GDCC
GDCC copied to clipboard
Proper float rounding
Currently, all float ops truncate rather than round in a controlled manner. This results in the accumulation of single ULP errors. Not a huge problem, but something to address eventually.