Assembler to Ada Reverse Ratio o

Assembler to Ada Reverse Ratio o

[gamache]

I'm working another assignment, so naturally I have another question to which
I am again, shall we say, clueless. ;-)

Here goes:
If I have some existing legacy code in assembler, there seems to be some
intuitive relationship between that existing code and code that performs
the same function written in Ada.  This intuition is derived (at least
for me) from the expansion ratio between Ada and assembler.  (This ratio
is not only complier/target dependent; but also it is application dependent)
OK. Is there any emperical data on what this reverse relationship (from
legacy code to Ada) would be?  Before any "good design" folks even get
a chance - I am not considering simple recode of the existing code.  I'm
simply trying to scope the size of the Ada effort.  This effort will (with
almost certainty) create a design that is different from the assembler
one.  But at that time the ratio I seek will be easily calculatable.

adTHANKSvance

---------------------------------------------------------------------
Joe Gamache
Lockheed Sanders

Re: Assembler to Ada Reverse Ratio o

[Pug 156]

I can't say where he got it, but my manager uses a 6:1 ratio.  Of course,
that isn't for a RISC machine.  So, he'd say 4K SLOCs (source lines of
code) of Ada should make about 24K instructions.  I tend to get about
twice he's rate, but I'm a big fan of generics and nested function calls. 
He counts semicolons to generate a SLOC report.  Hope it helps.

Re: Assembler to Ada Reverse Ratio o

[Robert Dewar]

It is not the case with optimized code for typical RISC machines that the
number of instructions generated for typical Ada code should be noticeably
higher on a RISC machine than on a CISC machine. On the contrary the larger
number of registers, if well used, may easily DECREASE the total machine
instruction count.