This topic describes the Firebolt implementation of the NUMERIC data type.

​
Overview

The NUMERIC data type is an exact numeric data type defined by its precision (total number of digits) and scale (number of digits to the right of the decimal point).

NUMERIC has two optional input parameters: NUMERIC(precision, scale)

The maximum precision is 38. The scale value is bounded by the precision value (scale<=precision).

The precision must be positive, while the scale can be zero or positive.

The DECIMAL data type is a synonym to the NUMERIC data type.

​
Storage size for precision

The storage size of the data type depends on the precision, as described in the following table:

precisionbytes
1-94
10-188
18-3816

For example, if the precision is between 1 and 9 digits, the data type will use 4 bytes of storage.

​
Default values for precision and scale

If the scale is not specified when declaring a column of NUMERIC data type, then it defaults to NUMERIC(precision, min(9, precision))

If both the precision and scale are not specified, then it defaults to NUMERIC(38, 9)

​
Precision vs. scale

If the scale of a value to be stored is greater than the declared scale of the column, the system will round the value to the specified number of fractional digits. If the number of digits to the left of the decimal point exceeds the declared precision, minus the declared scale, an error results.

SELECT CAST(100.76 AS NUMERIC(5,2)); -- 100.76
SELECT CAST(100.76 AS NUMERIC(5,1)); -- 100.8
SELECT(100.76 AS NUMERIC(3,1)); -- error

​
Type conversion (type promotion)

Note: NUMERIC precision and scale does not change with any operation, i.e., there is no automatic widening of the result type to prevent overflows. If a wider type is desired, cast the inputs to perform the operations on wider types.

  1. P1≠P2 or S1≠S2 (casting required)

    Any operation between two NUMERICs with different precision and/or scale requires explicit casting of the input to the desired precision and scale.

    NUMERIC(P1, S1) <OP> NUMERIC(P2, S2) -> ERROR (P1≠P2 or S1≠S2)

    where P1≠P2 or S1≠S2

  2. P1=P2 and S1=S2 (casting optional)

    If the two NUMERICs have the same precision and scale, the result will implicitly cast to the same precision and scale. You can still explicitly cast to any other precision and scale.

    NUMERIC(P1, S1) <OP> NUMERIC(P2, S2) -> NUMERIC(P1, S1)

​
Supported operators

Operators:

+
-
*
/