[ARM] Turn on by default interleaved access vectorization

[sbaranga]

Summary

This change turns on by default interleaved access vectorization on ARM,
as it has shown to be beneficial on ARM.

[sbaranga]

Tested with lnt, spec2000 and some other internal benchmarks.

Performance Regressions - Execution Time
lnt.SingleSource/Benchmarks/Misc/himenobmtxpa 3.03%
lnt.SingleSource/Benchmarks/Shootout-C++/sieve 1.81%
lnt.MultiSource/Benchmarks/McCat/12-IOtest/iotest 1.36%

Performance Improvements - Execution Time
lnt.MultiSource/Benchmarks/PAQ8p/paq8p -18.90%
lnt.SingleSource/Benchmarks/Shootout-C++/EH/except -2.90%
lnt.MultiSource/Applications/ClamAV/clamscan -2.50%
lnt.SingleSource/UnitTests/Vectorizer/gcc-loops -1.59%
lnt.MultiSource/Applications/siod/siod -1.41%
lnt.SingleSource/Benchmarks/CoyoteBench/fftbench -1.33%
lnt.MultiSource/Benchmarks/VersaBench/bmm/bmm -1.19%

I think the paq8p change might be a performance variation, so overall no significant change in lnt/spec2000. I have see good improvements in other benchmarks though.

[rengolin]

LGTM. Thanks!

[sbaranga]

Committed in r246541.

Thanks,
Silviu

lib/Target/ARM/ARMTargetTransformInfo.h

public:
 
   // Provide value semantics. MSVC requires that we spell all of these out.
   ARMTTIImpl(const ARMTTIImpl &Arg)
       : BaseT(static_cast<const BaseT &>(Arg)), ST(Arg.ST), TLI(Arg.TLI) {}
   ARMTTIImpl(ARMTTIImpl &&Arg)
       : BaseT(std::move(static_cast<BaseT &>(Arg))), ST(std::move(Arg.ST)),
         TLI(std::move(Arg.TLI)) {}
 
+  bool enableInterleavedAccessVectorization() { return true; }
+
   /// \name Scalar TTI Implementations
   /// @{
 
   using BaseT::getIntImmCost;
   int getIntImmCost(const APInt &Imm, Type *Ty);
 
   /// @}
 

test/Transforms/LoopVectorize/ARM/interleaved_cost.ll

-; RUN: opt -S -debug-only=loop-vectorize -loop-vectorize -instcombine -enable-interleaved-mem-accesses=true  < %s 2>&1 | FileCheck %s
+; RUN: opt -S -debug-only=loop-vectorize -loop-vectorize -instcombine  < %s 2>&1 | FileCheck %s
 ; REQUIRES: asserts
 
 target datalayout = "e-m:e-i64:64-i128:128-n32:64-S128"
 target triple = "armv8--linux-gnueabihf"
 
 @AB = common global [1024 x i8] zeroinitializer, align 4
 @CD = common global [1024 x i8] zeroinitializer, align 4