circuit-fibsqrt（电路 fibsqrt）

Circuit-Fibsqrt

Overview

This skill provides guidance for implementing mathematical computations as gate-level circuits. It covers combinational logic (adders, comparators, multiplexers) and sequential logic (feedback-based iteration) in text-based circuit simulators that use event-driven simulation.

When to Use This Skill

Use this skill when:

• Building circuits that compute mathematical functions (square root, Fibonacci, etc.)
• Working with text-based gate netlists (e.g., gates.txt format)
• Implementing multi-bit arithmetic operations at the gate level
• Debugging circuits in event-driven simulators with feedback loops
• Optimizing gate counts to meet resource constraints

Approach: Component-First Development

Step 1: Understand the Simulator Semantics

Before writing any circuit code:

1. Read example files carefully - Examine any provided example gate files to understand:

- Input signal handling (e.g., out{i} = out{i} pattern for preserving inputs) - Gate syntax and argument ordering - How feedback loops are represented

1. Establish conventions - Document clearly:

- Mux semantics: Does mux(sel, a, b) select a when sel=0 or sel=1? - Signal naming conventions - Bit ordering (LSB vs MSB first)

1. Test simulator behavior - Create minimal test circuits to verify:

- Feedback loop behavior (toggle tests) - Multi-cycle convergence - Event-driven vs synchronous semantics

Step 2: Paper-Trace Algorithms First

Before implementing complex algorithms:

1. Work through small examples by hand - For integer square root, trace through isqrt(16), isqrt(17), isqrt(100)
2. Verify mathematical correctness - Ensure formulas are correct before coding (e.g., Newton-Raphson, binary search bounds)
3. Identify iteration counts - Determine how many iterations are needed for the input range

Step 3: Estimate Resource Usage

Before implementation:

1. Calculate expected gate counts - A 32-bit multiplier may require 30,000+ gates
2. Compare against limits - If limit is 32,000 gates, avoid multiplication-heavy approaches
3. Choose appropriate algorithms - Binary search isqrt avoids multiplication; comparison-based approaches are gate-efficient

Step 4: Build and Test Components Independently

Implement in isolation before combining:

1. Primitive gates first:

- AND, OR, XOR, NOT - MUX (multiplexer)

1. Arithmetic building blocks:

- Half adder, full adder - N-bit ripple-carry adder - N-bit subtractor (adder with inverted operand + carry-in) - N-bit comparator (A < B, A == B)

1. Test each component:

- Unit test adders with known values - Verify comparators with edge cases (0, max value, equal values) - Test mux selection in both directions

Step 5: Implement Algorithm-Specific Logic

For isqrt (integer square root):

• Binary search approach: Start with bounds [0, 2^16], narrow by comparing mid^2 with input
• Avoid direct multiplication if gate-constrained; use iterative addition or precomputed squares
• Handle edge cases: isqrt(0)=0, isqrt(1)=1

For Fibonacci:

• Implement as sequential state machine using feedback
• Two registers: fib_prev and fib_curr
• Iterate based on iteration count from isqrt result
• Handle edge cases: fib(0)=0, fib(1)=1

Step 6: Combine and Integrate

When combining components:

• Use clear signal naming to track data flow
• Verify interface widths match between components
• Test the combined circuit with known input/output pairs

Verification Strategies

Unit Testing

1. Test primitives in isolation - Create separate test scripts for each component
2. Use known test vectors - For adders: 0+0=0, 1+1=2, max+1=overflow
3. Edge case coverage:

- Zero inputs - Maximum value inputs - Boundary conditions (e.g., perfect squares for isqrt)

Integration Testing

1. Test with small inputs first - Verify fib(isqrt(1)), fib(isqrt(4)), fib(isqrt(9))
2. Verify intermediate values - Check isqrt output before Fibonacci computation
3. Test large inputs - Ensure overflow handling works correctly

Debugging Techniques

1. Add debug outputs - Temporarily expose internal signals for inspection
2. Trace signal propagation - Follow a known input through the circuit manually
3. Binary search for bugs - If output is wrong, test intermediate stages to isolate the issue

Common Pitfalls

Input Signal Handling

Mistake: Not preserving input signals in the gate file.

Solution: Many simulators require explicit input preservation:

out0 = out0   # Preserve input bit 0
out1 = out1   # Preserve input bit 1
...

Read example files to identify this pattern before implementation.

Algorithm Implementation Errors

Mistake: Implementing formulas without verification.

Example: Using test_val = 2*res + 1 when the correct formula is different for the chosen iteration method.

Solution: Paper-trace the algorithm with concrete values before coding.

Gate Count Explosion

Mistake: Using multiplication without estimating gate cost.

Example: A 32x32 multiplier can require 30,000+ gates, exceeding typical limits.

Solution:

• Estimate gates before implementation
• Prefer comparison-based or additive approaches when possible
• Consider iterative algorithms that reuse circuitry

Off-by-One Errors

Mistake: Getting fib(k-1) instead of fib(k) due to iteration count errors.

Solution:

• Clearly define what iteration 0, 1, 2,... produce
• Trace through fib(0), fib(1), fib(2) by hand to verify indexing

Feedback Loop Confusion

Mistake: Misunderstanding how feedback stabilizes in event-driven simulation.

Example: A toggle test showing 0 after even iterations is correct, not broken.

Solution:

• Test feedback loops with minimal circuits first
• Understand that value after N steps depends on initial value and operation

Mux Argument Order

Mistake: Confusing which input is selected when selector is 0 vs 1.

Solution:

• Establish and document convention at the start
• Test with a minimal mux circuit before using in larger designs

Recommended Workflow

1. Read all provided examples and documentation
2. Paper-trace the algorithm with test values
3. Estimate gate counts for chosen approach
4. Build and test primitive gates
5. Build and test arithmetic components
6. Build and test algorithm-specific logic
7. Integrate components
8. Test with edge cases
9. Optimize if needed (reduce gates, fix bugs)

Testing Checklist

• Input signals preserved correctly
• Adder produces correct sums
• Comparator handles all cases (less, equal, greater)
• isqrt(0) = 0
• isqrt(1) = 1
• isqrt(perfect square) = exact root
• fib(0) = 0
• fib(1) = 1
• Combined circuit matches expected outputs
• Gate count within limits