We can prove that some delinearized subscripts do not wrap around to become
negative by the fact that they are from inbound geps of load/store locations.
This helps improve the delinearisation in cases where we can't prove that they
are non-negative from SCEV alone.
Differential Revision: https://reviews.llvm.org/D48481
llvm-svn: 335481
AArch64 was only setting costs for SK_Transpose, which meant that many of the simpler shuffles (e.g. SK_Select and SK_PermuteSingleSrc for larger vector elements) was being severely overestimated by the default shuffle expansion.
This patch adds costs to help improve SLP performance and avoid a regression in reductions introduced by D48174.
I'm not very knowledgeable about AArch64 shuffle lowering so I've kept the extra costs to a minimum - someone who knows this code can add extra costs which should improve vectorization a lot more.
Differential Revision: https://reviews.llvm.org/D48172
llvm-svn: 335329
Summary:
This initiates a discussion on changing Polly accordingly while re-applying r335197 (D48338).
I have never worked on Polly. The proposed change to param_div_div_div_2.ll is not educated, but just patterns that match the output.
All LLVM files are already reviewed in D48338.
Reviewers: jdoerfert, bollu, efriedma
Subscribers: jlebar, sanjoy, hiraditya, llvm-commits, bixia
Differential Revision: https://reviews.llvm.org/D48453
llvm-svn: 335292
Summary:
Also explicitly port over some tests in llvm.amdgcn.image.* that were
missing. Some tests are removed because they no longer apply (i.e.
explicitly testing building an address vector via insertelement).
This is in preparation for the eventual removal of the old-style
intrinsics.
Some additional notes:
- constant-address-space-32bit.ll: change some GCN-NEXT to GCN because
the instruction schedule was subtly altered
- insert_vector_elt.ll: the old test didn't actually test anything,
because %tmp1 was not used; remove the load, because it doesn't work
(Because of the amdgpu_ps calling convention? In any case, it's
orthogonal to what the test claims to be testing.)
Change-Id: Idfa99b6512ad139e755e82b8b89548ab08f0afcf
Reviewers: arsenm, rampitec
Subscribers: MatzeB, qcolombet, kzhuravl, wdng, yaxunl, dstuttard, tpr, t-tye, javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D48018
llvm-svn: 335229
This enables da-delinearize in Dependence Analysis for delinearizing array
accesses into multiple dimensions. This can help to increase the power of
Dependence analysis on multi-dimensional arrays and prevent having to fall
back to the slower and less accurate MIV tests. It adds static checks on the
bounds of the arrays to ensure that one dimension doesn't overflow into
another, and brings our code in line with our tests.
Differential Revision: https://reviews.llvm.org/D45872
llvm-svn: 335217
These were being over cautious for costs for one/two op general shuffles - VSHUFPD doesn't have to replicate the same shuffle in both lanes like VSHUFPS does.
llvm-svn: 335216
Summary:
Try to match udiv and urem patterns, and sink zext down to the leaves.
I'm not entirely sure why some unrelated tests change, but the added <nsw>s seem right.
Reviewers: sanjoy
Subscribers: jlebar, hiraditya, bixia, llvm-commits
Differential Revision: https://reviews.llvm.org/D48338
llvm-svn: 335197
Summary:
Related to https://bugs.llvm.org/show_bug.cgi?id=37793, https://reviews.llvm.org/D46760#1127287
We'd like to do this canonicalization https://rise4fun.com/Alive/Gmc
But it is currently restricted by rL155136 / rL155362, which says:
```
// This is a constant shift of a constant shift. Be careful about hiding
// shl instructions behind bit masks. They are used to represent multiplies
// by a constant, and it is important that simple arithmetic expressions
// are still recognizable by scalar evolution.
//
// The transforms applied to shl are very similar to the transforms applied
// to mul by constant. We can be more aggressive about optimizing right
// shifts.
//
// Combinations of right and left shifts will still be optimized in
// DAGCombine where scalar evolution no longer applies.
```
I think these tests show that for *constants*, SCEV has no issues with that canonicalization.
Reviewers: mkazantsev, spatel, efriedma, sanjoy
Reviewed By: mkazantsev
Subscribers: sanjoy, javed.absar, llvm-commits, stoklund, bixia
Differential Revision: https://reviews.llvm.org/D48229
llvm-svn: 335101
This reverts r334428. It incorrectly marks some multiplications as nuw. Tim
Shen is working on a proper fix.
Original commit message:
[SCEV] Add nuw/nsw to mul ops in StrengthenNoWrapFlags where safe.
Summary:
Previously we would add them for adds, but not multiplies.
llvm-svn: 335016
In particular, when asked to print a MemoryAccess, we'll now print where
defs are optimized to, and we'll print optimized access types.
This patch also introduces an operator<< to make printing AliasResults
easier.
Patch by Juneyoung Lee!
Differential Revision: https://reviews.llvm.org/D47860
llvm-svn: 334760
Summary:
Specifically, we transform
zext(2^K * (trunc X to iN)) to iM ->
2^K * (zext(trunc X to i{N-K}) to iM)<nuw>
This is helpful because pulling the 2^K out of the zext allows further
optimizations.
Reviewers: sanjoy
Subscribers: hiraditya, llvm-commits, timshen
Differential Revision: https://reviews.llvm.org/D48158
llvm-svn: 334737
Summary:
Previously we would do this simplification only if it did not introduce
any new truncs (excepting new truncs which replace other cast ops).
