A Novel Address Scheme for Continuous-Flow Parallel Memory-Based Real-Valued FFT Processor
Round 1
Reviewer 1 Report
This is a very good paper that proposes a memory-based real-valued FFT architecture based on constant geometry.
The manuscript is well written and organized, and the theory seems correct. The comparisons with other prior works are interesting.
Overall it is a solid work with interesting performance in terms of circuit resources and frequency efficiency.
Please correct some minor typos e.g. at row 185 "deigned", and so on.
Reviewer 2 Report
The authors present a modified constant-geometry based signal flow graph for memory-based real-valued fast Fourier transform architecture. The proposed method exibits low memory requirements and low computaional complexity. In addition it meets the constraint of in-place operation, concurrent I/O, normal-order I/O, variable size, and parallel processing. The presented experimental results confirm the efficiency of the proposed address scheme.
The paper is mostly well written.
