[comp.arch] Code size vs. Data size

john@geac.UUCP (John Henshaw) (03/03/88)

In article <7520@apple.Apple.Com> bcase@apple.UUCP (Brian Case) writes:

>One thing is for sure though, code size is almost always much less important
>that data size.


I find this statement interesting.

The both code and data require memory accesses so that we can perform
useful work. It seems to me that we do many more accesses to fetch
instructions than we do to fetch data. This is of course, offset by the
fact that the code is held in very fast cache memory (if we do things
right :-)).

I would appreciate it if Brian, (or anyone else of course) would care
to expand on the above.

-john-
-- 
John Henshaw,			(mnetor, yunexus, utgpu !geac!john)
Geac Computers Ltd.		If we don't pay for education now, are we
Markham, Ontario, Canada, eh?	going to be able to pay for ignorance later?

bcase@Apple.COM (Brian Case) (03/05/88)

In article <2377@geac.UUCP> john@geac.UUCP (John Henshaw) writes:
>In article <7520@apple.Apple.Com> bcase@apple.UUCP (Brian Case) writes:
>>One thing is for sure though, code size is almost always much less important
>>that data size.
>
>The both code and data require memory accesses so that we can perform
>useful work. It seems to me that we do many more accesses to fetch
>instructions than we do to fetch data. This is of course, offset by the
>fact that the code is held in very fast cache memory (if we do things
>right :-)).
>
>I would appreciate it if Brian, (or anyone else of course) would care
>to expand on the above.

Er, I think I should have been more explicit.  I meant to say that "The
number of bytes of data is much greater than the number of bytes of code
in most applications."  There may be, understandably, some dissagreement
here.  I was not trying to make any points about caches or bus bandwidth
or whatever.  I was just saying that code can expand by a factor of two,
but the problem will still be one of data size.  I should say that I
have this opinion because I have written compilers, eg., that build very
big data structures to facilitate optimization.  Code size just didn't
matter.  I suspect spreadsheets and word processors and editors are
similar.

I would be happy to verbosely answer your question, but I don't understand
it.