This change weakens this condition: If the number of truncs stays the
same, but we're able to transform trunc(X + Y) to X + trunc(Y), that's
still simpler, and it may open up additional transformations.
While we're here, also clean up some duplicated code.
Reviewers: sanjoy
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D48160
llvm-svn: 334736
IndVarSimplify sometimes makes transforms basing on users that are trivially dead. In particular,
if DCE wasn't run before it, there may be a dead `sext/zext` in loop that will trigger widening
transforms, however it makes no sense to do it.
This patch teaches IndVarsSimplify ignore the mist trivial cases of that.
Differential Revision: https://reviews.llvm.org/D47974
Reviewed By: sanjoy
llvm-svn: 334567
As discussed on PR33744, this patch relaxes ShuffleKind::SK_Alternate which requires shuffle masks to only match an alternating pattern from its 2 sources:
e.g. v4f32: <0,5,2,7> or <4,1,6,3>
This seems far too restrictive as most SIMD hardware which will implement it using a general blend/bit-select instruction, so replaces it with SK_Select, permitting elements from either source as long as they are inline:
e.g. v4f32: <0,5,2,7>, <4,1,6,3>, <0,1,6,7>, <4,1,2,3> etc.
This initial patch just updates the name and cost model shuffle mask analysis, later patch reviews will update SLP to better utilise this - it still limits itself to SK_Alternate style patterns.
Differential Revision: https://reviews.llvm.org/D47985
llvm-svn: 334513
As discussed on D47985, identity shuffle masks should probably be free.
I've limited this to the case where the input and output types all match - but we could probably accept all cases.
Differential Revision: https://reviews.llvm.org/D47986
llvm-svn: 334506
Summary:
Previously we would add them for adds, but not multiplies.
Reviewers: sanjoy
Subscribers: llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D48038
llvm-svn: 334428
An expression like
(zext i2 {(trunc i32 (1 + %B) to i2),+,1}<%while.body> to i32)
will become zero exactly when the nested value becomes zero in its type.
Strip injective operations from the input value in howFarToZero to make
the value simpler.
Differential Revision: https://reviews.llvm.org/D47951
llvm-svn: 334318
Currently the loop branch heuristic is applied before the invoke heuristic which makes us overestimate the probability of the unwind destination of invokes inside loops. This in turn makes us grossly underestimate the frequencies of loops with invokes.
Reviewed By: skatkov, vsk
Differential Revision: https://reviews.llvm.org/D47371
llvm-svn: 334285
Similar to v4i32 SHL, convert v8i16 shift amounts to scale factors instead to improve performance and reduce instruction count. We were already doing this for constant shifts, this adds variable shift support.
Reduces the serial nature of the codegen, which relies on chains of plendvb/pand+pandn+por shifts.
This is a step towards adding support for vXi16 vector rotates.
Differential Revision: https://reviews.llvm.org/D47546
llvm-svn: 334023
Summary:
Getelementptr returns a vector of pointers, instead of a single address,
when one or more of its arguments is a vector. In such case it is not
possible to simplify the expression by inserting a bitcast of operand(0)
into the destination type, as it will create a bitcast between different
sizes.
Reviewers: majnemer, mkuper, mssimpso, spatel
Reviewed By: spatel
Subscribers: lebedev.ri, llvm-commits
Differential Revision: https://reviews.llvm.org/D46379
llvm-svn: 333783
Both weakZeroSrcSIV and weakZeroDstSIV are currently giving the same
direction vectors. Fix weakZeroSrcSIVtest by flipping the directions
it gives.
Differential Revision: https://reviews.llvm.org/D46678
llvm-svn: 333658
Summary:
The atomic variants of the memcpy/memmove/memset intrinsics can be treated
the same was as the regular forms, with respect to aliasing. Update the
AliasSetTracker to treat the atomic forms the same was as the regular forms.
llvm-svn: 333551
Summary:
A simple change to derive mod/ref info from the atomic memcpy
intrinsic in the same way as from the regular memcpy intrinsic.
llvm-svn: 333454
Summary:
Patch for capture tracking broke
bootstrap of clang with -fstict-vtable-pointers
which resulted in debbugging nightmare. It was fixed
https://reviews.llvm.org/D46900 but as it turned
out, there were other parts like inliner (computing of
noalias metadata) that I found after bootstraping with enabled
assertions.
Reviewers: hfinkel, rsmith, chandlerc, amharc, kuhar
Subscribers: JDevlieghere, eraman, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D47088
llvm-svn: 333070
This enables us to detect more fast path sdiv cases under cost analysis.
This patch also enables us to handle non-uniform-constant pow2 cases for X86 SDIV costs.
Found while working on D46276
Future patches can then extend the vectorizers to more fully support non-uniform pow2 cases.
Differential Revision: https://reviews.llvm.org/D46637
llvm-svn: 332969
Summary:
invariant.group.launder should not stop propagation
of nonnull and dereferenceable, because e.g. we would not be
able to hoist loads speculatively.
Reviewers: rsmith, amharc, kuhar, xbolva00, hfinkel
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D46972
llvm-svn: 332788
Summary:
Memdep had funny bug related to invariant.groups - because it did not
invalidated cache, in some very rare cases it was possible to show memory
dependence of the instruction that was deleted, but because other
instruction took it's place it resulted in call to vtable!
Thanks @amharc for repro!.
Reviewers: dberlin, kuhar, amharc
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D45320
Co-authored-by: Krzysztof Pszeniczny <krzysztof.pszeniczny@gmail.com>
llvm-svn: 332781
CanProveNotTakenFirstIteration utility does not handle the case when
condition of the branch is a constant. Add its handling.
