The Partition DDS architecture supplies, an additional 48 dB of spurious free dynamic range (SFDR) in comparison to a DDS without any Noise Shaping. The maximum SFDR that can be achieved with this architecture is ~141 dB (theoretical maximum is 24-bit mantissa * 6). This is achieved by breaking the SINE/COSINE table into a coarse and a fine table and reconstructing the full SINE/COSINE at runtime. This architecture significantly reduces the memory requirements (2^24 vs 2^13) as the cost of some additional logic resources required to implement the reconstruction.