Commit Graph

593 Commits

Author SHA1 Message Date
Arnold Schwaighofer
5f0d9dbdf4 X86 cost model: Adjust cost for custom lowered vector multiplies
This matters for example in following matrix multiply:

int **mmult(int rows, int cols, int **m1, int **m2, int **m3) {
  int i, j, k, val;
  for (i=0; i<rows; i++) {
    for (j=0; j<cols; j++) {
      val = 0;
      for (k=0; k<cols; k++) {
        val += m1[i][k] * m2[k][j];
      }
      m3[i][j] = val;
    }
  }
  return(m3);
}

Taken from the test-suite benchmark Shootout.

We estimate the cost of the multiply to be 2 while we generate 9 instructions
for it and end up being quite a bit slower than the scalar version (48% on my
machine).

Also, properly differentiate between avx1 and avx2. On avx-1 we still split the
vector into 2 128bits and handle the subvector muls like above with 9
instructions.
Only on avx-2 will we have a cost of 9 for v4i64.

I changed the test case in test/Transforms/LoopVectorize/X86/avx1.ll to use an
add instead of a mul because with a mul we now no longer vectorize. I did
verify that the mul would be indeed more expensive when vectorized with 3
kernels:

for (i ...)
   r += a[i] * 3;
for (i ...)
  m1[i] = m1[i] * 3; // This matches the test case in avx1.ll
and a matrix multiply.

In each case the vectorized version was considerably slower.

radar://13304919

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176403 91177308-0d34-0410-b5e6-96231b3b80d8
2013-03-02 04:02:52 +00:00
Benjamin Kramer
8611d4449a Cost model support for lowered math builtins.
We make the cost for calling libm functions extremely high as emitting the
calls is expensive and causes spills (on x86) so performance suffers. We still
vectorize important calls like ceilf and friends on SSE4.1. and fabs.

Differential Revision: http://llvm-reviews.chandlerc.com/D466

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176287 91177308-0d34-0410-b5e6-96231b3b80d8
2013-02-28 19:09:33 +00:00
Bill Wendling
351b7a10e2 Use references to attribute groups on the call/invoke instructions.
Listing all of the attributes for the callee of a call/invoke instruction is way
too much and makes the IR unreadable. Use references to attributes instead.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@175877 91177308-0d34-0410-b5e6-96231b3b80d8
2013-02-22 09:09:42 +00:00
Elena Demikhovsky
52981c4b60 I optimized the following patterns:
sext <4 x i1> to <4 x i64>
 sext <4 x i8> to <4 x i64>
 sext <4 x i16> to <4 x i64>
 
I'm running Combine on SIGN_EXTEND_IN_REG and revert SEXT patterns:
 (sext_in_reg (v4i64 anyext (v4i32 x )), ExtraVT) -> (v4i64 sext (v4i32 sext_in_reg (v4i32 x , ExtraVT)))
 
 The sext_in_reg (v4i32 x) may be lowered to shl+sar operations.
 The "sar" does not exist on 64-bit operation, so lowering sext_in_reg (v4i64 x) has no vector solution.

I also added a cost of this operations to the AVX costs table.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@175619 91177308-0d34-0410-b5e6-96231b3b80d8
2013-02-20 12:42:54 +00:00
Bill Wendling
7ab6c76ad1 Modify the LLVM assembly output so that it uses references to represent function attributes.
This makes the LLVM assembly look better. E.g.:

     define void @foo() #0 { ret void }
     attributes #0 = { nounwind noinline ssp }


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@175605 91177308-0d34-0410-b5e6-96231b3b80d8
2013-02-20 07:21:42 +00:00
Tim Northover
f13a7e2b21 AArch64: adjust tests which rely on a default JIT
Profiling tests *do* need a JIT. They'll pass if a cross-compiler targetting
AArch64 by default has been built, but fail if a native AArch64 compiler has
been build. Therefore XFAIL is inappropriate and we mark them unsupported.

ExecutionEngine tests are JIT by definition, they should also be unsupported.