Reviewers: reames, anna, mkazantsev
Reviewed By: reames
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D46996
llvm-svn: 332695
Summary:
There was some unfinished work started for offset tracking in CFLGraph by the author of implementation of Andersen algorithm. This work was completed and support for field sensitivity was added to the core of Andersen algorithm.
The performance results seem promising.
SPEC2006 int_base score was increased by 1.1 % (I compared clang 6.0 with clang 6.0 with this patch). The avergae compile time was increased by +- 1 % according my measures with small and medium C/C++ projects (I did not tested it on the large projects with milions of lines of code)
Reviewers: chandlerc, george.burgess.iv, rja
Reviewed By: rja
Subscribers: rja, llvm-commits
Differential Revision: https://reviews.llvm.org/D46282
llvm-svn: 332657
Summary:
A recent patch ([[ https://reviews.llvm.org/rL331587 | rL331587 ]]) to Capture Tracking taught it that the `launder_invariant_group` intrinsic captures its argument only by returning it. Unfortunately, BasicAA still considered every call instruction as a possible escape source and hence concluded that the result of a `launder_invariant_group` call cannot alias any local non-escaping value. This led to [[ https://bugs.llvm.org/show_bug.cgi?id=37458 | bug 37458 ]].
This patch updates the relevant check for escape sources in BasicAA.
Reviewers: Prazek, kuhar, rsmith, hfinkel, sanjoy, xbolva00
Reviewed By: hfinkel, xbolva00
Subscribers: JDevlieghere, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D46900
llvm-svn: 332466
Summary:
After r332167 we started to sort the IDF blocks inside IDF calculation, so
there is no need to re-sort them on the user site. The test changes are due to
a slightly different order we're using now (originally we used DFSInNumber and
now the blocks are sorted by a pair (LevelFromRoot, DFSInNumber)).
Reviewers: dberlin, mgrang
Subscribers: Prazek, hiraditya, george.burgess.iv, llvm-commits
Differential Revision: https://reviews.llvm.org/D46899
llvm-svn: 332385
In general, it's difficult to poke the ConstantExpr code in CFLAA, since
LLVM is so great at eagerly reducing ConstantExprs. :)
Sadly, this only shows a functional difference from before the patch
because CFLAA has some special logic around taking loads of non-pointers
into account. Namely, with the broken select behavior, CFLAA will
completely fail to take note of @g3. Since CFLAA doesn't have any record
about @g3 when we do an alias query for @g3 and %a, it conservatively
answers MayAlias. When we properly take @g3 into account with the new
select logic, we get NoAlias for this query.
I suspect that the aforementioned "special logic" isn't completely
correct, but this test-case should prevent future wonky aliasing results
from appearing for these flavors of ConstantExprs, so I think it's still
worth having.
llvm-svn: 332017
With custom lowering for vector MULLH{S,U}, it is now profitable to
vectorize a divide by constant loop for the custom types (v16i8, v8i16,
and v4i32). The cost if based on TargetLowering::Build{S,U}DIV which
uses a multiply by constant plus adjustment to express a divide by
constant.
Both {u,s}mull{2} are expressed as Instruction::Mul and shifts by
Instruction::AShr.
llvm-svn: 331873
In order to set breakpoints on labels and list source code around
labels, we need collect debug information for labels, i.e., label
name, the function label belong, line number in the file, and the
address label located. In order to keep these information in LLVM
IR and to allow backend to generate debug information correctly.
We create a new kind of metadata for labels, DILabel. The format
of DILabel is
!DILabel(scope: !1, name: "foo", file: !2, line: 3)
We hope to keep debug information as much as possible even the
code is optimized. So, we create a new kind of intrinsic for label
metadata to avoid the metadata is eliminated with basic block.
The intrinsic will keep existing if we keep it from optimized out.
The format of the intrinsic is
llvm.dbg.label(metadata !1)
It has only one argument, that is the DILabel metadata. The
intrinsic will follow the label immediately. Backend could get the
label metadata through the intrinsic's parameter.
We also create DIBuilder API for labels to be used by Frontend.
Frontend could use createLabel() to allocate DILabel objects, and use
insertLabel() to insert llvm.dbg.label intrinsic in LLVM IR.
Differential Revision: https://reviews.llvm.org/D45024
Patch by Hsiangkai Wang.
llvm-svn: 331841
This patch was temporarily reverted because it has exposed bug 37229 on
PowerPC platform. The bug is unrelated to the patch and was just a general
bug in the optimization done for PowerPC platform only. The bug was fixed
by the patch rL331410.
This patch returns the disabled commit since the bug was fixed.
llvm-svn: 331427
It turned out that readonly argmemonly is not enough.
store 42, %p
%b = barrier(%p)
store 43, %b
the first store is dead, but because barrier was marked as
reading argument memory, it was considered alive. With
inaccessiblememonly it doesn't read the argument, but
it also can't be CSEd.
based on: https://reviews.llvm.org/D32006
llvm-svn: 331338
Summary:
The value tracking analysis uses function alignment to infer that the
least significant bits of function pointers are known to be zero.
Unfortunately, this is not correct for ARM targets: the least
significant bit of a function pointer stores the ARM/Thumb state
information (i.e., the LSB is set for Thumb functions and cleared for
ARM functions).
The original approach (https://reviews.llvm.org/D44781) introduced a
new field for function pointer alignment in the DataLayout structure
to address this. But it seems unlikely that optimizations based on
function pointer alignment would bring much benefit in practice to
justify the additional maintenance burden, so this patch simply
assumes that function pointer alignment is always unknown.
