Suppose you have the following processor as in your project.
Design a circuit to implement the 4-point FFT on a processor. You can assume x0, x1, x2 and x3
are always initially stored in R0, R1, R2 and R3
1. Using a Finite State Machine
2. Using additional hardware that could be implement the FFT In a single clock cycle. Draw
a diagram of a parallel circuit using as little hardware as possible (you can use
adders/registers/basic gates/multipexers)
3. Assume the following delays:
• Tsetup/thold/dq = negligible
• Tristate buffer = 1ns
• Adder = 10ns
• Bus delays = 1ns
• Basic Gates = 2ns
• Multiplexer = 2ns
What is the maximum clock frequency of each design?
4. Discuss advantages/disadvantages of each implementation.
5. Discuss how to use pipelining to speed up your answer to question 2, what is your new
maximum clock frequency?