I still hope that my cygwin machine will crash into an unresorable state,
so I can install Linux. I'm not doing this now, because all still works, and
there's enough real work around.

> > ==============================================================================
> > - HP-UX 11.00             perl-5.9.1               HP C-ANSI-C
> > 
> > a5:/pro/3gl/CPAN/DBI-1.41 113 > /pro/bin/perl5.9.1 -Iblib/lib -Iblib/arch 
> > t/41prof_dump.t
> 
> > not ok 6
> > # Test 6 got: '1.0.0' (t/41prof_dump.t at line 55)
> > #   Expected: '1.0'
> > #  t/41prof_dump.t line 55 is: ok($1, $DBI::ProfileDumper::VERSION);
> 
> Groan. 5.9.1 is being, er, clever with version numbers.
> I'll grab bleadperl take a look. Meanwhile, this'll probably work:
> 
> --- t/41prof_dump.t     (revision 56)
> +++ t/41prof_dump.t     (working copy)
> @@ -51,8 +51,9 @@
>  # has a header?
>  ok($prof =~ /^DBI::ProfileDumper\s+([\d.]+)/);
> 
> -# version matches VERSION?
> -ok($1, $DBI::ProfileDumper::VERSION);
> +# version matches VERSION? (DBI::ProfileDumper uses $self->VERSION so
> +# it's a stringified version object that looks like N.N.N)
> +ok($1, DBI::ProfileDumper->VERSION);
> 
>  # check that expected key is there
>  ok($prof =~ /\+\s+1\s+\Q$sql\E/m);

All tests successful, 9 tests skipped.
Files=34, Tests=1429, 14 wallclock secs (11.45 cusr +  0.98 csys = 12.43 CPU)

Made 100 connections in  0 wallclock secs ( 0.02 usr +  0.00 sys =  0.02 CPU)

Testing handle creation speed...
7462 NullP sth/sec on this perl 5.009001 PA-RISC2.0 (10000 in 1.3 cpu+sys secs)

> Tim.

-- 
H.Merijn Brand        Amsterdam Perl Mongers (http://amsterdam.pm.org/)
using perl-5.6.1, 5.8.0 & 633 on HP-UX 10.20 & 11.00, AIX 4.2, AIX 4.3,
     WinNT 4, Win2K pro & WinCE 2.11 often with Tk800.024 &/| DBD-Unify
ftp://ftp.funet.fi/pub/languages/perl/CPAN/authors/id/H/HM/HMBRAND/

Reply via email to