Hello,
I have a question regarding Tesla c1060 or T10 series. Is there only one ALU for both integer operation and floating point operation on one SP (or one core)? In other words, can we do integer operation and floating point operation at the same time (if we have two separate ALUs) or we can only do one at a time?
Any idea?
They are the same, so you can either do an integer operation or a floating point operation (including fmad, that usually counts as two operations) on the ALU.
However the special function unit can also do floating point multiplication. So if that is not used for something else, you should be able to do an integer operation and a floating point multiplication at the same time.
I have not tried to confirm this in practice, though.