 Roman Lebedev
|
bce7b42941
[Constant] Add 'isElementWiseEqual()' method
|
6 years ago |
|
Roman Lebedev
|
2f0aeb5b36
[InstCombine] Try to reuse constant from select in leading comparison
|
6 years ago |
 Sanjay Patel
|
eb85e5df9f
[InstCombine] simplify min/max of min/max with same operands (PR35607)
|
6 years ago |
|
Sanjay Patel
|
be0783132f
[InstCombine] canonicalize a scalar-select-of-vectors to vector select
|
6 years ago |
|
Roman Lebedev
|
fe1dee6170
[InstCombine] Non-canonical clamp-like pattern handling
|
6 years ago |
|
Roman Lebedev
|
a6ff94bf91
[InstCombine][NFC] Rename IsFreeToInvert() -> isFreeToInvert() for consistency
|
6 years ago |
|
Sanjay Patel
|
e2d5f7a219
[InstCombine] fold cmp+select using select operand equivalence
|
6 years ago |
|
Roman Lebedev
|
5ae3769ae0
[IR] SelectInst: add swapValues() utility
|
6 years ago |
 David Bolvansky
|
00d04e85f7
[InstCombine] Disable fold from D64285 for non-integer types
|
6 years ago |
|
David Bolvansky
|
e35805b819
[InstCombine] Fold select (icmp sgt x, -1), lshr (X, Y), ashr (X, Y) to ashr (X, Y))
|
6 years ago |
|
Sanjay Patel
|
0c13ef20a5
[InstCombine] canonicalize fcmp+select to minnum/maxnum intrinsics
|
6 years ago |
|
Sanjay Patel
|
04e978b788
[InstCombine] allow unordered preds when canonicalizing to fabs()
|
6 years ago |
|
Sanjay Patel
|
30d8a1f717
[InstCombine] fix bug in canonicalization to fabs()
|
6 years ago |
|
Sanjay Patel
|
dd29c50b32
[InstCombine] change canonicalization to fabs() to use FMF on fsub
|
6 years ago |
|
Sanjay Patel
|
79a352edc5
[InstCombine] allow unordered preds when canonicalizing to fabs()
|
6 years ago |
|
Sanjay Patel
|
a01a0c8205
[InstCombine] change canonicalization to fabs() to use FMF on fneg
|
6 years ago |
|
Sanjay Patel
|
fe9414d0cc
[InstCombine] sink FP negation of operands through select
|
6 years ago |
|
Sanjay Patel
|
f8e9cda8fc
[InstCombine] reduce code duplication; NFC
|
6 years ago |
 Nikita Popov
|
a600eb60fd
[InstCombine] Use uadd.sat and usub.sat for canonicalization
|
6 years ago |
|
Sanjay Patel
|
20aaebbdeb
[InstCombine] move add after smin/smax
|
6 years ago |
|
Sanjay Patel
|
77366ccba3
[InstCombine] move add after umin/umax
|
6 years ago |
|
Sanjay Patel
|
de27ddbd55
[InstCombine] canonicalize more unsigned saturated add with 'not'
|
6 years ago |
|
Sanjay Patel
|
92e8ea88cd
[InstCombine] reduce even more unsigned saturated add with 'not' op
|
6 years ago |
|
Sanjay Patel
|
783c1f039f
[InstCombine] rearrange saturated add folds; NFC
|
6 years ago |
|
Sanjay Patel
|
5f8e90cc7e
Revert "[InstCombine] reduce even more unsigned saturated add with 'not' op"
|
6 years ago |
|
Sanjay Patel
|
7e06b426b6
[InstCombine] reduce even more unsigned saturated add with 'not' op
|
6 years ago |
|
Sanjay Patel
|
6de5879eec
[InstCombine] reduce more unsigned saturated add with 'not' op
|
6 years ago |
|
Sanjay Patel
|
47d3493536
[InstCombine] reduce unsigned saturated add with 'not' op
|
6 years ago |
|
Sanjay Patel
|
6fa77014f0
[InstCombine] canonicalize cmp/select form of uadd saturate with constant
|
6 years ago |
 Chandler Carruth
|
6b547686c5
Update the file headers across all of the LLVM projects in the monorepo
|
6 years ago |
