- From: Paul Cheyrou-lagreze <Paul.Cheyrou-Lagreze@inrialpes.fr>
- Date: Tue, 11 Mar 2003 18:17:53 +0100
- To: Damian Steer <D.M.Steer@lse.ac.uk>
- Cc: www-amaya-dev@w3.org
> I assume it works on non-x86 linux?
Endianness hell...
Well, I think the incriminated code is in Amaya/thotlib/dialogue/openglfont.c
the p2() macro/function
If you could try with :
int p2(int p)
{
p -= 1;
p |= p >> 16;
p |= p >> 8;
p |= p >> 4;
p |= p >> 2;
p |= p >> 1;
return p + 1;
}
otherwise, identical, formulical, but VERY VERY VERY slower and surely the one that works on non-x86... :
int p2(int p)
{
return 1 << (int) ceilf(logf((float) p) / M_LN2);
}
or finally the slower than snail one :
int p2 (p)
{
return p*p;
}
- If you know an optimized version of p*p on non-x86, feel free to try there -
-Paul
Received on Tuesday, 11 March 2003 12:17:42 UTC