G92 on a 486, still more clarifications needed



 E. M. EVLETH
 Dynamique des Interactions Moleculaires
 Universite Pierre et Marie Curie
 4 Place Jussieu, Tour 22, Paris 75005
 33-1-44-27-42-08 (work), 33 = France; 1 = Paris
 33-1-45-48-67-20 (home)
 FAX 33-1-44-27-41-17 (lab);44-27-38-66(University)
 e-mail UDIM018 at FRORS31.BITNET
 Dfor avid Moses' confirmation of the $600 price a G92-windows
 486 version is welcome. BUT for USA based e-mailers and my own,
 additional clarification of the recommended configuration is still
 needed.  It is the $3000 figure which is questioned? I contacted a
 friend in the USA on this (who actually has an business in constucting
 PCsfrom components) and his comment was -----.
 "The  $3000 mentioned in theGaussian was not tout compris but their
 estimate (a bit optimistic I think) for what a mail order 486
 would cost that could run Gaussian. I doubt you could do anything
 serious on a $3000 machine.  They say you need 100 Mb free space + 60
 for the program. We can't do beans in 100 Mb of scratch. I would say
 you need at least a 1 Gb disk and 32 Mb of memory. If you can get
 that for $3000 (with the rest of the machine), (also a tape backup)
 you are getting an incredible deal. 32 Mb of good memory costs about
 $1200 wholesale and a 1 gb drive about 1500. I would say $5k is more
 reasonable for a properly configured machine"
 We run G92 on workstations in this configuration.  But a reasonable
 estimate of the cost of doing something reasonable is necessary for
 those who outside the "loop" (thank you George Bush).  Former eastern
 block, Russia, and many other countries are tied into PCs currently,
 visiting colleagues are use to PC operating systems and and have a lot
 PC software.  Those who pass by here use to using PCs complain about
 Unix environments and would prefer doing things on their PCs. Informing
 them on what more they really need to run G92, Games, Spartan, Hondo8
 and semiempirical (AMPAC 4, MOPAC) or whatever will be of service.
 We also still need the low end workstation timings on the
 examples Moses shows, the Cray times were interesting,
 a Cray engineer here in France showed G92 running about 40 times
 faster than our RS/6000 32OH but something more specific comparing
 a HP735, RS/6000 something would be useful too with regard to
 the 486 times. The timings allow outside the USA people to
 estimate the local cost competitiveness of remaining with PCs
 or getting workstation Unix operating system. One of the cost
 estimate features is the G92 on a 486 at $600 vs. $2000 on a RS/6000
 is a significant incitation to remain in the PC mode. That is
 the only thing which will be cheaper, the $5000 in the USA will
 be more outside the USA, though I don't know how much the Taiwanese
 machines undercut US suppled machines. I'll leave it to others
 in the real know to now supply the information.