SLPVectorizer: add support for vectorization of diamond shaped trees. We now perform a preliminary traversal of the graph to collect values with multiple users and check where the users came from. 

llvm-svn: 179414

CostModel: increase the default cost of supported floating point operations from 1 to two. Fixed a few tests that changes because now the cost of one insert + a vector operation on two doubles is lower than two scalar operations on doubles.

llvm-svn: 179413

llvm-svn: 179412

patch by Veselov, Yuri <Yuri.Veselov@intel.com>.

llvm-svn: 179409

when parsing MS-style inline assembly.  No functional change intended.

llvm-svn: 179407

is a follow on to r179393 and r179399.  Test case to be added on
the clang side.
Part of rdar://13453209

llvm-svn: 179403

is a follow on to r179393.  Test case to be added on the clang side.
Part of rdar://13453209

llvm-svn: 179399

According to the ARM reference manual, constant offsets are mandatory for pre-indexed addressing modes.
The MC disassembler was not obeying this when the offset is 0.
It was producing instructions like: str r0, [r1]!.
Correct syntax is: str r0, [r1, #0]!.

This change modifies the dumping of operands so that the offset is always printed, regardless of its value, when pre-indexed addressing mode is used.

Patch by Mihail Popa <Mihail.Popa@arm.com>

llvm-svn: 179398

immediate displacement.  Specifically, add support for generating the proper IR.
We've been able to parse this for some time now.  Test case to be added on the
clang side.
Part of rdar://13453209

llvm-svn: 179393

TableGen will not combine nested list 'let' bindings into a single list, and
instead uses only the inner scope. As a result, several instruction definitions
were missing implicit register defs that were in outer scopes. This de-nests
these scopes and makes all instructions have only one let binding which sets
implicit register definitions.

llvm-svn: 179392

llvm-svn: 179391

llvm-svn: 179388

llvm-svn: 179387

The transform will execute like so:
(A & ~B) == 0 --> (A & B) != 0
(A & ~B) != 0 --> (A & B) == 0

llvm-svn: 179386

in r179325.  Test case coming shortly on the clang side.
Part of rdar://13453209

llvm-svn: 179383

Don't classify idiv/udiv as a reduction operation. Integer division is lossy.
For example : (1 / 2) * 4 != 4/2.

Example:

int a[] = { 2, 5, 2, 2}
int x = 80;

for()
  x /= a[i];

Scalar:
  x /= 2 // = 40
  x /= 5 // = 8
  x /= 2 // = 4
  x /= 2 // = 2

Vectorized:

 <80, 1> / <2,5> //= <40,0>
 <40, 0> / <2,2> //= <20,0>

 20*0 = 0

radar://13640654

llvm-svn: 179381

You can't copy an OwningPtr, and move semantics aren't available in C++98.

llvm-svn: 179374

llvm-svn: 179373

Fix a disconcerting bug in Value::isUsedInBasicBlock, which gave wrong answers for blocks larger than 3 instrs.

Also add a unit test. PR15727.

llvm-svn: 179370

This is prep. work for the implementation of optimizeCompare. Many PPC
instructions have 'record' forms (in almost all cases, this means that the RC
bit is set) that cause the result of the instruction to be compared with zero,
and the result of that comparison saved in a predefined condition register. In
order to add the record forms of the instructions without too much
copy-and-paste, the relevant functions have been refactored into multiclasses
which define both the record and normal forms.

Also, two TableGen-generated mapping functions have been added which allow
querying the instruction code for the record form given the normal form (and
vice versa).

No functionality change intended.

llvm-svn: 179356

llvm-svn: 179355

When debugging performance regressions we often ask ourselves if the regression
that we see is due to poor isel/sched/ra or due to some micro-architetural
problem.  When comparing two code sequences one good way to rule out front-end
bottlenecks (and other the issues) is to force code alignment. This pass adds
a flag that forces the alignment of all of the basic blocks in the program.

llvm-svn: 179353

Original message:

Print more information about relocations.

With this patch llvm-readobj now prints if a relocation is pcrel, its length,
if it is extern and if it is scattered.

It also refactors the code a bit to use bit fields instead of shifts and
masks all over the place.

llvm-svn: 179345

variables that use namespace alias qualifiers.  Test case coming on clang side
shortly.
Part of rdar://13499009

llvm-svn: 179343

namespace alias qualifiers.  Test case coming on clang side shortly.
Part of rdar://13499009

llvm-svn: 179339

Added PathAliases to check if two struct-path tags can alias.
Added command line option -struct-path-tbaa.

llvm-svn: 179337

can build up the identifier string.  No test case as support for looking up
these type of identifiers hasn't been implemented on the clang side.
Part of rdar://13499009

llvm-svn: 179336

specific logic.  This makes the code much less fragile.  Test case coming on the
clang side in a moment.
rdar://13634327

llvm-svn: 179323

A64Imms::isLogicalImmBits and A64Imms::isLogicalImm will attempt to
execute shifts that perform undefined behavior. Instead of attempting
to perform the 64-bit rotation, treat it as a no-op.

llvm-svn: 179317

Allows LLVM to optimize sequences like the following:

%add = add nsw i32 %x, 1
%cmp = icmp sgt i32 %add, %y

into:

%cmp = icmp sge i32 %x, %y

as well as:

%add1 = add nsw i32 %x, 20
%add2 = add nsw i32 %y, 57
%cmp = icmp sge i32 %add1, %add2

into:

%add = add nsw i32 %y, 37
%cmp = icmp sle i32 %cmp, %x

llvm-svn: 179316

multiply instructions in MipsSEDAGToDAGISel.

This patch was supposed to be part of r178403.

llvm-svn: 179314

- Rename function.
- Pass iterator by value.
- Remove header include.

No functionality changes.

llvm-svn: 179312

llvm-svn: 179303

With this patch llvm-readobj now prints if a relocation is pcrel, its length,
if it is extern and if it is scattered.

It also refactors the code a bit to use bit fields instead of shifts and
masks all over the place.

llvm-svn: 179294

When trying to collapse sequences of insertelement/extractelement
instructions into single shuffle instructions, there is one specific
case where the Instruction Combiner wrongly updates the resulting
Mask of shuffle indexes.

The problem is in function CollectShuffleElments.

If we have a sequence of insert/extract element instructions
like the one below:

  %tmp1 = extractelement <4 x float> %LHS, i32 0
  %tmp2 = insertelement <4 x float> %RHS, float %tmp1, i32 1
  %tmp3 = extractelement <4 x float> %RHS, i32 2
  %tmp4 = insertelement <4 x float> %tmp2, float %tmp3, i32 3

Where:
  . %RHS will have a mask of [4,5,6,7]
  . %LHS will have a mask of [0,1,2,3]

The Mask of shuffle indexes is wrongly computed to [4,1,6,7]
instead of [4,0,6,7].
When analyzing %tmp2 in order to compute the Mask for the
resulting shuffle instruction, the algorithm forgets to update
the mask index at position 1 with the index associated to the
element extracted from %LHS by instruction %tmp1.

Patch by Andrea DiBiagio!

llvm-svn: 179291

This will be used in clang to decide if it should create an @file or not. It
will be tested on the clang side.

Patch by Nathan Froyd.

llvm-svn: 179285

llvm-svn: 179280

llvm-svn: 179275

Rename the C function to create a SLPVectorizerPass to something sane and expose it in the header file.

llvm-svn: 179272

As packed comparisons in AVX/SSE produce all 0s or all 1s in each SIMD lane,
vector select could be simplified to AND/OR or removed if one or both values
being selected is all 0s or all 1s.

llvm-svn: 179267