1996-12-17 - Re: ASM vs portable code [WAS: Re: Java DES breaker?]

Header Data

From: Adamsc@io-online.com (Adamsc)
To: “Dale Thorn” <trei@process.com>
Message Hash: 6cd245ace2536f9dbb24c827d0d92935a4444e80e2bc9ed64242b1203e37c2c5
Message ID: <19961217022038093.AAA188@gigante>
Reply To: N/A
UTC Datetime: 1996-12-17 02:23:32 UTC
Raw Date: Mon, 16 Dec 1996 18:23:32 -0800 (PST)

Raw message

From: Adamsc@io-online.com (Adamsc)
Date: Mon, 16 Dec 1996 18:23:32 -0800 (PST)
To: "Dale Thorn" <trei@process.com>
Subject: Re: ASM vs portable code [WAS: Re: Java DES breaker?]
Message-ID: <19961217022038093.AAA188@gigante>
MIME-Version: 1.0
Content-Type: text/plain


On Sat, 14 Dec 1996 21:28:46 -0800, Dale Thorn wrote:

>> Averaging several runs:
>> "C":   102,300 crypts/sec
>> ASM:   238,000 crypts/sec

>Someone on this list should know if it is possible to maximize speed in
>a typical 'C' routine, using Register variables (particularly for loops),
>inlining everything possible, etc., to get executable code much closer
>than a factor of 2x difference.  Can it be done on a PC, and how hard
>would it be to explain, to cover a representative variety of techniques?

A good optimizing compiler comes _close_ to hand coded, especially with the
many new concerns on Pentium+ processors (pipeline optimization, fpu tricks,
etc).

#  Chris Adams  <adamsc@io-online.com> | http://www.io-online.com/adamsc/adamsc.htp
#  <cadams@acucobol.com>                 | send mail with subject "send PGPKEY"
"That's our advantage at Microsoft; we set the standards and we can change them."
   --- Karen Hargrove, Microsoft (quoted in the Feb 1993 Unix Review editorial)







Thread