LabVIEW Idea Exchange
FIR Filter is almost the same as convolution, except that it has a init/cont terminal while convolution has an input for algorithm (Direct, frequency domain). FIR filter always uses direct convolution.
If "cont" is not used, convolution based FIR filtering could be orders of magnitude faster because it scales much less steeply with input sizes. Examples have been discussed where switching algorithms from "Direct" to "frequency domain" can turn minutes into seconds (e.g. 1M points and 5k filter).
While the knowledgeable programmer can of course make his own using the convolution primitives (also programming around other limitations because this idea is not implemented :(), it might be more intuitive if the FIR Filter had an "algorithm" input where we can select between the same choices as for convolution. (From my casual understanding, "frequency domain" would ignore the "cont" input because it is incompatible. This can just be mentioned in the help.)
