Comments re Convex Performance

[P Tyers]

Any comments regarding the performance of Convex Vector Processors
when used in a 64+ user environment would be appreciated.

The Telecom Australia Research Laboratories is considering upgrading its'
computer facilities. One option that is open to us is to purchase a single
Convex upon which both number crunching and general purpose program
development (of number crunchy type programs) will take place.

I have a feeling that the two dont mix; the context switching involved
in a large number of users (at least 64 probably 96) carrying out
tasks for which the machine was not designed heuristically seems dicey.

Any comments will be appreciated. Reply by mail and I'll summarise
for the net.

-- 
------------------------------------------------------------------------------
P Tyers,	Telecom Australia Research Laboratories
ACSnet 	tyers at trlamct.oz	(Australian Computer Science Network)
CSnet	tyers at trlamct.oz	ARPAnet tyers%trlamct.oz at seismo.arpa
UUCP	{decvax,vax135,eagle,pesnta}!mulga!trlamct.oz!tyers
PaperMAIL:	P Tyers, Telecom Australia (Research), 
			 P.O Box 249, Clayton, VICTORIA 3168, AUSTRALIA

D
D