justin on Wed, 9 Jul 2003 14:31:56 -0700


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Performance question


Hi, all,

There may be something obvious that I'm overlooking, but I am seeing a performance difference between two machines that seems counter-intuitive.

I have gp on two Mac OS X systems (10.2.6 and, um, another one). 10.2.6 is on a 450Mhz G4, while the other one is on a 300Mhz G3. When I ran 'make bench' on the G3, it seemed to do much better than on the G4.

G4:
Making bench in Odarwin-ppc
* Testing objets        for gp-sta..TIME=80     for gp-dyn..TIME=80
* Testing analyz        for gp-sta..TIME=370    for gp-dyn..TIME=350
* Testing number        for gp-sta..TIME=250    for gp-dyn..TIME=260
* Testing polyser       for gp-sta..TIME=130    for gp-dyn..TIME=140
* Testing linear        for gp-sta..TIME=170    for gp-dyn..TIME=170
* Testing elliptic      for gp-sta..TIME=220    for gp-dyn..TIME=250
* Testing sumiter       for gp-sta..TIME=420    for gp-dyn..TIME=460
* Testing graph         for gp-sta..TIME=160    for gp-dyn..TIME=180
* Testing program       for gp-sta..TIME=210    for gp-dyn..TIME=240
* Testing trans         for gp-sta..TIME=650    for gp-dyn..TIME=670
* Testing nfields       for gp-sta..TIME=1920   for gp-dyn..TIME=1920
+++ Total bench for gp-sta is 3044
+++ Total bench for gp-dyn is 3184
============================================================

G3:
Making bench in Odarwin-ppc
* Testing objets        for gp-sta..TIME=10
* Testing analyz        for gp-sta..TIME=270
* Testing number        for gp-sta..TIME=170
* Testing polyser       for gp-sta..TIME=70
* Testing linear        for gp-sta..TIME=80
* Testing elliptic      for gp-sta..TIME=180
* Testing sumiter       for gp-sta..TIME=340
* Testing graph         for gp-sta..TIME=100
* Testing program       for gp-sta..TIME=130
* Testing trans         for gp-sta..TIME=530
* Testing nfields       for gp-sta..TIME=1630
+++ Total bench for gp-sta is 2206

Any thoughts on the difference?  Here is the 'banner' for each:

G3:
============================================================
          GP/PARI CALCULATOR Version 2.2.6 (development CHANGES-1.741)
                  Power PC (PPC/GMP-4.1 kernel) 32-bit version
              compiled: Apr 19 2003, gcc-3.1 20020420 (prerelease)
                (readline v4.1 enabled, extended help available)

                       Copyright (C) 2003 The PARI Group

PARI/GP is free software, covered by the GNU General Public License, and
comes WITHOUT ANY WARRANTY WHATSOEVER.

Type ? for help, \q to quit.
Type ?12 for how to get moral (and possibly technical) support.

   realprecision = 28 significant digits
   seriesprecision = 16 significant terms
   format = g0.28

parisize = 4000000, primelimit = 500000
============================================================

G4:
============================================================
          GP/PARI CALCULATOR Version 2.2.7 (development CHANGES-1.786)
          Power PC running darwin (PPC/GMP-4.1 kernel) 32-bit version
compiled: Jul 9 2003, gcc-3.3 20030304 (Apple Computer, Inc. build 1435)
            (readline not compiled in, extended help not available)

                       Copyright (C) 2003 The PARI Group

PARI/GP is free software, covered by the GNU General Public License, and
comes WITHOUT ANY WARRANTY WHATSOEVER.

Type ? for help, \q to quit.
Type ?12 for how to get moral (and possibly technical) support.

   realprecision = 28 significant digits
   seriesprecision = 16 significant terms
   format = g0.28

parisize = 4000000, primelimit = 500000
============================================================

FWIW, gcc on the G4 is 3.1, and on the G3 is 3.3.

Regards,

Justin

--
Justin C. Walker, Curmudgeon-At-Large  *
Institute for General Semantics        |    Men are from Earth.
                                       |    Women are from Earth.
                                       |       Deal with it.
*--------------------------------------*-------------------------------*