Re: Fastcode CharPosRev B&V 0.6.0
- From: "John O'Harrow" <john@xxxxxxxxxxxxxxxxxxxx>
- Date: Mon, 27 Nov 2006 18:56:53 -0000
The negative offsets are only use to set base addresses. At no time is
memory outside of that allocated to the string ever referenced.
I am no concerned about needing the $R-. All this means is that the
programmer must take responsibility for range checking instead of the
compiler. The need for R- is very common where performance is the main
criteria (try compiling the RTL with range checking on!).
--
regards,
John
The Fastcode Project:
http://www.fastcodeproject.org/
"Davy Landman" <davy.landman@xxxxxxxxx> wrote in message
news:456b1192@xxxxxxxxxxxxxxxxxxxxxxxxx
Hi John,
Nice idea, but would it be an good idea to read then bytes for the start
of the string?
-4 = size
-8 = refcount...
but -12/10 ?
Looking at the code I can see it will probably be faster..
my code generates to an [esi+eax+$01-$04], while your code will lead to an
[esi + $00-$00].. nice idea :)
allthough too bad the R+ is needed..
Regards,
Davy
.
- Follow-Ups:
- Re: Fastcode CharPosRev B&V 0.6.0
- From: Davy Landman
- Re: Fastcode CharPosRev B&V 0.6.0
- References:
- Fastcode CharPosRev B&V 0.6.0
- From: Dennis
- Re: Fastcode CharPosRev B&V 0.6.0
- From: Davy Landman
- Re: Fastcode CharPosRev B&V 0.6.0
- From: John O'Harrow
- Re: Fastcode CharPosRev B&V 0.6.0
- From: Davy Landman
- Fastcode CharPosRev B&V 0.6.0
- Prev by Date: Re: Fastcode CharPosRev B&V 0.6.0
- Next by Date: Re: Fastcode CharPosRev B&V 0.6.0
- Previous by thread: Re: Fastcode CharPosRev B&V 0.6.0
- Next by thread: Re: Fastcode CharPosRev B&V 0.6.0
- Index(es):
Relevant Pages
|