Re: [help-GIFT] Re: Patches AMD 64 next try...

[risc]

On Wed, Feb 28, 2007 at 03:33:45PM +0100, Wolfgang Müller wrote:
> On Tuesday 27 February 2007 21:11, you wrote:
> 
> Great! GDB is good :-) .
> 
> > this goes away if you change x or y into a uint64_t.
> >
> > i'm scared. i'd love an answer to this one. other than "magic."
> 
> No it's not magic :-) in one case you do a 32 bit addition to a 64 bit value. 
> Bit 31 overflows, but it is ignored. In fact, you have done a addition of a 
> negative 32 bit number to the 64 bit base address.
> 
> If you do a 64 bit addition, bit 31 overflows into bit 32 and is taken into 
> account. Everything right. "Alles palletti :-) " as some people say over 
> here. Any protests?
> 
> This seems to be it, correct? We just have to check that all indexes stay in 
> our 256x256 square.
> 
> Cheers,
> Wolfgang
> 
> -- 
> Dr. Wolfgang Müller
> LS Medieninformatik
> Universität Bamberg
> Check out the SIG MM web site http://www.sigmm.org


OK, i'm dumb. ;)

its worth mentioning, the fix for this should likely be the same as the "hack" 
in my patch.

uint_fast32_t is 32bit on ia32, and 64bit on AMD64.

i'll try to compile gift on my PPC sometime soon (playstation 3 power/cell), 
and make sure everything compiles right.

Julia Longtin <address@hidden>