Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!rutgers!sri-unix!hplabs!decwrl!sun!gorodish!guy
From: guy@gorodish.UUCP
Newsgroups: comp.arch,comp.lang.c
Subject: Re: String Processing Instruction
Message-ID: <15793@sun.uucp>
Date: Sat, 28-Mar-87 02:34:03 EST
Article-I.D.: sun.15793
Posted: Sat Mar 28 02:34:03 1987
Date-Received: Sun, 29-Mar-87 09:03:53 EST
References: <15292@amdcad.UUCP> <978@ames.UUCP> <15694@sun.uucp> <18036@ucbvax.BERKELEY.EDU> <4504@columbia.UUCP>
Sender: news@sun.uucp
Reply-To: guy@sun.UUCP (Guy Harris)
Distribution: na
Organization: Sun Microsystems, Mountain View
Lines: 24
Keywords: instruction set architectures, Am29000, printf
Xref: utgpu comp.arch:690 comp.lang.c:1365

>>The most obvious use of new string processing instructions would be to
>>improve the performance of printf.  Has anyone done this?

The version of "_doprnt" that comes with 4BSD and comes in the VAX System III
source (althought whether it's the one that's in the library or not,
I don't know), and may very well have come with UNIX 32V, is written
in assembler.  The 4.3BSD version, at least, does use the string
*and* EDITPC instructions, although it doesn't use the string
instructions for handling things like "%s".

VAX System V doesn't use this version (and no, it's probably not
because the assembly version doesn't handle all the formatting
operators in the S5 version - I know it handles most of them, and I
think it handles all of them), so I'm curious whether they dropped it
because it didn't buy much.

>We have experimented with a number of solutions, including a hardware
>printf coprocessor for the 68020.  Predictably, it improved performance
>quite a bit. We also have tried using specially coded versions of printf
>in assembly language that we passed through a (very time expensive)
>exhaustive search optimiser and the results were also pretty good.

How much did the coprocessor cost, and how much better was it than the
assembly-language version?