31 lines
1.3 KiB
Text
31 lines
1.3 KiB
Text
###############################################################################
|
|
Software Floating Point Implementation Notes
|
|
###############################################################################
|
|
|
|
Big thanks to MageofMystra for providing a C implementation from which to base.
|
|
And for his ongoing assistance with mathematics questions.
|
|
|
|
For targets that do not have native support for floating-point operations, GDCC
|
|
will emulate them. Where possible and reasonably efficient, the implementation
|
|
follows the IEEE-754 standard.
|
|
|
|
===============================================================================
|
|
ZDoom ACS
|
|
===============================================================================
|
|
|
|
Supports both single and double precision formats.
|
|
|
|
Subnormal numbers are not supported and are never generated. Other special
|
|
exponent values (infinity, signed zero, NaN) are supported.
|
|
|
|
NaNs are supported, but may not be detected unless at least one mantissa bit in
|
|
the highest word is set. Generated NaNs always meet this requirement.
|
|
|
|
Rounding mode is always truncation.
|
|
|
|
Exceptions are not recorded and all NaNs are quiet.
|
|
|
|
###############################################################################
|
|
EOF
|
|
###############################################################################
|
|
|