|Beginner's Question... mihai@A-aod.resnet.ucsb.edu (Mihai Christodorescu) (1997-01-16)|
|Re: Beginner's Question... firstname.lastname@example.org (1997-01-16)|
|Re: Beginner's Question... email@example.com (John Lilley) (1997-01-16)|
|Re: Beginner's Question... firstname.lastname@example.org (Kurt Svensson) (1997-01-17)|
|Re: Beginner's Question... email@example.com (William D Clinger) (1997-01-17)|
|Re: Beginner's Question... firstname.lastname@example.org (John Lilley) (1997-01-19)|
|From:||email@example.com (Daniel J. Salomon)|
|Date:||16 Jan 1997 20:08:29 -0500|
|Organization:||Computer Science, University of Manitoba, Winnipeg, Canada|
Mihai Christodorescu <mihai@A-aod.resnet.ucsb.edu> wrote:
|> Why? It seems to me that it is more important for a compiler to
|> generate good and/or fast code, rather than compiling fast. Or maybe
|> there is a link between the two.
There are two marketing factors that affect the design of a compiler,
one is competition among computer manufacturers and the other is
competition among compiler writers.
Hardly anyone programs in assembler any more, so the way programmers
compare the speed of two processors is to try them on benchmark
programs in some high-level language. The trade magazines report the
results of these tests in order to compare the speed of the
processors. If the compiler does a sloppy job, it will make the
processor look bad. So the processor manufacturers put a lot of
pressure on their compiler writers to generate very efficient code so
that the processor will look faster and sell more.
In the competition among independent compiler writers, the speed of
both compilation and execution of the generated code are typically
reported in the trade magazines. Then the buyer can decide which is
more important to him. Presumably, for real software projects, the
generated code will be executed more often and by more people than the
source is compiled.
Daniel J. Salomon -- salomon@cs.UManitoba.CA
Dept. of Computer Science / University of Manitoba
Winnipeg, Manitoba, Canada R3T 2N2 / (204) 474-8687
Return to the
Search the comp.compilers archives again.