From: |
peeter \\at// chem.ut.ee (Peeter Burk) |
Date: |
Fri, 25 Nov 1994 12:13:46 +0200 (EET) |
Subject: |
Gaussian benchmarks (summary) |
Hi!
Some time ago I asked on this forum about Gaussian benchmarks
for different computers. However, to my astonishment I got
quite a few replays. Still, if somebody have some data to,
I will be most thankfull. Anyways, here comes the original
inquiry and the answers I got.
______________________________________________________________________
From peeter Tue Nov 8 17:11:02 1994
Subject: Gaussian benchmark request
Dear all,
We have now lucky possibility to buy new workstation for our
group. We will use it primarily for running gaussian (and other
QM programs like this). Can somebody point me to the places in
the net, where one can obtain gaussian benchmark information for
various computers? When such information is not available, can you
please send to me the timings of some testjobs?
Please send your answers directly to me and I will later summarize
for net.
Best regards,
Peeter
----------------------------------------------------------------------
From hommes ( ( at ) ) ccc.uni-erlangen.de
Dr. N.J.R. van Eikema Hommes
Computational Chemistry Center
University of Erlangen
Naegelsbachstrasse 25
D-91052 Erlangen, F.R.G.
hommes "-at-" ccc.uni-erlangen.de
Ab Initio Benchmarks Gaussian 92, Cyclohexane (D3d)
Platform HF/6-31G* HF/6-31G* MP2(fu)/6-31G* MP4sdtq/6-31G*
Optimization Frequencies Optimization Single Point
CDC CD4680fs 4986 21172 41673 -
(0.52) (0.59) (0.38)
Convex C210 2570 12500 15900 33900
(1.00) (1.00) (1.00) (1.00)
SGI R4000 2445 12300 24300 92700
100 MHz (1.05) (1.02) (0.65) (0.37)
SGI R4400 1594 7528 15483 -
150 MHz (1.61) (1.66) (1.03)
HP 730 2377 11058 - -
(1.08) (1.13)
HP 735 1286 5872 12640 32034
99 MHz (2.0) (2.1) (1.26) (1.06)
IBM RS6000/580 976 4758 8891 -
(2.6) (2.6) (1.79)
DecAXP 3000/800 912 4296 6900 22582
(2.8) (2.9) (2.3) (1.50)
IBM RS6000/590 555 2337 4327 14110
(4.6) (5.3) (3.7) (2.4)
Cray YMP 392 1520 1832 7069
(6.6) (8.2) (8.7) (4.8)
Keywords :
HF Optimization
%mem=4000000
#P RHF/6-31G* OPT SCF=DIRECT TEST POP=NONE
HF Frequencies
%mem=4000000
#P RHF/6-31G* FREQ GUESS=READ GEOM=CHECK SCF=DIRECT POP=NONE
MP2 Optimization
%mem=8000000
#P RMP2=(FULL,DIRECT,MAXDISK=60000000)/6-31G* OPT=READFC SCF=DIRECT POP=NONE
GUESS=READ GEOM=CHECK TEST
MP4 Energy
%mem=8000000
#P RMP4SDTQ/6-31G* GUESS=READ GEOM=CHECK POP=NONE TEST
The full inputs are available upon request.
__________________________________________________________________________
From: jmuilu (+ at +) joyke7.joensuu.fi (Juha Muilu)
Here is short summary of g92 execution times (test job 178)
I have compiled from several sources..
Computer Proc MHz Nproc LP 178 G92 Version
Mflops min
--------------------------------------------------------------------
SGI Indigo/32Mb R4000 50/100 1 16 111 RevC.3
SGI Indigo/32Mb R4400 75/150 1 24 80 RevC.3
IBM RS6000 590/256MB Power2 66 1 132 20 RevE
IBM RS6000 590/256MB Power2 66 1 132 17 RevG3
IBM RS6000 380/32MB Power2 57 1 50 45 RevE
IBM RS6000 370 Power 62 1 26 57
IBM RS6000 550 Power 42 1 26 61
IBM RS6000 540 Power 30 1 19 83
IBM RS6000 32H/16MB Power 25 1 12 134 RevB
DEC 3000 AXP 800/Unix Alpha 200 1 44 RevE.2
DEC 3000 300 Alpha 150 1 144
DEC 3000 500X Alpha 200 1 59
DEC 3000 AXP 400/VMS Alpha 133 1 26 90 RevG.2
Cray X-MP EA/432 117 1 66 17
CRAY C90 8/256 ~250 1 4.25 RevC.3
CRAY C90 8/256(incore) ~250 1 1.5 RevC.3
CRAY YMP EL 34 1 39
Convex C3840 37 1 39
Convex C3840 37 2 39
Nproc) Number of processors used.
LP ) Linpack DP (100x100).
178) Test 178 from g92 test suite.
Elapsed wall clock time in minutes.
(Should be close to CPU+system time, when Nproc = 1)
Number of basis functions: 300
SCF iterations: 8
Input stream:
%mem=2000000
#P RHF/6-31G** scf=direct test pop=npa prop=fit
Gaussian 92 Test Job 178:
TATB rhf/6-31g**//hf/6-31g**
Energy should be -1006.2213391
....
c
Best Regards,
Juha Muilu
Department of Chemistry,University of Joensuu,P.O. Box 111
80101 Joensuu,FINLAND
Email:Juha.Muilu ^at^ joensuu.fi (NeXT mail ok)
Phone:358-73-1513312,Fax:358-73-1513390
------------------------------------------------------------------------
From Mckelvey ( ( at ) ) Kodak.COM
Several addenda for recent benchmarkings of Gaussian G3 and TEST0178
SGI/R8000 16 Min (By Roberto Gomperts at SGI)
IBM 590/pwr2 17 Min (Kodak)
IBM 390/pwr2 41 Min (Kodak; Same executable images as for 590/pwr2)
90 Mhz Pentium 600 Min (Kodak; Micronics mb with 512K cache and 32MB memory)
Regards
--
John M. McKelvey email: mckelvey;at;Kodak.COM
Computational Science Laboratory phone: (716) 477-3335
2nd Floor, Bldg 83, RL
Eastman Kodak Company
Rochester, NY 14650-2216
Thank you all who found the time to read or even reply.
Regards,
Peeter
--
Peeter Burk, Ph.D. Jakobi 2, EE2400 Tartu, Estonia
Institute of Chemical Physics Phone (372-7) 441-453
Tartu University Fax (372-7) 441-453
Estonia E-mail peeter (+ at +) chem.ut.ee
Similar Messages
11/15/1994: Re: CCL:Gaussian-92 system performance numbers?
06/09/1994: SUMMARY of the g92 test job nro 178.
06/09/1994: G92 test 178/ revised
11/10/1994: Gaussian benchmark request
Raw Message Text
|