Transforms/LICM only uses the interpreter to check the output is still sane
after optimisation. It can be switched to use an interpreter.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@175433 91177308-0d34-0410-b5e6-96231b3b80d8
2013-02-18 11:08:37 +00:00
Arnold Schwaighofer
6851623c54 ARM cost model: Add vector reverse shuffle costs
A reverse shuffle is lowered to a vrev and possibly a vext instruction (quad
word).

radar://13171406

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@174933 91177308-0d34-0410-b5e6-96231b3b80d8
2013-02-12 02:40:39 +00:00
Bill Schmidt
0f58dbae4a Refine fix to bug 15041.
Thanks to help from Nadav and Hal, I have a more reasonable (and even
correct!) approach.  This specifically penalizes the insertelement
and extractelement operations for the performance hit that will occur
on PowerPC processors.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@174725 91177308-0d34-0410-b5e6-96231b3b80d8
2013-02-08 18:19:17 +00:00
Arnold Schwaighofer
fb55a8fd7c ARM cost model: Address computation in vector mem ops not free
Adds a function to target transform info to query for the cost of address
computation. The cost model analysis pass now also queries this interface.
The code in LoopVectorize adds the cost of address computation as part of the
memory instruction cost calculation. Only there, we know whether the instruction
will be scalarized or not.
Increase the penality for inserting in to D registers on swift. This becomes
necessary because we now always assume that address computation has a cost and
three is a closer value to the architecture.

radar://13097204

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@174713 91177308-0d34-0410-b5e6-96231b3b80d8
2013-02-08 14:50:48 +00:00
Arnold Schwaighofer
66f535a273 ARM cost model: Add costs for vector selects
Vector selects are cheap on NEON. They get lowered to a vbsl instruction.

radar://13158753

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@174631 91177308-0d34-0410-b5e6-96231b3b80d8
2013-02-07 16:10:15 +00:00
Arnold Schwaighofer
e2d5590c33 ARM cost model: Cost for scalar integer casts and floating point conversions
Also adds some costs for vector integer float conversions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@174371 91177308-0d34-0410-b5e6-96231b3b80d8
2013-02-05 14:05:55 +00:00
Arnold Schwaighofer
a7ad84851b ARM cost model: Penalize insertelement into D subregisters
Swift has a renaming dependency if we load into D subregisters. We don't have a
way of distinguishing between insertelement operations of values from loads and
other values. Therefore, we are pessimistic for now (The performance problem
showed up in example 14 of gcc-loops).

radar://13096933

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@174300 91177308-0d34-0410-b5e6-96231b3b80d8
2013-02-04 02:52:05 +00:00
Hal Finkel
a8b289b70d Initial implementation of PPCTargetTransformInfo
This provides a place to add customized operation cost information and
control some other target-specific IR-level transformations.

The only non-trivial logic in this checkin assigns a higher cost to
unaligned loads and stores (covered by the included test case).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@173520 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-25 23:05:59 +00:00
Nadav Rotem
b338d897f9 Make opt grab the triple from the module and use it to initialize the target machine.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171341 91177308-0d34-0410-b5e6-96231b3b80d8
2013-01-01 08:00:32 +00:00
Dmitri Gribenko
a6542923b8 Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID
This is done to avoid odd test failures, like the one fixed in r171243.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171250 91177308-0d34-0410-b5e6-96231b3b80d8
2012-12-30 02:33:22 +00:00
Dmitri Gribenko
a51edf0986 Add a check to the test Analysis/ScalarEvolution/2010-09-03-RequiredTransitive.ll
This test did not test anything at all (except for opt crashing, but that was
not the reason why it was added).


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171248 91177308-0d34-0410-b5e6-96231b3b80d8
2012-12-30 01:42:34 +00:00
Dmitri Gribenko
b4912b9dcc Tests: rewrite 'opt ... %s' to 'opt ... < %s' so that opt does not emit a ModuleID
This is done to avoid odd test failures, like the one fixed in r171243.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171246 91177308-0d34-0410-b5e6-96231b3b80d8
2012-12-30 01:28:40 +00:00
Nadav Rotem
f85ec865f0 We are not ready to estimate the cost of integer expansions based on the number of parts. This test is too noisy.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@170999 91177308-0d34-0410-b5e6-96231b3b80d8
2012-12-23 09:11:07 +00:00
Nadav Rotem
f5637c3997 Improve the X86 cost model for loads and stores.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@170830 91177308-0d34-0410-b5e6-96231b3b80d8
2012-12-21 01:33:59 +00:00
Jakub Staszak
270bfbd3d1 Reverse order of checking SSE level when calculating compare cost, so we check
AVX2 before AVX.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@170464 91177308-0d34-0410-b5e6-96231b3b80d8
2012-12-18 22:57:56 +00:00
Arnold Schwaighofer
2b475922e6 Optimistically analyse Phi cycles
Analyse Phis under the starting assumption that they are NoAlias. Recursively
look at their inputs.
If they MayAlias/MustAlias there must be an input that makes them so.

