Where should the type information be 120
Where should the type information be 121
Stephen, There WAS some (relatively minor) interplay, as the stack architecture determined which instructions were likely to occur one after the other, which in turn affected how the systems...
If you mean that you must consider every aspect of the Burroughs large scale system architecture together and not try to analyze them separately, then I disagree. For example, one could imagine a computer that used the Burroughs memory addressing scheme, but not its stack oriented instruction set. Or one that had those two, but added a "privledged" end mode for some of the code in order to eliminate the requirement of giving special privledges to any program that generates code (to insure that it doesn't crash the system).
Sort of. The hardware still has to access the array descriptor in memory prior to computing the real memory address. Then there is a second memory access to get the actual data. These cannot be overlapped as you need the output from the first to begin teh second. The result is an extra memory access (albeit not programmer visible, but still taking real time) for the array access. This mattered less (in terms of performance) when the CPU was relativly slower compared to the memory (i.e. when said systems were designed), than it does now. In terms of array subscript calculations, today a multiply is faster than a memory access. But you do lose the protection aspects. The trick would be to design a mechanism that gained the protection without the performance cost.
-- - Stephen Fuld e-mail address disguised to prevent spam