[ont.general] strings statistics site

jstewart@sce.carleton.ca (John Stewart) (01/30/90)

I find it pretty amazing that vendors who are selling thousands of workstations
couldn't be bothered to optimize their string functions.  Last year I wrote
an ANSI C runtime library for our mainframe and it was trivial to write all
the string functions in assembler.  On a CISC machine most string functions 
effectively translate down to one or two instructions.  Our Sun-4 still 
blows the mainframe away though simply because it is a much faster machine.

-- 
preferred address: John_Stewart@carleton.ca	uucp: jstewart@sce.uucp
"Support the President's War On Long Usenet Signatures"

henry@utzoo.uucp (Henry Spencer) (02/07/90)

In article <770@sce.carleton.ca> jstewart@sce.UUCP (John Stewart) writes:
>I find it pretty amazing that vendors who are selling thousands of workstations
>couldn't be bothered to optimize their string functions.  Last year I wrote
>an ANSI C runtime library for our mainframe and it was trivial to write all
>the string functions in assembler.  On a CISC machine most string functions 
>effectively translate down to one or two instructions...

And my experimental C code beats them!  "Work smart, not hard."

It *is* annoying that vendors put so little effort into their libraries.
Actually, I get more annoyed about stdio than strings.  String functions
are seldom a major part of a program's execution, but slow implementations
of stdio waste billions of dollars of CPU time every year.
-- 
SVR4:  every feature you ever |     Henry Spencer at U of Toronto Zoology
wanted, and plenty you didn't.| uunet!attcan!utzoo!henry henry@zoo.toronto.edu