At the very least you need to
rehash toolbox
However, if you are copying flops from an old version of MATLAB then flops.m is very likely just the documentation for a built-in operation.
There has not been any meaningful way to experimentally measure Floating Point Operations Per Second (FlOPs) since 1993 when out-of-order execution and multiple arithmetic units were introduced in MIPS (tm) CPUs. Some would push the date further back to 1990 when IBM introduced fused multiplication and addition. flops never were particularly meaningful for performance measurements, as they deem a multiplication to be the same cost as an addition, which has never been the case. (Oddly enough, there have been a number of systems in which a floating point multiplication was faster than an integer addition -- because a lot of work was put into making floating point fast but integer work is a relatively small proportion of computation so it was not as important to speed up.)
If you are asked to calculation algorithmic complexity, you cannot do that by taking instruction measurements in any relatively modern system.
