- Jul 12, 1990
- Jul 10, 1990
-
-
wlott authored
define-primitive-object. Replaced the silly (eval `(make-... stuff with a call to make-.... Moved float-format-digits into the code sources, 'cause it's needed at run-time.
-
wlott authored
-
wlott authored
which included commits to RCS files with non-trunk default branches.
-
wlott authored
-
wlott authored
-
ram authored
lifetime problems.
-
ram authored
-
ram authored
which included commits to RCS files with non-trunk default branches.
-
ram authored
the number is positive.
-
ram authored
float.
-
- Jul 09, 1990
-
-
ram authored
-
- Jul 08, 1990
-
-
wlott authored
supposed to have compile-time parameters in it.
-
- Jul 07, 1990
-
-
ram authored
bignum and then divide 0 by it.
-
ram authored
FLOAT-DIGITS to use new constants. Macroized zerop, plusp, etc.
-
ram authored
-
ram authored
-
ram authored
-
ram authored
FLOAT-BIGNUM-RATIO. Now floating is done by grabbing high bits out of the bignum, rounding, and making them into a float with the correct exponent. This guarantees a correct result, and is also more efficient for large bignums.
-
wlott authored
instead of fixnum, 'cause some of the bignum code wants to use it to generate single-digit bignums.
-
wlott authored
-
wlott authored
also handle moving from descriptor-regs to fp-regs. This way we can pass floats in fp-regs in local call even if it starts in a descriptor-reg. Added sap-ref-{single,double} and setters for them.
-
wlott authored
-
wlott authored
which included commits to RCS files with non-trunk default branches.
-
wlott authored
alien-access stuff for single-floats and double-floats.
-
wlott authored
-
- Jul 06, 1990
- Jul 05, 1990
-
-
wlott authored
NIL. This way, we can use internal errors inside of assembly-routines, which don't have a vop to pass in.
-
wlott authored
multiply-fixnums so that it will not negate 0.
-
ram authored
Installed new interface to bignum floating, including use of FLOAT-BIGNUM-RATIO which ensures we don't cause spurious FP overflows.
-
ram authored
making a two-word bignum. Also, fixed the Lisp code for digit-0-or-plusp, which had negated sense.
-
ram authored
MULTIPLY-BIGNUMS/MAKE-SMALL-BIGNUM. Fixed fixnum GCD.
-
wlott authored
instead of single-float.
-
wlott authored
-