Where should the type information be 116
YES, another GE-Honeybucket jock!
Where should the type information be 117
Nick Maclaren Some further comments, that may be out of date as google is not showing recent articles. The PL-I article makes it clear that the dope vector terminology was used in...
Don't forget the Repeat Double instruction that allowed you to construct your own instructions that ran at the same speed as hardwired instructions! There were two instruction end units, one for even locations and one for odd locations. The Repeat Double instruction had to be located on an odd location, and locked the next two instructions into their respective end units and executed them over and over until a particular stated condition occurred. These were used for most of the moves, searches, type conversions, etc. needed to support COBOL. With the addition of character-level indirect addressing, it was an INCREDIBLY powerful string processing machine, easily outrunning its compebreastion with similar clock speeds by between 2:1 and 3:1.
This same technology could probably be applied today to gain a LOT more speed for small loops, as it eliminates the loop overhead while still being able to perform each iteration differently, e.g. searching for some pattern, etc. This provides nearly array-processor speed but without the usual limitations of vectorizability.
These were originally engineered to replace 16 IBM-7094s used to perform range control that were on lease, which they did very well. They even included a Gray-To-Binary instruction to support their initial application.
Then came the 360s and the faster honeybuckets faded into history.
Steve Richfie1d ===============