Quote:
Originally Posted by Greg McKaskle
The LV compiler supports optimization transforms, but I don't expect there to be one that will turn the select and negate into the XOR. So I expect the XOR to be just a bit more efficient. I put the code into timing sequences and ran it 100,000,000 times and indeed, at least for intel code gen, the XOR is just a bit faster.
|
Thanks Greg. I realize the difference, if any, is likely miniscule, but would it be reasonable to assume (based on your testing above) that the bottom sketch in the attached PNG is also slightly more efficient than the top one?