Accuracy Analysis

Analyze accuracy of inaccurate adder/inaccurate multiplier by comparing it with accurate adder/accurate multiplier using two user provided unsigned numbers

Adder Accuracy Analysis

Accuracy Analysis

⚠️

Only integer inputs are accepted. Will prompt an error if the input are not valid integer

  1. Type of Accuracy Analysis to carry out, choose one of [Adder Accuracy Analysis, Multiplier Accuracy Analysis]

    • Default Options:
      • Adder Accuracy Analysis

  2. Total number of bits for approximate computing module

    • Default Options:
      • 4
    • Constraints:
      • 4 <= total_bits <= 32

  3. Accurate bits and Inaccurate bits selection. Move slider to adjust.

    • Default Options:
      • acc_bits = 1
      • inacc_bits = 3
    • Constraints:
      • 3 <= inacc_bits <= total_bits - 1

  4. Two unsigned decimal numbers to be added using accurate adder and to be compared against inaccurate adder for accuracy

    • Default Options: 12
    • Constraints:
      • First number: 1 <= first_number <= 2^(total_bits) -1
      • Second number: 1 <= second_number <= 2^(total_bits) -1
💡

Accurate sum cannot be zero in Accuracy analysis because they appear in denominator in the formula and will cause division by zero error. Both inputs for adder cannot be zero simulataneously

  1. Type of hardware module to analyse error, chose one of [HEAA, HOERAA, HOAANED, M-HERLOA]

    • Default Options: HEAA

  2. Start Analysing Accuracy

Multiplier Accuracy Analysis

Multiplier Accuracy Analysis

  1. Total Multiplicand bits and Total Multiplier bits

    • Default Options:
      • 4
    • Constraints:
      • multiplicand_bits >= 3
      • multiplier_bits >=3

  2. V-cut

    • Default Options:
      • 3
    • Constraints:
      • 0 <= v_cut <= multiplicand_bits + multiplier_bits - 3

  3. Two unsigned decimal numbers to be multiplied using accurate multiplier and compared against MxN AAM01 with V-cut for accuracy

    • Default Options:
      • 12
    • Constraints:
      • First number (multiplicand): 1 <= decimal_number <= 2^(multiplicand_bits) -1
      • Second number (multiplier): 1 <= decimal_number <= 2^(multiplier_bits) -1
💡

Accurate product cannot be zero in Accuracy analysis because they appear in denominator in the formula and will cause division by zero error. Either inputs for multiplier cannot be zero

  1. Type of hardware module to analyse accuracy

    • Default Options:
      • MxN AAM01 with V=cut

  2. Start Analysing Accuracy