tom@stl.stc.co.uk (Tom Thomson) (06/28/90)
Well, we still make an enourmous profit out of selling mainframes. As to terminology, I think a mainframe is a system which can sustain high throughput while running a mixed-mode workload; most of the "supers" can usefully carry out only one type of work at a time (usually because their OS is simplified to eliminate overheads when running crunchinmg jobs). Other characteristics include high data throughput compared to compute speed, vast (disc, tape) amounts of file store (high file-store to main store size ratio), complex db/transaction monitor. Tom
eugene@wilbur.nas.nasa.gov (Eugene N. Miya) (06/28/90)
More: From: dwells@fits.CX.NRAO.EDU (Don Wells) Hugh wrote: >Mainframe. A long way to go before dead. Today, a mainframe is a >machine which >can do at least 100 MBytes/sec of I/O. All the mini/micro systems that >I know of >think 8-10 MB/sec is a big deal... >. The first step is memory bandwidth. Was >~1Gbyte/sec. Next time, 5-10 GBytes/sec. Still far from Y-MP 40 GBytes/sec. > >It is too bad that no one builds decent Unix mainframes, ... That last sentence ignores things like Amdahl's "UTS", which runs native on any 370 architecture, and IBM's "AIX" which at least runs under VM. And -- amazing! -- it ignores CRI's "Unicos" on CRI "mainframes". It also ignores the mini-supercomputers, which are effectively mainframes by your definition, and which mostly run Unix. Also, their kernels are customized and have sophisticated I/O systems which are, of course, analogous to the smart channel controllers of traditional mainframes. There is nothing about Unix per se that inhibits vendors from providing kernels with functionality and performance analogous to traditional kernels, and numerous vendors are systematically following the obvious strategy. Surely, Eugene, you are not ignorant of these things!?! [We have big Amdahls front-ending our mass-storage system --enm] Donald C. Wells, Associate Scientist | NSFnet: dwells@nrao.edu [192.33.115.2] National Radio Astronomy Observatory | SPAN: NRAO::DWELLS [6654::] Edgemont Road | BITnet: DWELLS@NRAO Charlottesville, VA 22903-2475 USA | UUCP: ...!uunet!nrao.edu!dwells Lat: 38:02.2N Long: 78:31.1W | Tel:+1-804-296-0277 Fax:+1-804-296-0278 ================== From: dik@cwi.nl In article <6965@amelia.nas.nasa.gov> you write: > The thing to clearly watch for the the appearance of a COBOL compiler > for the Cray. Then it's curtains. > But isn't that already the case? As far as I understand it, Crays are nowadays mainframes. Just as inaccessable for the average student as a 'mainframe' was 10-15 years ago (at least here, the time scale may be different in the US). -- dik t. winter, cwi, amsterdam, nederland ================== From: tony@comperex@comperex.cx.oz.au (Tony Williams) This is indeed an interesting subject. I previously worked for Control Data Australia on mainframes like CDC 6600, 7600, Cyber 205, Cyber 990, ETA10 and also Vax .... through the Third Party Maintenance area. Now I work for the distributor of MIPS systems. The compute power comparison is very interesting based on physical size, I/O bandwidth etc. Even better think about how you would classify a 386/486 with a couple of Transputer array boards plugged into it. --e. nobuo miya, NASA Ames Research Center, eugene@orville.nas.nasa.gov {uunet,mailrus,other gateways}!ames!eugene