Reviewers: javed.absar, efriedma
Reviewed By: efriedma
Subscribers: kristof.beyls, llvm-commits, hfinkel, rogfer01
Differential Revision: https://reviews.llvm.org/D46110
llvm-svn: 331025
This patch adds a new shuffle kind useful for transposing a 2xn matrix. These
transpose shuffle masks read corresponding even- or odd-numbered vector
elements from two n-dimensional source vectors and write each result into
consecutive elements of an n-dimensional destination vector. The transpose
shuffle kind is meant to model the TRN1 and TRN2 AArch64 instructions. As such,
this patch also considers transpose shuffles in the AArch64 implementation of
getShuffleCost.
Differential Revision: https://reviews.llvm.org/D45982
llvm-svn: 330941
This reverts commit 023c8be90980e0180766196cba86f81608b35d38.
This patch triggers miscompile of zlib on PowerPC platform. Most likely it is
caused by some pre-backend PPC-specific pass, but we don't clearly know the
reason yet. So we temporally revert this patch with intention to return it
once the problem is resolved. See bug 37229 for details.
llvm-svn: 330893
This patch adds a cost model test case for vector shuffles having transpose
masks. The given costs are inaccurate and will be updated in a follow-on patch.
llvm-svn: 330625
Improve the alias analysis to account for cases where we
know that src/dst pairs cannot alias due to things like
TBAA. As we know they are noalias, we know no dependency
can occur. Also fixes issues around the size parameter
to AA being incorrect.
Differential Revision: https://reviews.llvm.org/D42381
llvm-svn: 329692
The script allows the auto-generation of checks for cost model tests to speed up their creation and help improve coverage, which will help a lot with PR36550.
If the need arises we can add support for other analyze passes as well, but the cost models was the one I needed to get done - at the moment it just warns that any other analysis mode is unsupported.
I've regenerated a couple of x86 test files to show the effect.
Differential Revision: https://reviews.llvm.org/D45272
llvm-svn: 329390
Summary:
These new image intrinsics contain the texture type as part of
their name and have each component of the address/coordinate as
individual parameters.
This is a preparatory step for implementing the A16 feature, where
coordinates are passed as half-floats or -ints, but the Z compare
value and texel offsets are still full dwords, making it difficult
or impossible to distinguish between A16 on or off in the old-style
intrinsics.
Additionally, these intrinsics pass the 'texfailpolicy' and
'cachectrl' as i32 bit fields to reduce operand clutter and allow
for future extensibility.
v2:
- gather4 supports 2darray images
- fix a bug with 1D images on SI
Change-Id: I099f309e0a394082a5901ea196c3967afb867f04
Reviewers: arsenm, rampitec, b-sumner
Subscribers: kzhuravl, wdng, yaxunl, dstuttard, tpr, llvm-commits, t-tye
Differential Revision: https://reviews.llvm.org/D44939
llvm-svn: 329166
Current implementation of `computeExitLimit` has a big piece of code
the only purpose of which is to prove that after the execution of this
block the latch will be executed. What it currently checks is actually a
subset of situations where the exiting block dominates latch.
This patch replaces all these checks for simple particular cases with
domination check over loop's latch which is the only necessary condition
of taking the exiting block into consideration. This change allows to
calculate exact loop taken count for simple loops like
for (int i = 0; i < 100; i++) {
if (cond) {...} else {...}
if (i > 50) break;
. . .
}
Differential Revision: https://reviews.llvm.org/D44677
Reviewed By: efriedma
llvm-svn: 329047
We use a `DenseMap<MemoryLocOrCall, MemlocStackInfo>` to keep track of
prior work when optimizing uses in MemorySSA. Because we weren't
accounting for callsite arguments in either the hash code or equality
tests for `MemoryLocOrCall`s, we optimized uses too aggressively in
some rare cases.
Fix by Daniel Berlin.
Should fix PR36883.
llvm-svn: 328748
Currently, `getExact` fails if it sees two exit counts in different blocks. There is
no solid reason to do so, given that we only calculate exact non-taken count
for exiting blocks that dominate latch. Using this fact, we can simply take min
out of all exits of all blocks to get the exact taken count.
This patch makes the calculation more optimistic with enforcing our assumption
with asserts. It allows us to calculate exact backedge taken count in trivial loops
like
for (int i = 0; i < 100; i++) {
if (i > 50) break;
. . .
}
Differential Revision: https://reviews.llvm.org/D44676
Reviewed By: fhahn
llvm-svn: 328611
Add fdiv costs for Goldmont using table 16-17 of the Intel Optimization Manual. Also add overrides for FSQRT for Goldmont and Silvermont.
Reviewers: RKSimon
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D44644
llvm-svn: 328451
Most basic possible test for the logic used by LICM.
Also contains a speculative build fix for compiles which complain about a definition of a stuct K; followed by a declaration as class K;
llvm-svn: 328058
As suggested in the original review (https://reviews.llvm.org/D44524), use an annotation style printer instead.
Note: The switch from -analyze to -disable-output in tests was driven by the fact that seems to be the idiomatic style used in annoation passes. I tried to keep both working, but the old style pass API for printers really doesn't make this easy. It invokes (runOnFunction, print(Module)) repeatedly. I decided the extra state wasn't worth it given the old pass manager is going away soonish anyway.
llvm-svn: 328015
Many of our loop passes make use of so called "must execute" or "guaranteed to execute" facts to prove the legality of code motion. The basic notion is that we know (by assumption) an instruction didn't fault at it's original location, so if the location we move it to is strictly post dominated by the original, then we can't have introduced a new fault.
