mpgetflt(Mpflt *a)
{
int s, i;
- uvlong v;
+ uvlong v, vm;
double f;
if(a->val.ovf)
// independently or in the 6g half of the compiler
// pick up the mantissa in a uvlong
- s = 63;
+ s = 53;
v = 0;
for(i=Mpnorm-1; s>=Mpscale; i--) {
v = (v<<Mpscale) | a->val.a[i];
s -= Mpscale;
}
+ vm = v;
+ if(s > 0)
+ vm = (vm<<s) | (a->val.a[i]>>(Mpscale-s));
+
+ // continue with 64 more bits
+ s += 64;
+ for(; s>=Mpscale; i--) {
+ v = (v<<Mpscale) | a->val.a[i];
+ s -= Mpscale;
+ }
if(s > 0)
v = (v<<s) | (a->val.a[i]>>(Mpscale-s));
- // 63 bits of mantissa being rounded to 53
- // should do this in multi precision
- if((v&0x3ffULL) != 0x200ULL || (v&0x400) != 0)
- v += 0x200ULL; // round toward even
+//print("vm=%.16llux v=%.16llux\n", vm, v);
+ // round toward even
+ if(v != (1ULL<<63) || (vm&1ULL) != 0)
+ vm += v>>63;
- v >>= 10;
- f = (double)(v);
+ f = (double)(vm);
f = ldexp(f, Mpnorm*Mpscale + a->exp - 53);
if(a->val.neg)