Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.forth > #42623

Re: Problem with GForth

Newsgroups comp.lang.forth
Date 2015-10-13 12:34 -0700
References (2 earlier) <b5e0c255-5ffc-416d-a269-d8edb4f9625b@googlegroups.com> <2015Oct8.094515@mips.complang.tuwien.ac.at> <083ceee9-89a3-4600-8169-f4828f442a8f@googlegroups.com> <87lhbc6hpu.fsf@jester.gateway.sonic.net> <2015Oct10.130320@mips.complang.tuwien.ac.at>
Message-ID <cb669f1f-e294-4e1b-a903-822bf100ead9@googlegroups.com> (permalink)
Subject Re: Problem with GForth
From lehs <skydda.barnen@gmail.com>

Show all headers | View raw

Den lördag 10 oktober 2015 kl. 13:26:16 UTC+2 skrev Anton Ertl:
> Paul Rubin <no.email@nospam.invalid> writes:
> >lehs <skydda.barnen@gmail.com> writes:
> >> https://github.com/Lehs/BIG-INTEGER-ANS-FORTH
> >> [ three 1022-bit operands ]    b**mod
> >> takes a half second in both SP-Forth and GForth64 at Win 10.
> >
> >Nice!  That's fast enough to be useful for some purposes.  I'd be
> >interested in seeing measurements with gforth-fast on various cpus, and
> >with other Forth implementations.
> 
> gforth-fast on a Core i7-4790K (running at 4.4GHz thanks to Turbo mode):
> real    0m0.117s
> user    0m0.112s
> sys     0m0.000s
> 
> vfxlin 4.71 on the same machine
> real    0m0.117s
> user    0m0.112s
> sys     0m0.000s
> 
> Strange.   Let's see what the performance counters say:
> 
> perf stat -e instructions:u -e cycles:u vfxlin-4.71 "include /nfs/nfstmp/anton/VfxLinEval-4.71/Lib/Ndp387.fth include bench.fs" >/dev/null
>         1041236122      instructions:u            #    2.08  insns per cycle        
>          499547195      cycles:u                 
> 
>        0.116635759 seconds time elapsed
> 
> perf stat -e instructions:u -e cycles:u /localtmp/anton/gforth/gforth-fast bench.fs >/dev/null
>         1595291884      instructions:u            #    3.14  insns per cycle        
>          507562915      cycles:u                 
> 
>        0.117390969 seconds time elapsed
> 
> While VFX executes fewer instructions, apparently the dependencies are
> such that this does not result in significantly better performance for
> this benchmark.
> 
> - anton
> -- 
> M. Anton Ertl  http://www.complang.tuwien.ac.at/anton/home.html
> comp.lang.forth FAQs: http://www.complang.tuwien.ac.at/forth/faq/toc.html
>      New standard: http://www.forth200x.org/forth200x.html
>    EuroForth 2015: http://www.rigwit.co.uk/EuroForth2015/

I suppose it was Linux?

OBS! There was a typo in the barrett reduction word 'barmod' that caused a lot of errors for small numbers, but more rare errors for sufficiently big numbers.

Back to comp.lang.forth | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-09-21 11:09 -0700
  Re: Problem with GForth Bernd Paysan <bernd.paysan@gmx.de> - 2015-09-21 20:14 +0200
    Re: Problem with GForth "Elizabeth D. Rather" <erather@forth.com> - 2015-09-21 12:36 -1000
      Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-09-21 16:20 -0700
        Re: Problem with GForth Bernd Paysan <bernd.paysan@gmx.de> - 2015-09-22 02:37 +0200
          Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-09-21 22:59 -0700
          Re: Problem with GForth "Elizabeth D. Rather" <erather@forth.com> - 2015-09-21 20:04 -1000
    Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-09-21 16:17 -0700
      Re: Problem with GForth Bernd Paysan <bernd.paysan@gmx.de> - 2015-09-22 02:35 +0200
    Re: Problem with GForth hughaguilar96@gmail.com - 2015-09-21 23:30 -0700
    Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-10-07 18:33 -0700
      Re: Problem with GForth Paul Rubin <no.email@nospam.invalid> - 2015-10-07 19:03 -0700
      Re: Problem with GForth anton@mips.complang.tuwien.ac.at (Anton Ertl) - 2015-10-08 07:45 +0000
        Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-10-08 05:06 -0700
          Re: Problem with GForth Howerd <howerdo@yahoo.co.uk> - 2015-10-08 14:24 -0700
            Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-10-08 22:49 -0700
              Re: Problem with GForth Paul Rubin <no.email@nospam.invalid> - 2015-10-08 23:11 -0700
          Re: Problem with GForth Coos Haak <chforth@hccnet.nl> - 2015-10-09 01:01 +0200
            Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-10-08 22:53 -0700
              Re: Problem with GForth Coos Haak <chforth@hccnet.nl> - 2015-10-09 10:04 +0200
                Re: Problem with GForth Paul Rubin <no.email@nospam.invalid> - 2015-10-09 01:23 -0700
                Re: Problem with GForth Coos Haak <chforth@hccnet.nl> - 2015-10-09 12:32 +0200
          Re: Problem with GForth "WJ" <w_a_x_man@yahoo.com> - 2015-10-08 23:34 +0000
            Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-10-08 22:55 -0700
          Re: Problem with GForth Paul Rubin <no.email@nospam.invalid> - 2015-10-08 16:52 -0700
            Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-10-08 23:03 -0700
            Re: Problem with GForth Coos Haak <chforth@hccnet.nl> - 2015-10-09 10:21 +0200
              Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-10-09 08:04 -0700
            Re: Problem with GForth anton@mips.complang.tuwien.ac.at (Anton Ertl) - 2015-10-10 11:03 +0000
              Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-10-13 12:34 -0700
            Re: Problem with GForth Marcos Cruz <use_the_contact_page@programandala.net> - 2015-10-13 18:31 +0000
              Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-10-13 12:40 -0700
                Re: Problem with GForth Marcos Cruz <use_the_contact_page@programandala.net> - 2015-10-14 20:20 +0000
          Re: Problem with GForth "WJ" <w_a_x_man@yahoo.com> - 2015-10-09 05:58 +0000
            Re: Problem with GForth lehs <skydda.barnen@gmail.com> - 2015-10-08 23:08 -0700
            Re: Problem with GForth mhx@iae.nl - 2015-10-09 00:30 -0700

csiph-web