That depends on the filter order. Evaluating a FIR is O(N M^2) where N is the number of data points and M is the size of the FIR. Doing the same with an FFT is

That depends on the filter order. Evaluating a FIR is O(N M^2) where N is the number of data points and M is the size of the FIR. Doing the same with an FFT is O(N Mlog(M)). Neither is particularly "performant." FFTs and FIRs are big guns!