 Andy Wingo
|
d1cf892880
Optimize fixnum or s64 -> f64 conversions
|
vor 6 Jahren |
|
Andy Wingo
|
9b3c4612bd
CPS type analysis support for mutable vs immutable vectors
|
vor 7 Jahren |
|
Andy Wingo
|
29fee39c2a
Variable renaming in type-fold.scm
|
vor 7 Jahren |
|
Andy Wingo
|
afb0a92d50
$branch is now a distinct CPS term type
|
vor 7 Jahren |
|
Andy Wingo
|
108ade6b0e
Re-add support for logbit?
|
vor 7 Jahren |
|
Andy Wingo
|
64acf24b40
Fix NaN handling in <= and >=
|
vor 7 Jahren |
|
Andy Wingo
|
40dac99d42
Integer comparison folding refactors
|
vor 7 Jahren |
|
Andy Wingo
|
695362a830
Better type folding for = on exact numbers
|
vor 7 Jahren |
|
Andy Wingo
|
6f3ae92b37
Improve type and range inference on bignums
|
vor 7 Jahren |
|
Andy Wingo
|
6a11fb1532
Add &exact-number helper definition
|
vor 7 Jahren |
|
Andy Wingo
|
c22e61a7ff
Fix u64/s64 typesafety around fixnum (un)tagging
|
vor 7 Jahren |
|
Andy Wingo
|
a48735f56c
Fix s64->scm reducer
|
vor 7 Jahren |
|
Andy Wingo
|
294dbaad35
Add support for comparisons against integer immediates
|
vor 7 Jahren |
|
Andy Wingo
|
4a0a930f1c
Don't fold comparisons that might contain a NaN.
|
vor 7 Jahren |
|
Andy Wingo
|
7e79a3291e
Better untagging of fixnums
|
vor 7 Jahren |
|
Andy Wingo
|
e8226be0c0
Type folding has "macro reduction" phase
|
vor 7 Jahren |
|
Andy Wingo
|
7a7f71de5c
Canonicalize <=, >=, and > primcalls to <
|
vor 7 Jahren |
|
Andy Wingo
|
17bd5a8938
Add lsh, rsh instructions
|
vor 7 Jahren |
|
Andy Wingo
|
c54c151eb6
$primcall has a "param" member
|
vor 7 Jahren |
|
Andy Wingo
|
0d42f5467f
Lower logtest branches to instead be 'zero? logand'
|
vor 7 Jahren |
|
Andy Wingo
|
d1c69b5c95
Remove compiler support for u64-scm comparisons
|
vor 7 Jahren |
|
Andy Wingo
|
31e7f44340
Add compiler support for fixnum? primcall predicate
|
vor 7 Jahren |
|
Andy Wingo
|
79a2748f83
Add compiler support for s64 comparisons.
|
vor 7 Jahren |
|
Andy Wingo
|
73d1502630
Lower eqv? and equal? to new instructions.
|
vor 7 Jahren |
|
Andy Wingo
|
c2fa345093
Add missing compiler support for heap-object? primcall et al.
|
vor 7 Jahren |
|
Andy Wingo
|
c119623e45
Simplify $branch to always take a $primcall
|
vor 7 Jahren |
|
Andy Wingo
|
1139c10e09
Heap type predicates preceded by heap-object?
|
vor 7 Jahren |
|
Andy Wingo
|
cd947a1161
Model all special immediates under one type bit (with range)
|
vor 7 Jahren |
|
Andy Wingo
|
2ca88789b1
Type inference distinguishes &fixnum and &bignum types
|
vor 7 Jahren |
 David Thompson
|
35a9059250
Add unboxed floating point comparison instructions.
|
vor 8 Jahren |