At the moment, the testing for this logic is somewhat adhoc and done mostly through LICM. Since I'm working on that code, I want to improve the testing. This patch is the first step in that direction. It doesn't actually test the variant used by the loop passes - I need to move that to the Analysis library first - but instead exercises an alternate implementation used by SCEV. (I plan on merging both implementations.)
Note: I'll be replacing the printing logic within this with an annotation based version in the near future. Anna suggested this in review, and it seems like a strictly better format.
Differential Revision: https://reviews.llvm.org/D44524
llvm-svn: 328004
This is re-land of https://reviews.llvm.org/rL327362 with a fix
and regression test.
The crash was due to it is possible that for found MDL loop,
LHS or RHS may contain an invariant unknown SCEV which
does not dominate the MDL. Please see regression
test for an example.
Reviewers: sanjoy, mkazantsev, reames
Reviewed By: mkazantsev
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D44553
llvm-svn: 327822
This patch provides an implementation of getArithmeticReductionCost for
AArch64. We can specialize the cost of add reductions since they are computed
using the 'addv' instruction.
Differential Revision: https://reviews.llvm.org/D44490
llvm-svn: 327702
This patch considers the experimental vector reduce intrinsics in the default
implementation of getIntrinsicInstrCost. The cost of these intrinsics is
computed with getArithmeticReductionCost and getMinMaxReductionCost. This patch
also adds a test case for AArch64 that indicates the costs we currently compute
for vector reduce intrinsics. These costs are inaccurate and will be updated in
a follow-on patch.
Differential Revision: https://reviews.llvm.org/D44489
llvm-svn: 327698
Since there is no instruction for integer vector division, factor in the
cost of singling out each element to be used with the scalar division
instruction.
Differential revision: https://reviews.llvm.org/D43974
llvm-svn: 326955
It's been quite some time the Dependence Analysis (DA) is broken,
as it uses the GEP representation to "identify" multi-dimensional arrays.
It even wrongly detects multi-dimensional arrays in single nested loops:
from test/Analysis/DependenceAnalysis/Coupled.ll, example @couple6
;; for (long int i = 0; i < 50; i++) {
;; A[i][3*i - 6] = i;
;; *B++ = A[i][i];
DA used to detect two subscripts, which makes no sense in the LLVM IR
or in C/C++ semantics, as there are no guarantees as in Fortran of
subscripts not overlapping into a next array dimension:
maximum nesting levels = 1
SrcPtrSCEV = %A
DstPtrSCEV = %A
using GEPs
subscript 0
src = {0,+,1}<nuw><nsw><%for.body>
dst = {0,+,1}<nuw><nsw><%for.body>
class = 1
loops = {1}
subscript 1
src = {-6,+,3}<nsw><%for.body>
dst = {0,+,1}<nuw><nsw><%for.body>
class = 1
loops = {1}
Separable = {}
Coupled = {1}
With the current patch, DA will correctly work on only one dimension:
maximum nesting levels = 1
SrcSCEV = {(-2424 + %A)<nsw>,+,1212}<%for.body>
DstSCEV = {%A,+,404}<%for.body>
subscript 0
src = {(-2424 + %A)<nsw>,+,1212}<%for.body>
dst = {%A,+,404}<%for.body>
class = 1
loops = {1}
Separable = {0}
Coupled = {}
This change removes all uses of GEP from DA, and we now only rely
on the SCEV representation.
The patch does not turn on -da-delinearize by default, and so the DA analysis
will be more conservative in the case of multi-dimensional memory accesses in
nested loops.
I disabled some interchange tests, as the DA is not able to disambiguate
the dependence anymore. To make DA stronger, we may need to
compute a bound on the number of iterations based on the access functions
and array dimensions.
The patch cleans up all the CHECKs in test/Transforms/LoopInterchange/*.ll to
avoid checking for snippets of LLVM IR: this form of checking is very hard to
maintain. Instead, we now check for output of the pass that are more meaningful
than dozens of lines of LLVM IR. Some tests now require -debug messages and thus
only enabled with asserts.
Patch written by Sebastian Pop and Aditya Kumar.
Differential Revision: https://reviews.llvm.org/D35430
llvm-svn: 326837
The range of SCEVUnknown Phi which merges values `X1, X2, ..., XN`
can be evaluated as `U(Range(X1), Range(X2), ..., Range(XN))`.
Reviewed By: sanjoy
Differential Revision: https://reviews.llvm.org/D43810
llvm-svn: 326418
Agner's tables indicate that for SSE42+ targets (Core2 and later) we can reduce the FADD/FSUB/FMUL costs down to 1, which should fix the Himeno benchmark.
Note: the AVX512 FDIV costs look rather dodgy, but this isn't part of this patch.
Differential Revision: https://reviews.llvm.org/D43733
llvm-svn: 326133
If we have a loop like this:
int n = 0;
while (...) {
if (++n >= MAX) {
n = 0;
}
}
then the body of the 'if' statement will only be executed once every MAX
iterations. Detect this by looking for branches in loops where taking the branch
makes the branch condition evaluate to 'not taken' in the next iteration of the
loop, and reduce the probability of such branches.
This slightly improves EEMBC benchmarks on cortex-m4/cortex-m33 due to making
better choices in if-conversion, but has no effect on any other cpu/benchmark
that I could detect.