Addresses bug 14351.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@169788 91177308-0d34-0410-b5e6-96231b3b80d8
2012-12-10 23:02:41 +00:00
Nadav Rotem
0602bb4659 Cost Model: change the default cost of control flow instructions (br / ret / ...) to zero.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@169423 91177308-0d34-0410-b5e6-96231b3b80d8
2012-12-05 21:21:26 +00:00
Preston Briggs
a4eff77e37 Modified dump() to provide a little
more information for dependences between
instructions that don't share a common loop.

Updated the test results appropriately.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@168965 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-30 00:44:47 +00:00
Preston Briggs
3c1cc3888b Modified depends() to recognize that when all levels are "=" and
there's no possible loo-independent dependence, then there's no
dependence.

Updated all test result appropriately.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@168719 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-27 19:12:26 +00:00
Preston Briggs
6ee74f52e9 Modify depends(Src, Dst, PossiblyLoopIndependent).
If the Src and Dst are the same instruction,
no loop-independent dependence is possible,
so we force the PossiblyLoopIndependent flag to false.

The test case results are updated appropriately.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@168678 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-27 06:41:46 +00:00
Preston Briggs
72a2c0622a Corrects a problem where we reply exclusively of GEPs to drive
analysis.  Better is to look for cases with useful GEPs and use them
when possible.  When a pair of useful GEPs is not available, use the
raw SCEVs directly. This approach supports better analysis of pointer
dereferencing.

In parallel, all the test cases are updated appropriately.
Cases where we have a store to *B++ can now be analyzed!


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@168474 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-21 23:50:04 +00:00
Hal Finkel
c37f502d48 Phi speculation improvement for BasicAA
This is a partial solution to PR14351. It removes some of the special
significance of the first incoming phi value in the phi aliasing checking logic
in BasicAA. In the context of a loop, the old logic assumes that the first
incoming value is the interesting one (meaning that it is the one that comes
from outside the loop), but this is often not the case.  With this change, we
now test first the incoming value that comes from a block other than the parent
of the phi being tested.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@168245 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-17 02:33:15 +00:00
Benjamin Kramer
563e8fce2e DependenceAnalysis: Print all dependency pairs when dumping. Update all testcases.
Part of a patch by Preston Briggs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167827 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-13 12:12:02 +00:00
Nadav Rotem
a6fb97a49a CostModel: add another known vector trunc optimization.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167488 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-06 21:17:17 +00:00
Nadav Rotem
b042868c01 Cost Model: add tables for some avx type-conversion hacks.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167480 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-06 19:33:53 +00:00
Nadav Rotem
7ae3bcca45 CostModel: Add tables for the common x86 compares.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167421 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-05 23:48:20 +00:00
Nadav Rotem
2d1528b358 Code Model: Improve the accuracy of the zext/sext/trunc vector cost estimation.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167412 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-05 22:20:53 +00:00
Nadav Rotem
a4ab5290e6 Cost Model: Normalize the insert/extract index when splitting types
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167402 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-05 21:12:13 +00:00
Nadav Rotem
75138f58b0 Cost Model: teach the cost model about expanding integers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167401 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-05 21:11:10 +00:00
Nadav Rotem
e623702c22 Implement the cost of abnormal x86 instruction lowering as a table.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167395 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-05 19:32:46 +00:00
Richard Osborne
544533301e Don't infer whether a value is captured in the current function from the
'nocapture' attribute.

The nocapture attribute only specifies that no copies are made that
outlive the function. This isn't the same as there being no copies at all.
This fixes PR14045.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167381 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-05 10:48:24 +00:00
Nadav Rotem
b4b04c3fa0 X86 CostModel: Add support for a some of the common arithmetic instructions for SSE4, AVX and AVX2.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167347 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-03 00:39:56 +00:00
Nadav Rotem
0c31e43ff3 Add a stub for the x86 cost model impl. Implement a basic cost rule for inserting/extracting from XMM registers.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167333 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-02 23:27:16 +00:00
Nadav Rotem
f149567160 CostModel: add support for Vector Insert and Extract.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167329 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-02 22:31:56 +00:00
Nadav Rotem
6bed58ef24 Add a cost model analysis that allows us to estimate the cost of IR-level instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167324 91177308-0d34-0410-b5e6-96231b3b80d8
2012-11-02 21:48:17 +00:00
Benjamin Kramer
b8b3f6081f Remove LoopDependenceAnalysis.
It was unmaintained and not much more than a stub. The new DependenceAnalysis
pass is both more general and complete.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@166810 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-26 20:25:01 +00:00
Sebastian Pop
ad43499fc4 dependence analysis
Patch from Preston Briggs <preston.briggs@gmail.com>.

This is an updated version of the dependence-analysis patch, including an MIV
test based on Banerjee's inequalities.

