Speeding up Unix

[reiher at ucla-cs.UUCP]

What kernel modifications do all you wizards think would produce the
greatest speedup in Unix (let's assume Sys V, to give a common base for
discussion)?  How difficult to you think they would be to make?  How much
can be done without introducing incompatibilites, at least at the source code
level?  Are there any potential speedups of such overriding benefit that
minor incompatibilities might be worth it?
-- 

        			Peter Reiher
        			reiher at ucla-cs.arpa
        			{...ihnp4,ucbvax,sdcrdcf}!ucla-cs!reiher