Dmitry Venikov 0bb74a59ff [InstSimplify] Missed optimization in math expression: squashing exp(log), log(exp) ...

Summary: This patch enables folding following expressions under -ffast-math flag: exp(log(x)) -> x, exp2(log2(x)) -> x, log(exp(x)) -> x, log2(exp2(x)) -> x

Reviewers: spatel, hfinkel, davide

Reviewed By: spatel, hfinkel, davide

Subscribers: scanon, llvm-commits

Differential Revision: https://reviews.llvm.org/D41381

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@321710 91177308-0d34-0410-b5e6-96231b3b80d8

2018-01-03 14:37:42 +00:00

..

2010-12-20-Boolean.ll

…

2011-01-14-Thread.ll

…

2011-02-01-Vector.ll

…

2011-09-05-InsertExtractValue.ll

InstructionSimplify: 'extractelement' with an undef index is undef

2017-12-06 17:51:46 +00:00

2011-10-27-BinOpCrash.ll

…

2011-11-23-MaskedBitsCrash.ll

…

2013-04-19-ConstantFoldingCrash.ll

…

add-mask.ll

Revert @llvm.assume with operator bundles (r289755-r289757)

2016-12-19 08:22:17 +00:00

addsub.ll

[InstSimplify] Add test cases for mixing add/sub i1 with xor of i1. Seems we can simplify in one direction but not the other.

2017-04-06 05:48:06 +00:00

and-icmps-same-ops.ll

[InstSimplify] fixed (?) to not mutate icmps

2016-12-06 22:09:52 +00:00

AndOrXor.ll

[InstSimplify] Make m_Not work for xor -1, X

2017-05-18 20:27:32 +00:00

assume.ll

[ValueTracking] emit a remark when we detect a conflicting assumption (PR31809)

2017-02-06 18:26:06 +00:00

bitcast-vector-fold.ll

[ConstantFold] Improve the bitcast folding logic for constant vectors.

2016-09-13 14:50:47 +00:00

bitreverse-fold.ll

Move tests associated with transforms moved in r321467

2017-12-30 03:13:00 +00:00

bitreverse.ll

[ValueTracking] Extend known bits to understand @llvm.bitreverse.

2017-01-17 17:23:51 +00:00

bswap.ll

…

call-callconv.ll

…

call.ll

InstSimplify: canonicalize is idempotent

2017-09-07 01:21:43 +00:00

cast.ll

[InstSimplify] Fold gep (gep V, C), (xor V, -1) to C-1

2016-08-16 06:13:46 +00:00

compare.ll

[InstCombine] remove unnecessary vector select fold; NFCI

2017-08-30 14:04:57 +00:00

dead-code-removal.ll

…

div.ll

[InstSimplify] fix some test names; NFC

2017-09-11 20:38:31 +00:00

exact-nsw-nuw.ll

[InstSimplify] fold sdiv/srem based on compare of dividend and divisor

2017-09-14 14:59:07 +00:00

exp2-intrinsic.ll

[InstSimplify] Missed optimization in math expression: squashing exp(log), log(exp)

2018-01-03 14:37:42 +00:00

exp-intrinsic.ll

[InstSimplify] Missed optimization in math expression: squashing exp(log), log(exp)

2018-01-03 14:37:42 +00:00

extract-element.ll

[instsimplify] consistently handle undef and out of bound indices for insertelement and extractelement

2017-12-30 05:54:22 +00:00

fast-math.ll

…

fdiv.ll

[InstSimplify] add constant folding for fdiv/frem

2017-04-01 19:05:11 +00:00

floating-point-arithmetic.ll

[ValueTracking] Implement SignBitMustBeZero correctly for sqrt.

2017-01-26 00:10:26 +00:00

floating-point-compare.ll

[InstSimplify] More fcmp cases when comparing against negative constants.

