This is just to show a few of the results we've been getting running CNS
(CCP4 examples to follow next week) jobs on different CPU's.
Its by no means an exhaustive benchmarking test.
We ran two CNS jobs from the nsf sad phasing tutorial, just to test our
own systems, but I thought you may be interested in the results.
Especially the on the AMD ATHLON CPU.
All compilation was carried out using the default optimization
parameters (ie. no extra
optimization tried).
SGI: compilers 7.3
PC/MAC fort77/f2c gave a faster executable compared which gcc/g77.
Commercial Portland group compilers gave a very fast executable for CNS.
The PC/MAC had 128MB RAM and were running under Linux Redhat 6.1 and
linuxPPC respectively.
Thanks to Ilya Goldberg and Piotr Sliz for help in compiling and
testing.
|
Computer
|
CNS job
|
Time
|
Total Time (mins) |
|
Alpha 21264 (CNS tutorial)
|
heavy.inp /sad_phase.inp
|
2940/720
|
61
|
|
SGI IndigoII R10000/175Mhz
|
"
|
5754/1243
|
117
|
|
SGI Octane R10000/175Mhz
|
"
|
5591/1224
|
114
|
|
SGI Origin 200 R10000/225Mhz
|
"
|
4350/960
|
88.5
|
|
PentiumII 400Hz
|
"
|
6949/2452
|
157
|
|
PentiumII 400Mhz (Portland group compilers)
|
"
|
5302/1654
|
116
|
|
Mac 400Mhz G4/linuxppc
|
"
|
5277/not run
|
--
|
|
Pentium III 500MHz (fort77/f2c))
|
"
|
5491/1955
|
124
|
|
AMD ATHLON 700Mhz PC (fort77/f2c)
|
"
|
3998/1165
|
86
|
|
AMD ATHLON 700MHz PC (Portland compilers)
|
|
2611/749
|
56
|
begin:vcard
n:Litster;Steve
tel;fax:617-355-3506
tel;work:617-355-6563
x-mozilla-html:FALSE
url:http://crystal.harvard.edu
version:2.1
email;internet:litster@crystal.harvard.edu
title:Computer Systems Manager
org;quoted-printable:Howard Hughes Medical Institute/Harvard University;Departments of Biological Chemistry=0D=0A and Molecular Pharmacology and Pediatrics (HMS)
adr;quoted-printable:;;Enders Research Building=0D=0A320, Longwood Avenue;Boston;MA;02115;USA
x-mozilla-cpt:;0
fn:Steve Litster
end:vcard