> Here is an opposite take on claiming fixes. In the mid-1970s, I > became involved in a system that had just come online with what was > already an antiquated architecture: CDC 6400 computers running SDC's > TDMS for the database, with dumb terminals and other peripherals wired > to "controllers", in use for mission-critical, real time data receipt, > processing, and human analysis and reporting. On several occasions, > when the system broke down for mysterious reasons, and while running > in degraded mode on the smaller of the two CPUs, the hardware people, > applications programmers, and systems programmers would all spend > hours finding problems and fixing them until the system returned to > normal operations. At the end, I could find no one to take credit for > detecting the root cause of the breakdown and fixing it.
The CDC 6600, and probably the CDC 6400 before that, had a fundamental flaw: the core memory was NOT protected by parity checking. Indeed, parity checking might not have been propagated throughout the various processor units. It was indeed possible, even very likely, that several "runs" of the same program and data could and would produce different results. I was, then, professionally involved in the development of a "finite element method" computer-aided engineering program system for the design of electric utility transmission towers, the large, lattice-type structures which, in my specific case, was intended to carry this Nation's largest municipal utility power over great distances, and at 500 kV ac and/or 1000 kV dc. It was insanity to use any computer system which could not be depended upon to produce the same results, given the same data, every time. EVERY TIME. So, it was my task to take a CDC 6600 program which could design these transmission towers and to convert it to IBM System/360. This conversion was completely successful and my program was certified by the structural engineering group of my employer. The System/360 had several things going for it, although its floating point was several binary digits less than CDC's: partity checking on each and every unit and bus, and on main storage. Also, a three binary digit system of "guard digits" to enhance (or maintain) floating-point resolution before the final, "normalized" result was written. -- You received this message because you are a member of G-Group, a group for those using G3, G4, and G5 desktop Macs - with a particular focus on Power Macs. The list FAQ is at http://lowendmac.com/lists/g-list.shtml and our netiquette guide is at http://www.lowendmac.com/lists/netiquette.shtml To post to this group, send email to g3-5-list@googlegroups.com For more options, visit this group at http://groups.google.com/group/g3-5-list
