From: "Lawrence Rust" <lvr AT NOsoftSPsystemAM DOT co DOT uk>
Newsgroups: comp.os.msdos.djgpp
References: <Pine DOT SUN DOT 3 DOT 91 DOT 1010808155014 DOT 25973A-100000 AT is> <9krgph$496$04$1 AT news DOT t-online DOT com>
Subject: Re: 128 bit integer
Lines: 43
Organization: Software Systems
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-Mimeole: Produced By Microsoft MimeOLE V5.50.4522.1200
Message-ID: <Ysdc7.12907$da2.54030@news11-gui.server.ntli.net>
Date: Wed, 8 Aug 2001 17:07:39 +0100
NNTP-Posting-Host: 213.107.106.28
X-Complaints-To: abuse AT ntlworld DOT com
X-Trace: news11-gui.server.ntli.net 997286840 213.107.106.28 (Wed, 08 Aug 2001 17:07:20 BST)
NNTP-Posting-Date: Wed, 08 Aug 2001 17:07:20 BST
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com


"Kai Dietrich" <toepferei DOT dietrich AT t-online DOT de> wrote in message
news:9krgph$496$04$1 AT news DOT t-online DOT com...
Message by Eli Zaretskii< eliz AT is DOT elta DOT co DOT il> on: 08.08.01 15:51:59
>
>>I want to do simple 128 bit integer math (+ and -). Do I have to combine
>>two long long int by myself or is there something like a library or a
>>little bit free source code which does that job?
>

[snip]

> I want the solution to behave like long long but with 128 bits
> (256 or more would be OK, too) instead of 64. I know that it is
> possible, it is the same solution like getting long long working
> on a 32bit machine. I would be able to combined two long long but
> I can only write + not - :-(. I need a little bit help writing
>
> _int128 add(_int128 a, _int128 b);
>
> _int128 sub(_int128 a, _int128 b);

Extended precision addition can be achieved in C as follows:

unsigned char a[2], b[2], sum[2], carry;

sum[0] = a[0] + b[0];
carry = sum[0] < a[0] ? 1 : 0;
sum[1] = a[1] + b[1] + carry;

Subtraction follows similarly.

This approach can be extended to unsigned integers of any size.  For best
efficiency replace char with the longest unsigned integer supported by the
implementation i.e. unsigned long long for DJGPP   Signed integers can be
represented by the 2's complement of the unsigned value.

NB it's not good etiquette to post in HTML to usenet groups.

-- Lawrence Rust
Software Systems www.softsystem.co.uk