3 ms·
Thanks for the kind words! Your best bet for floating point determinism is to stick with doubles. Then, in 64 bit code, the double math will be done with the X
by WalterBright 2y ago
Thanks for the kind words!
Your best bet for floating point determinism is to stick with doubles. Then, in 64 bit code, the double math will be done with the XMM registers and instructions, which will stick with 64 bit arithmetic.