It's a fairly complete implementation of the paper

    Practical Dependence Testing
    Gina Goff, Ken Kennedy, and Chau-Wen Tseng
    PLDI 1991

It cannot yet propagate constraints between coupled RDIV subscripts (discussed
in Section 5.3.2 of the paper).

It's organized as a FunctionPass with a single entry point that supports testing
for dependence between two instructions in a function. If there's no dependence,
it returns null. If there's a dependence, it returns a pointer to a Dependence
which can be queried about details (what kind of dependence, is it loop
independent, direction and distance vector entries, etc). I haven't included
every imaginable feature, but there's a good selection that should be adequate
for supporting many loop transformations. Of course, it can be extended as
necessary.

Included in the patch file are many test cases, commented with C code showing
the loops and array references.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@165708 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-11 07:32:34 +00:00
James Molloy
9e36496eb3 Add default JIT LIT variable.
Patch by David Tweed!



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164996 91177308-0d34-0410-b5e6-96231b3b80d8
2012-10-02 10:57:08 +00:00
Duncan Sands
44401b7c80 Now that invoke of an intrinsic is possible (for the llvm.do.nothing intrinsic)
teach the callgraph logic to not create callgraph edges to intrinsics for invoke
instructions; it already skips this for call instructions.  Fixes PR13903.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@164707 91177308-0d34-0410-b5e6-96231b3b80d8
2012-09-26 17:16:01 +00:00
Arnold Schwaighofer
3d5f96ee1b BasicAA: Recognize cyclic NoAlias phis
Enhances basic alias analysis to recognize phis whose first incoming values are
NoAlias and whose other incoming values are just the phi node itself through
some amount of recursion.

Example: With this change basicaa reports that ptr_phi and ptr_phi2 do not alias
each other.

bb:
 ptr = ptr2 + 1

loop:
  ptr_phi = phi [bb, ptr], [loop, ptr_plus_one]
  ptr2_phi = phi [bb, ptr2], [loop, ptr2_plus_one]
  ...
  ptr_plus_one = gep ptr_phi, 1
  ptr2_plus_one = gep ptr2_phi, 1

This enables the elimination of one load in code like the following:

extern int foo;

int test_noalias(int *ptr, int num, int* coeff) {
  int *ptr2 = ptr;
  int result = (*ptr++) * (*coeff--);
  while (num--) {
    *ptr2++ = *ptr;
    result +=  (*coeff--) * (*ptr++);
  }
  *ptr = foo;
  return result;
}

Part 2/2 of fix for PR13564.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163319 91177308-0d34-0410-b5e6-96231b3b80d8
2012-09-06 14:41:53 +00:00
Arnold Schwaighofer
029032693f BasicAA: GEPs of NoAlias'ing base ptr with equivalent indices are NoAlias
If we can show that the base pointers of two GEPs don't alias each other using
precise analysis and the indices and base offset are equal then the two GEPs
also don't alias each other.
This is primarily needed for the follow up patch that analyses NoAlias'ing PHI
nodes.

Part 1/2 of fix for PR13564.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@163317 91177308-0d34-0410-b5e6-96231b3b80d8
2012-09-06 14:31:51 +00:00
NAKAMURA Takumi
c3c237d8be llvm/test/Analysis/Profiling: Mark 3 of them as REQUIRES: loadable_module.
FIXME: profile_rt.dll could be built on win32.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162811 91177308-0d34-0410-b5e6-96231b3b80d8
2012-08-29 00:37:46 +00:00
Manman Ren
d26200423e Profile: set branch weight metadata with data generated from profiling.
This patch implements ProfileDataLoader which loads profile data generated by
-insert-edge-profiling and updates branch weight metadata accordingly.

Patch by Alastair Murray.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162799 91177308-0d34-0410-b5e6-96231b3b80d8
2012-08-28 22:21:25 +00:00
Manman Ren
1a710fdde1 BranchProb: modify the definition of an edge in BranchProbabilityInfo to handle
the case of multiple edges from one block to another.

A simple example is a switch statement with multiple values to the same
destination. The definition of an edge is modified from a pair of blocks to
a pair of PredBlock and an index into the successors.

Also set the weight correctly when building SelectionDAG from LLVM IR,
especially when converting a Switch.
IntegersSubsetMapping is updated to calculate the weight for each cluster.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162572 91177308-0d34-0410-b5e6-96231b3b80d8
2012-08-24 18:14:27 +00:00
Benjamin Kramer
4e81d40545 Fix broken check lines.
I really need to find a way to automate this, but I can't come up with a regex
that has no false positives while handling tricky cases like custom check
prefixes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@162097 91177308-0d34-0410-b5e6-96231b3b80d8
2012-08-17 12:28:26 +00:00