Why should I decrease the input signal frequency when increasing the FFT size

Question

I have a dds compiler with an IFFT block. The objective is to compute an IFFT of size 2048 of IQ data generated from dds Configuration of DDS compiler: System clock frequenc