Differential Revision: https://reviews.llvm.org/D35804
llvm-svn: 325925
There are too many perf regressions resulting from this, so we need to
investigate (and add tests for) targets like ARM and AArch64 before
trying to reinstate.
llvm-svn: 325658
This change was mentioned at least as far back as:
https://bugs.llvm.org/show_bug.cgi?id=26837#c26
...and I found a real program that is harmed by this:
Himeno running on AMD Jaguar gets 6% slower with SLP vectorization:
https://bugs.llvm.org/show_bug.cgi?id=36280
...but the change here appears to solve that bug only accidentally.
The div/rem costs for x86 look very wrong in some cases, but that's already true,
so we can fix those in follow-up patches. There's also evidence that more cost model
changes are needed to solve SLP problems as shown in D42981, but that's an independent
problem (though the solution may be adjusted after this change is made).
Differential Revision: https://reviews.llvm.org/D43079
llvm-svn: 325515
The current implementation of `getPostIncExpr` invokes `getAddExpr` for two recurrencies
and expects that it always returns it a recurrency. But this is not guaranteed to happen if we
have reached max recursion depth or refused to make SCEV simplification for other reasons.
This patch changes its implementation so that now it always returns SCEVAddRec without
relying on `getAddExpr`.
Differential Revision: https://reviews.llvm.org/D42953
llvm-svn: 324866
This patch implements analysis for new-format TBAA access tags
with aggregate types as their final access types.
Differential Revision: https://reviews.llvm.org/D41501
llvm-svn: 324092
Summary:
This change is part of step five in the series of changes to remove alignment argument from
memcpy/memmove/memset in favour of alignment attributes. In particular, this changes the Lint
analysis to cease using the old getAlignment() API of MemoryIntrinsic in favour of getting
source & dest specific alignments through the new API.
Steps:
Step 1) Remove alignment parameter and create alignment parameter attributes for
memcpy/memmove/memset. ( rL322965, rC322964, rL322963 )
Step 2) Expand the IRBuilder API to allow creation of memcpy/memmove with differing
source and dest alignments. ( rL323597 )
Step 3) Update Clang to use the new IRBuilder API. ( rC323617 )
Step 4) Update Polly to use the new IRBuilder API. ( rL323618 )
Step 5) Update LLVM passes that create memcpy/memmove calls to use the new IRBuilder API,
and those that use use MemIntrinsicInst::[get|set]Alignment() to use [get|set]DestAlignment()
and [get|set]SourceAlignment() instead.
Step 6) Remove the single-alignment IRBuilder API for memcpy/memmove, and the
MemIntrinsicInst::[get|set]Alignment() methods.
Reference
http://lists.llvm.org/pipermail/llvm-dev/2015-August/089384.htmlhttp://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20151109/312083.html
llvm-svn: 323886
We're getting bug reports:
https://bugs.llvm.org/show_bug.cgi?id=35807https://bugs.llvm.org/show_bug.cgi?id=35840https://bugs.llvm.org/show_bug.cgi?id=36045
...where we blow up the stack in value tracking because other passes are sending
in selects that have an operand that is itself the select.
We don't currently have a reliable way to avoid analyzing dead code that may take
non-standard forms, so bail out when things go too far.
This mimics the recursion depth limitations in other parts of value tracking.
Unfortunately, this pushes the underlying problems for other passes (jump-threading,
simplifycfg, correlated-propagation) into hiding. If someone wants to uncover those
again, the first draft of this patch on Phab would do that (it would assert rather
than bail out).
Differential Revision: https://reviews.llvm.org/D42442
llvm-svn: 323331
An alternative (and probably better) fix would be that of
making `Scale` an APInt, and there's a patch floating around
to do this. As we're still discussing it, at least stop crashing
in the meanwhile (added bonus, we now have a regression test for
this situation).
Fixes PR35843.
Thanks to Eli for suggesting the fix and Simon for reporting and
reducing the bug.
llvm-svn: 322467
Summary:
See D37528 for a previous (non-deferred) version of this
patch and its description.
Preserves dominance in a deferred manner using a new class
DeferredDominance. This reduces the performance impact of
updating the DominatorTree at every edge insertion and
deletion. A user may call DDT->flush() within JumpThreading
for an up-to-date DT. This patch currently has one flush()
at the end of runImpl() to ensure DT is preserved across
the pass.
LVI is also preserved to help subsequent passes such as
CorrelatedValuePropagation. LVI is simpler to maintain and
is done immediately (not deferred). The code to perform the
preversation was minimally altered and simply marked as
preserved for the PassManager to be informed.
This extends the analysis available to JumpThreading for
future enhancements such as threading across loop headers.
Reviewers: dberlin, kuhar, sebpop
Reviewed By: kuhar, sebpop
Subscribers: mgorny, dmgreen, kuba, rnk, rsmith, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D40146
llvm-svn: 322401
Summary:
There are few oddities that occur due to v1i1, v8i1, v16i1 being legal without v2i1 and v4i1 being legal when we don't have VLX. Particularly during legalization of v2i32/v4i32/v2i64/v4i64 masked gather/scatter/load/store. We end up promoting the mask argument to these during type legalization and then have to widen the promoted type to v8iX/v16iX and truncate it to get the element size back down to v8i1/v16i1 to use a 512-bit operation. Since need to fill the upper bits of the mask we have to fill with 0s at the promoted type.
It would be better if we could just have the v2i1/v4i1 types as legal so they don't undergo any promotion. Then we can just widen with 0s directly in a k register. There are no real v4i1/v2i1 instructions anyway. Everything is done on a larger register anyway.
