Re: re: Folk Theorem: Assemblers are superior to Compilers

jm@tao.univ-paris8.fr (Jean Mehat) writes:
|> I think these folk theorems may as well be true. I thought at first that
|> it may be false when the processor is designed for the compiler (like the
|> RISC delayed branch), but I think it's just that no one, except a
|> compiler's writer, cares to use delayed branch ; with reasonable training,
|> anyone should be able to use it like a compiler.


This suggests that the folk theorems may be stated incompletely: You must
ask not only for an `actual real-world application' but also for an
`actual real-world programmer'. This makes proving or disproving the
theorems quite difficult.


I know of colleagues of mine here in Karlsruhe, who significantly
outperformed the Sun-OS cc as well as the gcc on a SPARC with their
assembler routines. But these were only the very small kernels of their
cryptography application, which are only a very small fraction of the code
yet consume a very large amount of overall run time. It took them weeks
to tune this assembler code. I believe a similar thing is absolutely
impractical for complete applications.


As you see, I neither give quantitative figures of the effects nor do I
have any citations to present. Maybe they have published something about
their efforts, I don't know. Perhaps you want to contact Steffen Stempel
(stempel@ira.uka.de) to ask.


    Lutz


Lutz Prechelt (email: prechelt@ira.uka.de)
Institut fuer Programmstrukturen und Datenorganisation
Universitaet Karlsruhe; 76128 Karlsruhe; Germany
(Voice: ++49/721/608-4068, FAX: ++49/721/694092)
--

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