Re: Folk Theorem: Assemblers are superior to Compilers

synaptx!daveg@uunet.UU.NET writes:
>Drawing again from the fertile ground of i860 programming :-), one of the
>keys to getting good performance on that chip is taking advantage of its
>huge memory bandwidth. It can load or store a quadword (four IEEE single
>floats) in a single clock cycle. BUT... it can only do that if the
>address is quadword aligned. How many languages have a declaration that
>tells the compiler that a given pointer, or even a given integer, is a
>multiple of 16?


When it comes to languages, there are standards and there are standards.


In the case of the C language, the PRIMARY standard the the ISO C
standard, and I must agree that this standard does not mandate alignment
of various data types (nor does it specify any sort of pragma by which
shuch alignment could be effected).


It appears however that perhaps Dave Gillespie may perhaps be unfamilar
with yet another kind of standard (as are a number of other people), i.e.
an "industry standard" (as differentiated from, say, and international
standard or a national standard).


In the case of the C language, we are (I think) fortunate to have certain
"industry standards", which, in many cases, go beyond the requirements
laid down by the international ISO C standard.


I am speaking of the System V Release 4 processor-specific ABI
supplements.


In the case of the i860 (in particular) the ps-ABI for this processor does
indeed require compilers to align all data objects (and members of struct
and union types) which have type `long double' to 16 bytes boundaries.


-- Ronald F. Guilmette, Sunnyvale, California -------------------------------
------ domain address: rfg@netcom.com ---------------------------------------
------ uucp address: ...!uunet!netcom.com!rfg -------------------------------
--

Return to the comp.compilers page.
Search the comp.compilers archives again.