kirchner@informatik.uni-kl.de (Reinhard Kirchner) (10/12/90)
From article <11791@pucc.Princeton.EDU>, by EGNILGES@pucc.Princeton.EDU (Ed Nilges): > > On page 27 of Andrew S. Tanenbaum's book STRUCTURED COMPUTER ORGANIZA- > TION (Prentice-Hall 1976), there is a list of computers that have > been sold commercially and their word size. All are even numbers > save for one. This is the "Electrologica X8", with "27 bits per cell." > I have never heard anything else about this machine, which sounds > like a vacuum cleaner. When I started studiing at Karlsruhe University in Winter 1969 their 'Mainframe' was such a X8. I still have some paper about it, but not at hand, so I will brouse through my memory: It had 64k words of 27 bits, addresslength was really 15 bits, so we had low core and high core, programms run in one, don't remember how and why. It had a stack, and some registers, an integer akku and a FP akku. One interesting thing: FP-numbers where stored denormalized, so integers where converted by just moving to the FP akku and vice versa. There was a sophisticated inderect addressing scheme, so the lexical levels of ALGOL60 programs could be easily addressed. ALGOL was the only language we had ( exept assembler, which I tried one or two times, and still must have the listings -:) ) There was a purely batch operating system, as far as I know not the original one from Phillips, but may be developed in Karlsruhe univ. We had RJE via TTY and normel batch via operator. There where no permanent user files. I don't know anything about disk capacity or speed, may have been in the 100k - 300k ips. The RJE was handled by a PDP-8. They also had a very nice system implementation language called LOLA, where I made my first experiments with addresses etc. This for now, there must be still some system administrators at the computer center in Karlsruhe ( if not already retired ). R. Kirchner Kaiserslautern Univ. ( kirchner@uklirb.informatik.uni-kl.de )
dik@cwi.nl (Dik T. Winter) (10/13/90)
In article <6924@uklirb.informatik.uni-kl.de> kirchner@informatik.uni-kl.de (Reinhard Kirchner) writes: ... > One interesting thing: FP-numbers where stored denormalized, so integers > where converted by just moving to the FP akku and vice versa. This requires some clarification. FP numbers were stored in what was called 'standardized' format. This implied that the absolute value of the exponent was minimized (under the constraint of course that no precision is lost). I think the inventor was Grau, but I am not sure about the spelling of his name. The advantage is that integers are a priory in the correct format to be used as FP numbers (note: both integers and FP's were stored in one's complement). The reverse is not true. Yes, if the result of a FP operation was integral the result was stored in the correct integer format; but not if the result was not integral. (And I still omit some details.) -- dik t. winter, cwi, amsterdam, nederland dik@cwi.nl
ncjuul@diku.dk (Niels Christian Juul) (10/14/90)
For the benefit of history: This Electrologica X8 was the hardware underneath Dijkstra's classical operating system kernel: THE Multiprogramming System (THE=Technische Hogeschool Eindhoven) Described in Dijkstra's cacm article from 1968 where the concepts of a hierachical kernel design is presented, including semaphore with operations 'P' and 'V'. Niels Chr. -- ------------------------------------------------------------------------------- Niels Christian Juul Email: ncjuul@diku.dk DIKU (aka Dept.Comp.Sci. Univ.Copenhagen) Phone: +45 31 39 64 66 ext.405 Universitetsparken 1 -- DK 2100 Copenhagen Direct: +45 31 39 33 11 -- 405