This also fixes an issue that we couldn't implement a masked vextractf32x4 from zmm to xmm properly.
We now have to support widening more compares to 512-bit to get a mask result out so new tablegen patterns got added.
I had to hack the legalizer for widening the operand of a setcc a bit so it didn't try create a setcc returning v4i32, extract from it, then try to promote it using a sign extend to v2i1. Now we create the setcc with v4i1 if the original setcc's result type is v2i1. Then extract that and don't sign extend it at all.
There's definitely room for improvement with some follow up patches.
Reviewers: RKSimon, zvi, guyblank
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D41560
llvm-svn: 321967
Summary:
See D37528 for a previous (non-deferred) version of this
patch and its description.
Preserves dominance in a deferred manner using a new class
DeferredDominance. This reduces the performance impact of
updating the DominatorTree at every edge insertion and
deletion. A user may call DDT->flush() within JumpThreading
for an up-to-date DT. This patch currently has one flush()
at the end of runImpl() to ensure DT is preserved across
the pass.
LVI is also preserved to help subsequent passes such as
CorrelatedValuePropagation. LVI is simpler to maintain and
is done immediately (not deferred). The code to perfom the
preversation was minimally altered and was simply marked
as preserved for the PassManager to be informed.
This extends the analysis available to JumpThreading for
future enhancements. One example is loop boundary threading.
Reviewers: dberlin, kuhar, sebpop
Reviewed By: kuhar, sebpop
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D40146
llvm-svn: 321825
Summary:
When using byval, the data is effectively copied as part of the call
anyway, so we aren't actually passing the pointer and thus there is no
reason to issue a warning.
Reviewers: rnk
Reviewed By: rnk
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D40118
llvm-svn: 321478
This is fix for the crash caused by ScalarEvolution::getTruncateExpr.
It expects that if it checked the condition that SCEV is not in UniqueSCEVs cache in
the beginning that it will not be there inside this method.
However during recursion and transformation/simplification for sub expression,
it is possible that these modifications will end up with the same SCEV as we started from.
So we must always check whether SCEV is in cache and do not insert item if it is already there.
Reviewers: sanjoy, mkazantsev, craig.topper
Reviewed By: sanjoy
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D41380
llvm-svn: 321472
Summary:
Make MemorySSA allow reordering of two loads that may alias, when one is volatile.
This makes MemorySSA less conservative and behaving the same as the AliasSetTracker.
For more context, see D16875.
LLVM language reference: "The optimizers must not change the number of volatile operations or change their order of execution relative to other volatile operations. The optimizers may change the order of volatile operations relative to non-volatile operations. This is not Java’s “volatile” and has no cross-thread synchronization behavior."
Reviewers: george.burgess.iv, dberlin
Subscribers: sanjoy, reames, hfinkel, llvm-commits, Prazek
Differential Revision: https://reviews.llvm.org/D41525
llvm-svn: 321382
Summary:
Add an additional bit to ModRefInfo, ModRefInfo::Must, to be cleared for known must aliases.
Shift existing Mod/Ref/ModRef values to include an additional most
significant bit. Update wrappers that modify ModRefInfo values to
reflect the change.
Notes:
* ModRefInfo::Must is almost entirely cleared in the AAResults methods, the remaining changes are trying to preserve it.
* Only some small changes to make custom AA passes set ModRefInfo::Must (BasicAA).
* GlobalsModRef already declares a bit, who's meaning overlaps with the most significant bit in ModRefInfo (MayReadAnyGlobal). No changes to shift the value of MayReadAnyGlobal (see AlignedMap). FunctionInfo.getModRef() ajusts most significant bit so correctness is preserved, but the Must info is lost.
* There are cases where the ModRefInfo::Must is not set, e.g. 2 calls that only read will return ModRefInfo::NoModRef, though they may read from exactly the same location.
Reviewers: dberlin, hfinkel, george.burgess.iv
Subscribers: llvm-commits, sanjoy
Differential Revision: https://reviews.llvm.org/D38862
llvm-svn: 321309
Summary:
As suggested by Eli Friedman, don't try to handle array allocas here,
because of possible overflows, instead rely on instcombine converting
them to allocations of array types.
Reviewers: efriedma
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D41398
llvm-svn: 321159
Enhance LVI to analyze the ‘ashr’ binary operation. This leverages the infrastructure in ConstantRange for the ashr operation.
Patch by Surya Kumari Jangala!
Differential Revision: https://reviews.llvm.org/D40886
llvm-svn: 320983
Based on the names of the check lines, features seems more appropriate that cpu.
Spotted while prototyping my patch to make 512-bit vectors illegal on SKX sometimes.
llvm-svn: 320959
Summary:
For byval arguments, the number of dereferenceable bytes is equal to
the size of the pointee, not the pointer.
Reviewers: hfinkel, rnk
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D41305
llvm-svn: 320939
Summary:
See D37528 for a previous (non-deferred) version of this
patch and its description.
Preserves dominance in a deferred manner using a new class
DeferredDominance. This reduces the performance impact of
updating the DominatorTree at every edge insertion and
deletion. A user may call DDT->flush() within JumpThreading
for an up-to-date DT. This patch currently has one flush()
at the end of runImpl() to ensure DT is preserved across
the pass.
LVI is also preserved to help subsequent passes such as
CorrelatedValuePropagation. LVI is simpler to maintain and
is done immediately (not deferred). The code to perfom the
preversation was minimally altered and was simply marked
as preserved for the PassManager to be informed.
