4 ms·
The conversion is optimized out by gcc-4.6 at all positive optimization levels, the addition is optimized at -O2 and above. But the slowness is caused by the al
by jedbrown 15y ago
The conversion is optimized out by gcc-4.6 at all positive optimization levels, the addition is optimized at -O2 and above. But the slowness is caused by the algorithms being different, with one producing denormals while the other stays reasonable.