Arnold Schwaighofer 074e489dbc SLPVectorizer: Handle more horizontal reductions (disabled) ...

Match reductions starting at binary operation feeding into a phi. The code
handles trees like

 r += v1 + v2 + v3 ...

and

 r += v1
 r += v2
 ...

and

 r *= v1 + v2 + ...

We currently only handle associative operations (add, fadd fast).

The code can now also handle reductions feeding into stores.

 a[i] = v1 + v2 + v3 + ...

The code is currently disabled behind the flag "-slp-vectorize-hor".  The cost
model for most architectures is not there yet.

I found one opportunity of a horizontal reduction feeding a phi in TSVC
(LoopRerolling-flt) and there are several opportunities where reductions feed
into stores.

radar://14607682

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@191108 91177308-0d34-0410-b5e6-96231b3b80d8

2013-09-20 21:18:20 +00:00

..

BBVectorize.cpp

BBVectorize: Add initial stores to the write set when tracking uses

2013-08-13 23:34:32 +00:00

CMakeLists.txt

SLP Vectorizer: Implement multi-block slp-vectorization.

2013-06-22 21:34:10 +00:00

LLVMBuild.txt

…

LoopVectorize.cpp

Prevent LoopVectorizer and SLPVectorizer running if the target has no vector registers.

2013-09-18 12:43:35 +00:00

Makefile

…

SLPVectorizer.cpp

SLPVectorizer: Handle more horizontal reductions (disabled)

2013-09-20 21:18:20 +00:00

Vectorize.cpp

This patch breaks up Wrap.h so that it does not have to include all of

2013-05-01 20:59:00 +00:00