Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!hplabs!hpda!hpcuhc!hpsemc!gph
From: gph@hpsemc.HP.COM (Paul Houtz)
Newsgroups: comp.sys.ibm.pc
Subject: Re: Optimizers and RISC (Re: Why unix doesn't catch on)
Message-ID: <8090020@hpsemc.HP.COM>
Date: 3 May 89 17:34:27 GMT
References: <3181@looking.UUCP>
Organization: HP Technology Access Center, Cupertino, CA
Lines: 26

brad@looking.UUCP (Brad Templeton) writes:

>Did RISC machines cause a push in optimizing compilers?  I always thought
>the big advantage of the RISC was that it made the low level optimizations
>so much simpler -- there's only one way to code the thing, so no need to

  Your question is a little ambiguous.   Are you asking if RISC machines
  caused a change in  the INDUSTRY, or do they give impetus for a single
  vendor to start making optimizers?

  HP has RISC machines, and yes, all of a sudden we have optimizing
  compilers for the first time.

  You see, Optimizers become a big boon with RISC because instead of having
  complex instructions (which cannot be optimized at compile time), you 
  have little programs that do what the CISC instructions used to do.  
  Those little programs are software, and once they are dragged into 
  the code, they can be optimized along with everything else in the 
  program.

  Also, when you start talking about pipelining, delayed branch 
  slots, and single cycle instructions, it actually becomes more reasonable
  to allow the compiler to generate whatever construct is efficient for 
  a given function.   Then it is very simple for a separate pass to do 
  things like fill in branch delay slots, move non-dependent structures
  out of loops, etc.