Normalize a decimal or binary number into scientific notation and binary floating-point form, or solve F = N × 2(E − B) with IEEE bias presets.

Decimal accepts e notation; binary accepts 0b and a binary point.

Optional settings

Related Calculators

Floating-Point Significand and Exponent Formula (IEEE-754)

The following formulas relate a signed finite value F to a signed significand N, exponent E and bias B. For an IEEE-754 normal encoding, 1 ≤ |N| < 2 and E must be an integer in the normal field range. Zero, subnormal values and special encodings require different rules. The calculator also permits general algebra with custom real E and B.

F = N × 2(E − B)
N = (F) / (2(E − B))

Variables:

  • N is the normalized significand (mantissa). For a binary normalized value, it is typically in the range 1 ≤ |N| < 2 (for nonzero “normal” numbers).
  • F is the signed real-number value. In these equations the sign is included in F and N; an IEEE encoding stores it separately.
  • E is the exponent field stored in the floating-point format (a biased exponent).
  • B is the exponent bias for the chosen format.

To get the normalized significand from the value, divide F by 2 raised to the power of (E − B), where (E − B) is the unbiased exponent. To reconstruct the value, multiply the significand by the same power of 2.

What is Floating-Point Normalization?

Floating-point normalization is the process of writing a nonzero number in a standard scientific form. For binary floating-point (such as IEEE-754), this means expressing the magnitude as N × 2k where the significand N is in the range 1 ≤ N < 2. Normalization ensures the leading binary digit is 1, which uses the available significand bits efficiently and maximizes precision for a given bit width. In IEEE-754 formats, the exponent is stored using a bias (E = k + B) so that the exponent field can represent both positive and negative exponents without a separate sign bit.

How to Calculate Floating-Point Normalization?

The following steps outline how to calculate the Floating-Point Normalization.

  1. First, determine the real-number value you want to represent (F). Track the sign separately if the value is negative.
  2. Next, convert the magnitude of F into normalized binary scientific form: F = N × 2k with 1 ≤ N < 2 (for nonzero normal numbers).
  3. Next, determine the bias (B) for the floating-point format you are using (for example, B = 127 for IEEE-754 single precision).
  4. Compute the stored (biased) exponent field as E = k + B.
  5. Finally, you can verify (or solve for) a missing variable using N = F / 2(E − B) or F = N × 2(E − B), and check your answer with the calculator above.

Example Problem : 

Use the following variables as an example problem to test your knowledge.

Floating-point Value (F) = 8.5

Exponent Field (E) = 4

Bias (B) = 1

Then the unbiased exponent is (E − B) = 3, so the normalized value is N = 8.5 / 23 = 1.0625 (which is 1.0001 in binary).