TO_BINARY_DOUBLE
TO_BINARY_DOUBLE converts its argument to double-precision BINARY_DOUBLE. It accepts numbers, binary floating-point values, and text. The one-argument text form accepts decimal and scientific notation. Supply a numeric format model as the second argument to interpret formatted text.
SELECT TO_BINARY_DOUBLE('1.25e200') AS scientific,
TO_BINARY_DOUBLE('9,876.5', '9,999.9') AS formatted,
TO_BINARY_DOUBLE(16777217::NUMBER) =
TO_BINARY_DOUBLE(16777216::NUMBER) AS equal_after_conversion
FROM dual;
The query returns 1.25e+200, 9876.5, and f. Double precision distinguishes both integers in the comparison. Converting an already rounded BINARY_FLOAT to double precision keeps the rounded value.