NEW
ZX Spectrum Floating-Point Number Explorer
Explore the ZX Spectrum's five-byte floating-point
format and its short-integer format.
Enter a number, change the hexadecimal bytes or
click individual bits to see how the Spectrum
represents the value.
Five Bytes
Edit the hexadecimal bytes directly. Changes are immediately reflected below.
Binary
Edit Individual Bits
Click any bit to toggle it.
How the Spectrum Stores It
Floating point:
Byte 0 contains the exponent plus 128. The first bit of byte 1 contains the sign, while the remaining 31 bits contain the mantissa.
Short integer:
Byte 0 is zero. Byte 1 identifies the sign, bytes 2 and 3 contain the 16-bit value and byte 4 is zero.
Byte 0 contains the exponent plus 128. The first bit of byte 1 contains the sign, while the remaining 31 bits contain the mantissa.
Short integer:
Byte 0 is zero. Byte 1 identifies the sign, bytes 2 and 3 contain the 16-bit value and byte 4 is zero.
The -65536 INT Bug
The Spectrum ROM is inconsistent in its
treatment of integer values around
-65536.
The famous example is:
which produces: -1.
Values such as
The explorer deliberately preserves these Spectrum-specific oddities rather than correcting them mathematically.
The famous example is:
PRINT INT -65536
which produces: -1.
Values such as
INT -65535.3
and
INT -65535.5
can instead produce the tiny floating-point
value:
-1E-38.
The explorer deliberately preserves these Spectrum-specific oddities rather than correcting them mathematically.
Interpretation
Decimal value
Exponent
Mantissa
Sign
Last INT Operation
Not applied