This extends the analysis available to JumpThreading for
future enhancements. One example is loop boundary threading.
Reviewers: dberlin, kuhar, sebpop
Reviewed By: kuhar, sebpop
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D40146
llvm-svn: 320612
This reverts part of r300656, which caused a regression in
propagateMassToSuccessors by counting edges n^2 times, where n is the
number of edges from the source basic block to the same successor basic
block. The result was both incorrect and very slow to compute for large
values of n (e.g. switches with multiple cases that go to the same basic
block).
Patch by Andrew Scheidecker!
llvm-svn: 320208
In this method, we invoke `SimplifyICmpOperands` which takes the `Cond` predicate
by reference and may change it along with `LHS` and `RHS` SCEVs. But then we invoke
`computeShiftCompareExitLimit` with Values from which the SCEVs have been derived,
these Values have not been modified while `Cond` could be.
One of possible outcomes of this is that we may falsely prove that an infinite loop ends
within some finite number of iterations.
In this patch, we save the original `Cond` and pass it along with original operands.
This logic may be removed in future once `computeShiftCompareExitLimit` works
with SCEVs instead of value operands.
Reviewed By: sanjoy
Differential Revision: https://reviews.llvm.org/D40953
llvm-svn: 320142
Given loops `L1` and `L2` with AddRecs `AR1` and `AR2` varying in them respectively.
When identifying loop disposition of `AR2` w.r.t. `L1`, we only say that it is varying if
`L1` contains `L2`. But there is also a possible situation where `L1` and `L2` are
consecutive sibling loops within the parent loop. In this case, `AR2` is also varying
w.r.t. `L1`, but we don't correctly identify it.
It can lead, for exaple, to attempt of incorrect folding. Consider:
AR1 = {a,+,b}<L1>
AR2 = {c,+,d}<L2>
EXAR2 = sext(AR1)
MUL = mul AR1, EXAR2
If we incorrectly assume that `EXAR2` is invariant w.r.t. `L1`, we can end up trying to
construct something like: `{a * {c,+,d}<L2>,+,b * {c,+,d}<L2>}<L1>`, which is incorrect
because `AR2` is not available on entrance of `L1`.
Both situations "`L1` contains `L2`" and "`L1` preceeds sibling loop `L2`" can be handled
with one check: "header of `L1` dominates header of `L2`". This patch replaces the old
insufficient check with this one.
Differential Revision: https://reviews.llvm.org/D39453
llvm-svn: 318819
Summary:
Add the following heuristics for irreducible loop metadata:
- When an irreducible loop header is missing the loop header weight metadata,
give it the minimum weight seen among other headers.
- Annotate indirectbr targets with the loop header weight metadata (as they are
likely to become irreducible loop headers after indirectbr tail duplication.)
These greatly improve the accuracy of the block frequency info of the Python
interpreter loop (eg. from ~3-16x off down to ~40-55% off) and the Python
performance (eg. unpack_sequence from ~50% slower to ~8% faster than GCC) due to
better register allocation under PGO.
Reviewers: davidxl
Reviewed By: davidxl
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D39980
llvm-svn: 318693
llvm.invariant.group.barrier may accept pointers to arbitrary address space.
This patch let it accept pointers to i8 in any address space and returns
pointer to i8 in the same address space.
Differential Revision: https://reviews.llvm.org/D39973
llvm-svn: 318413
This patch contains more accurate cost of interelaved load\store of stride 2 for the types int64\double on AVX2.
Reviewers: delena, RKSimon, craig.topper, dorit
Reviewed By: dorit
Differential Revision: https://reviews.llvm.org/D40008
llvm-svn: 318385
Summary:
This fixes PR35241.
When using byval, the data is effectively copied as part of the call
anyway, so the pointer returned by the alloca will not be leaked to the
callee and thus there is no reason to issue a warning.
Reviewers: rnk
Reviewed By: rnk
Subscribers: Ka-Ka, llvm-commits
Differential Revision: https://reviews.llvm.org/D40009
llvm-svn: 318279
Summary:
If a compare instruction is same or inverse of the compare in the
branch of the loop latch, then return a constant evolution node.
This shall facilitate computations of loop exit counts in cases
where compare appears in the evolution chain of induction variables.
Will fix PR 34538
Reviewers: sanjoy, hfinkel, junryoungju
Reviewed By: sanjoy, junryoungju
Subscribers: javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D38494
llvm-svn: 318050
Recommit:
This patch contains update of the costs of interleaved loads of v8f32 of stride 3 and 8.
fixed the location of the lit test it works with make check-all.
Differential Revision: https://reviews.llvm.org/D39403
llvm-svn: 317471
reverted my changes will be committed later after fixing the failure
This patch contains update of the costs of interleaved loads of v8f32 of stride 3 and 8.
Differential Revision: https://reviews.llvm.org/D39403
llvm-svn: 317433
This patch contains update of the costs of interleaved loads of v8f32 of stride 3 and 8.
Differential Revision: https://reviews.llvm.org/D39403
llvm-svn: 317432
Summary:
Currently the block frequency analysis is an approximation for irreducible
loops.
The new irreducible loop metadata is used to annotate the irreducible loop
headers with their header weights based on the PGO profile (currently this is
approximated to be evenly weighted) and to help improve the accuracy of the
block frequency analysis for irreducible loops.
This patch is a basic support for this.
Reviewers: davidxl
Reviewed By: davidxl
Subscribers: mehdi_amini, llvm-commits, eraman
Differential Revision: https://reviews.llvm.org/D39028
llvm-svn: 317278