0
\$\begingroup\$

I'm having trouble with analyzing a circuit with respect to time, given clock-to-q, setup, and delays of individual components. We are given that

  • \$t_{clk-to-q} = 3ps\$
  • \$t_{setup}=4\$
  • \$t_{shifter}=1\$
  • \$t_{adder}=5\$
  • \$t_{multiplier}=6\$
  • \$t_{subtractor}=4\$.

We are also given that the inputs A, B, and C take on their new values exactly at the rising edge of every clock cycle and that all registers are intialized to zero.

This is the pipelined circuit:

enter image description here

(Sorry, some of the components I don't know how to represent in the circuit maker thing)

Anyway, the question asks:

How long does it take to compute the output for a given set of inputs? Assume the clock period is 11ps.

And the answer and explanation are given as:

Technically, 30ps is the time it will take for the values of one set of inputs to propogate to the output, We also accepted 41ps (3 clock cycles + clk-to-q + adder)

I'm confused on how they got that. Currently, I am calculating 31. Im taking the path from input B to register 2 to the output, which is (setup + adder + clock-to-q + multiplier+ adder + clock-to-q + adder) (I think)

What am I doing wrong?

DISCLAIMER: THIS IS NOT A HW QUESTION, I'M JUST STUDYING FOR FINALS AND DOING PRACTICE PROBLEMS

\$\endgroup\$
4
  • \$\begingroup\$ Use \$ for inline MathJAX on EE.SE. (It's $ on other sites.) \$\endgroup\$
    – Transistor
    May 10, 2021 at 21:14
  • \$\begingroup\$ @Transistor ty for letting me know, fixing now! EDIT - it doesnt seem to work :( I think it was displaying correctly before though \$\endgroup\$
    – Manny
    May 10, 2021 at 21:20
  • \$\begingroup\$ Have a look at the "edited x minutes ago" link to see how I fixed it. \$\endgroup\$
    – Transistor
    May 10, 2021 at 21:37
  • \$\begingroup\$ None of the answers make any sense as there are setup violations at multiple paths in the circuit, for eg: Reg4 to Reg2. \$\endgroup\$
    – Mitu Raj
    May 11, 2021 at 7:57

1 Answer 1

1
\$\begingroup\$

You failed on two points:

  1. You did not recognize that the prof is playing games with your head.
  2. The delay from register 4 to register 2 contains an adder and a multiplier, which have an aggregate delay of 11ps, and the registers have clock-to-q and setup delays,.

Add these extra delays to the clock-to-q delay for register 4 and the setup delay for register 2, and -- when the temperature and clock frequency are just right and the process hasn't changed lately -- that'll add a reliable 'a bit more than 1 clock delay' to that signal path.

\$\endgroup\$
2
  • \$\begingroup\$ im confused, aren't we taking the full path from input B to Output? Or is it just from register 2 to 4? \$\endgroup\$
    – Manny
    May 11, 2021 at 4:33
  • \$\begingroup\$ The question says "from the inputs to the outputs". So you have to look at all paths from all inputs to the output. \$\endgroup\$
    – TimWescott
    May 14, 2021 at 16:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.