Arnold Schwaighofer a3fb330d05 LoopVectorizer: Recognize min/max reductions ...

A min/max operation is represented by a select(cmp(lt/le/gt/ge, X, Y), X, Y)
sequence in LLVM. If we see such a sequence we can treat it just as any other
commutative binary instruction and reduce it.

This appears to help bzip2 by about 1.5% on an imac12,2.

radar://12960601

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@179773 91177308-0d34-0410-b5e6-96231b3b80d8

2013-04-18 17:22:34 +00:00

..

ARM

…

X86

LoopVectorizer: Pass OperandValueKind information to the cost model

2013-04-04 23:26:27 +00:00

12-12-11-if-conv.ll

…

2012-10-20-infloop.ll

…

2012-10-22-isconsec.ll

…

bzip_reverse_loops.ll

…

calloc.ll

…

cast-induction.ll

…

cpp-new-array.ll

…

dbg.value.ll

LoopVectorizer: Ignore dbg.value instructions

2013-03-09 15:56:34 +00:00

flags.ll

…

float-reduction.ll

…

gcc-examples.ll

…

global_alias.ll

Test case hygiene.

2013-03-09 18:25:40 +00:00

i8-induction.ll

…

if-conv-crash.ll

…

if-conversion-reduction.ll

…

if-conversion.ll

…

increment.ll

…

induction_plus.ll

…

intrinsic.ll

LoopVectorize: Vectorize math builtin calls.

2013-02-27 15:24:19 +00:00

lcssa-crash.ll

…

lit.local.cfg

…

minmax_reduction.ll

LoopVectorizer: Recognize min/max reductions

2013-04-18 17:22:34 +00:00

no_idiv_reduction.ll

LoopVectorizer: integer division is not a reduction operation

2013-04-12 15:15:19 +00:00

no_int_induction.ll

…

nofloat.ll

…

non-const-n.ll

…

nsw-crash.ll

…

phi-hang.ll

LoopVectorize: Use a set to avoid longer cycles in the reduction chain too.

2013-04-18 14:29:13 +00:00

ptr_loops.ll

…

read-only.ll

…

reduction.ll

…

runtime-check.ll

…

same-base-access.ll

…

scalar-select.ll

…

simple-unroll.ll

…

small-loop.ll

…

start-non-zero.ll

…

struct_access.ll

…

vectorize-once.ll

PR14448 - prevent the loop vectorizer from vectorizing the same loop twice.

2013-03-02 01:33:49 +00:00

write-only.ll

…