2017-12-01 12:34:16 +00:00

fold-builtin-fma.ll

…

fold-intrinsics.ll

Move tests associated with transforms moved in r321467

2017-12-30 03:13:00 +00:00

gep.ll

…

icmp-bool-constant.ll

[InstSimplify] handle all icmp i1 X, C in one place; NFCI

2017-05-17 20:27:55 +00:00

icmp-constant.ll

[ValueTracking] return zero when there's conflict in known bits of a shift (PR34838)

2017-10-12 17:31:46 +00:00

icmp-ranges.ll

[InstSimplify] use ConstantRange to simplify or-of-icmps

2017-05-07 15:11:40 +00:00

implies.ll

…

insertelement.ll

[instsimplify] consistently handle undef and out of bound indices for insertelement and extractelement

2017-12-30 05:54:22 +00:00

load-relative-32.ll

Introduce llvm.load.relative intrinsic.

2016-04-22 21:18:02 +00:00

load-relative.ll

Introduce llvm.load.relative intrinsic.

2016-04-22 21:18:02 +00:00

load.ll

[ConstantFolding] Extend FoldReinterpretLoadFromConstPtr to handle negative offsets

2016-07-13 23:33:07 +00:00

log2-intrinsic.ll

[InstSimplify] Missed optimization in math expression: squashing exp(log), log(exp)

2018-01-03 14:37:42 +00:00

log-intrinsic.ll

[InstSimplify] Missed optimization in math expression: squashing exp(log), log(exp)

2018-01-03 14:37:42 +00:00

logic-of-fcmps.ll

[InstSimplify] fold and/or of fcmp ord/uno when operand is known nnan

2017-11-19 15:34:27 +00:00

maxmin.ll

…

mul.ll

[InstSimplify] Teach SimplifyMulInst to recognize vectors of i1 as And. Not just scalar i1.

2017-04-06 17:33:37 +00:00

negate.ll

[InstSimplify] fold negation of sign-bit

2016-10-19 21:23:45 +00:00

noalias-ptr.ll

…

or-icmps-same-ops.ll

[InstSimplify] fixed (?) to not mutate icmps

2016-12-06 22:09:52 +00:00

or.ll

[InstSimplify] Use m_APInt instead of m_ConstantInt in ((V + N) & C1) | (V & C2) handling in order to support splat vectors.

2017-05-26 19:03:53 +00:00

past-the-end.ll

…

phi.ll

…

pr28725.ll

Reintroduce r320049, r320014 and r319894.

2017-12-13 11:21:18 +00:00

pr33957.ll

[ConstantFolder] Don't try to fold gep when the idx is a vector.

2017-07-27 22:20:44 +00:00

ptr_diff.ll

…

reassociate.ll

…

rem.ll

[InstSimplify] fix some test names; NFC

2017-09-11 20:38:31 +00:00

require-dominator.ll

Add unittest for r280760

2016-09-08 16:53:40 +00:00

returned.ll

Pointer-comparison folding should look through returned-argument functions

2016-07-11 03:37:59 +00:00

select.ll

[InstSimplify] teach SimplifySelectInst() to fold more vector selects

2017-10-02 23:43:52 +00:00

shift-128-kb.ll

…

shift-knownbits.ll

[InstCombine / InstSimplify] add and move tests for lshr transforms; NFC

2017-01-13 22:54:12 +00:00

shr-nop.ll

[ValueTracking] improve ComputeNumSignBits for vector constants

2016-06-22 19:20:59 +00:00

shufflevector.ll

InstructionSimplify: Relanding r301766

2017-05-07 18:16:37 +00:00

signed-div-rem.ll

[InstSimplify] fold sdiv/srem based on compare of dividend and divisor

2017-09-14 14:59:07 +00:00

simplify-nested-bitcast.ll

[InstSimplify] Constant fold the new GEP in SimplifyGEPInst.

2017-06-06 10:17:14 +00:00

undef.ll

[InstCombine] replace divide-by-constant checks with asserts; NFC

2016-08-30 17:31:34 +00:00

vec-cmp.ll

[ValueTracking, InstSimplify] extend isKnownNonZero() to handle vector constants

2016-05-24 14:18:49 +00:00

vector_gep.ll

[ConstantFold] Support vector index when factoring out GEP index into preceding dimensions

2017-12-04 19:56:33 +00:00

vector_ptr_bitcast.ll

…