🚀 go-pugleaf

mash@mips.COM (John Mashey) writes:
>The REAL question is: how much difference does it really make?
>Everybody please remember this was a "what-if" analysis, not an
>analysis that claimed that it really cost 1.6X in performance.
>At least several vendors who have looked at this and did care about
>COBOL didn't think that it was this much degradation to overall system
>performance, or at least, in designing their RISCs, included a few
>simple instructions to help the issue, and chose to stop there.

  I wonder if there is enough evidence about COBOL use to validate
a technique used both in elderly CISC machines and tiny, archaic ICL
micros: doing arithmetic on ascii strings directly.

  I once had to recode an ICL package to increase the maximum
length of string that could be added (:-(), and noted that the
inner loop amounted to a very small number of instructions.  This
raises the possibility that one could drag this old technique out
and apply it to modern machines, possibly with a tad of hardware
assist, much like the decimal normalize instructions discussed before.
  But the ICL did ascii (well, excess-3) arithmetic because it was
easy, not because it ran COBOL (it did not), and the CISC had both
conversion and ascii-arithmetic instructions...  Which means that
these techniques were not necessarily backed up by evidence.

  Who has evidence?  And how "easy" is, for example,
	word *add(word *,word*); /* A "word" is 2 ascii characters. */
on a given RISC machine?

--dave
-- 
David Collier-Brown,  | davecb@yunexus, ...!yunexus!davecb or
72 Abitibi Ave.,      | {toronto area...}lethe!dave 
Willowdale, Ontario,  | Joyce C-B:
CANADA. 416-223-8968  |    He's so smart he's dumb.