With regard to the AMD vs Intel bug:

The previous results narrowed it down significantly, but I'm going to
need more info to fix this.
Please, somebody with AMD add this to line to
std.math.internal.gammafunction line  624, add an import
core.stdc.stdio; in the file,

            if( x == 0.0L ) {
                di = 0.5;
                x = x0  +  di * (x1 - x0);
                if( x == 0.0 )
                    goto under;
            }
            y = betaIncomplete( a, b, x );
+            printf("%d %La %La %La %La %La %La %La\n", i, x, y, di,
yl, yh, a, b);
            yp = (x1 - x0)/(x1 + x0);
            if( fabs(yp) < dithresh )
                goto newt;
            yp = (y-y0)/y0;
            if( fabs(yp) < dithresh )
                goto newt;

compile Phobos, and then run this program:

import std.mathspecial;
import std.stdio;

void main()
{
writefln("%a", betaIncompleteInverse(0x1.ff1275ae5b939bcap-41, 4.6713e18,
 0.0813601));
}
_______________________________________________
phobos mailing list
[email protected]
http://lists.puremagic.com/mailman/listinfo/phobos

Reply via email to