Path: chuka.playstation.co.uk!news
From: "Steve Dunn" <stevedunn@dial.pipex.com>
Newsgroups: scea.yaroze.programming.gnu_compiler
Subject: Re: long long ints for improved accuracy
Date: Mon, 16 Feb 1998 19:00:49 -0000
Organization: PlayStation Net Yaroze (SCEE)
Lines: 70
Message-ID: <6ca2bs$beo27@chuka.playstation.co.uk>
References: <673ein$9k37@chuka.playstation.co.uk>
NNTP-Posting-Host: usera830.uk.uudial.com
X-Newsreader: Microsoft Outlook Express 4.72.2106.4
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.2106.4

I've had a similar problem.  It's under the tree 'C++ Include paths' (near
the bottom(at the moment)).

If you have any luck resolving this, I'd appreciate a nudge.


Cheers John


Steve


john.rees@dial.pipex.com wrote in message
<673ein$9k37@chuka.playstation.co.uk>...
>I am trying to use long long ints to hold temporary values during
>fixed point calculations, but if I use divide and modulus operations
>then the linker complains that certain functions are not found.
>
>gcc basic.o  -Xlinker -Map -Xlinker mapfile -Xlinker -Ttext -Xlinker
>80100000 -o basic
>basic.o: In function `main':
>basic.c:11: undefined reference to `__divdi3'
>basic.c:11: relocation truncated to fit: JMPADDR __divdi3
>basic.c:12: undefined reference to `__moddi3'
>basic.c:12: relocation truncated to fit: JMPADDR __moddi3
>make.exe: *** [basic] Error 1
>
>It is strange that add, subtract and multiply seem to work, but divide
>and modulus are not provided. Even if these would be slow to execute
>on a 32 bit machine, it would be VERY useful.
>
>
>#include <stdio.h>
>
>#define printll(v) printf(#v "=%08x %08x\n", *(((int*)&v)+1),
>*(int*)&v )
>
>int main(void)
>{
> long long x = 0x123456;
> long long y = 0x100000;
> long long prod = (x-y)*y;
> long long quot = x/y;
> long long rem = x%y;
>
> printll(x);
> printll(y);
> printll(prod);
> printll(quot);
> printll(rem);
>
> return 0;
>}
>
>With the "quot" and "rem" lines commented out, the output is:
>
>x=00000000 00123456
>y=00000000 00100000
>prod=00000023 45600000
>
>Thanks to Alex Amsel for a throwaway remark in the More Pointless
>Optimisation Tricks thread that made me aware of long longs.
>After struggling with fixed point numbers for far too many hours I
>really hope someone can help with this.
>
>john
>
>