Path: utzoo!utgpu!water!watmath!clyde!bellcore!faline!thumper!ulysses!andante!mit-eddie!bu-cs!purdue!decwrl!labrea!sri-unix!garth!smryan
From: smryan@garth.UUCP
Newsgroups: comp.arch
Subject: Re: Memory latency / cacheing / scientific programs
Message-ID: <843@garth.UUCP>
Date: 30 Jun 88 21:58:08 GMT
References: <243@granite.dec.com> <779@garth.UUCP> <2033@pt.cs.cmu.edu> <803@garth.UUCP> <11023@ames.arc.nasa.gov> <6070@aw.sei.cmu.edu>
Reply-To: smryan@garth.UUCP (Steven Ryan)
Organization: INTERGRAPH (APD) -- Palo Alto, CA
Lines: 11

>                                 then the codegenerator really must be
>prepared to perform global register allocation, and must also address
>the problem of register partitioning between user code and shared
>library code.  Anything less, I would regard as an indication that the
>implementor hadn't bothered to understand the machine.

We understood the machine, but the problem isn't so simple. Current languages,
C and Fortran in particular, emphasise separate compilations but do not
provide a standard library mechanism to permit good interprocedure
optimisation. While work is been doing to overcome this, it is in spite of
rather than because of most languages.