Re: Code quality

polstra!jdp@uunet.UU.NET (John Polstra)
Tue, 12 Jan 1993 17:29:53 GMT

          From comp.compilers

Related articles
[5 earlier articles]
Re: Code quality prener@watson.ibm.com (1993-01-07)
Re: Code quality ssimmons@convex.com (1993-01-07)
Re: Code quality bill@amber.csd.harris.com (1993-01-07)
Re: Code quality tm@netcom.com (1993-01-07)
Re: Code quality grover@brahmand.Eng.Sun.COM (1993-01-07)
Re: Code quality drw@riesz.mit.edu (1993-01-08)
Re: Code quality polstra!jdp@uunet.UU.NET (1993-01-12)
Re: Code quality shebs@apple.com (1993-01-13)
Re: Code quality glew@pdx007.intel.com (1993-01-25)
Re: Code quality wjw@eb.ele.tue.nl (1993-02-01)
| List of all articles for this month |
Newsgroups: comp.compilers
From: polstra!jdp@uunet.UU.NET (John Polstra)
Organization: Polstra & Co., Inc., Seattle, WA
Date: Tue, 12 Jan 1993 17:29:53 GMT
References: 93-01-017
Keywords: optimize

drw@zermelo.mit.edu (Dale R. Worley) writes:
> Is there much of a market for another 10% in speed of generated code?


For a number of years I have been doing consulting work for the compiler
department of a large company that manufactures office-scale Unix systems
with bundled compilers. Run time performance of the compiled code is
*all* they care about. Their marketing people say that the price they can
get for a system directly depends on the SPEC rating the system can
attain. They literally have a formula that they use to calculate the
system price from the SPEC rating. Furthermore, for every additional 1%
of object code speed improvement that we might squeeze out of the
compiler, the marketing people can estimate how many additional millions
of dollars of revenue the product will produce for the company.


On the other hand, they couldn't care less about speed of compilation.
This is a complete reversal of the attitude that prevailed there a few
years ago.
--
John Polstra polstra!jdp@uunet.uu.net
John D. Polstra & Co., Inc. ...!uunet!polstra!jdp
Seattle, Washington USA Phone (206) 932-6482, FAX (206) 935-1262
--


Post a followup to this message

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