Shifts with a uniform but non-constant count were considered very expensive to
vectorize, because the splat of the uniform count and the shift would tend to
appear in different blocks. That made the splat invisible to ISel, and we'd
scalarize the shift at codegen time.
Since r201655, CodeGenPrepare sinks those splats to be next to their use, and we
are able to select the appropriate vector shifts. This updates the cost model to
to take this into account by making shifts by a uniform cheap again.
Differential Revision: https://reviews.llvm.org/D23049
llvm-svn: 277782
Summary:
TargetBaseAlign is no longer required since LSV checks if target allows misaligned accesses.
A constant defining a base alignment is still needed for stack accesses where alignment can be adjusted.
Previous patch (D22936) was reverted because tests were failing. This patch also fixes the cause of those failures:
- x86 failing tests either did not have the right target, or the right alignment.
- NVPTX failing tests did not have the right alignment.
- AMDGPU failing test (merge-stores) should allow vectorization with the given alignment but the target info
considers <3xi32> a non-standard type and gives up early. This patch removes the condition and only checks
for a maximum size allowed and relies on the next condition checking for %4 for correctness.
This should be revisited to include 3xi32 as a MVT type (on arsenm's non-immediate todo list).
Note that checking the sizeInBits for a MVT is undefined (leads to an assertion failure),
so we need to create an EVT, hence the interface change in allowsMisaligned to include the Context.
Reviewers: arsenm, jlebar, tstellarAMD
Subscribers: jholewinski, arsenm, mzolotukhin, llvm-commits
Differential Revision: https://reviews.llvm.org/D23068
llvm-svn: 277735
Update comment for isOutOfScope and add a testcase for uniform value being used
out of scope.
Differential Revision: https://reviews.llvm.org/D23073
llvm-svn: 277515
This patch enables the vectorizer to generate both scalar and vector versions
of an integer induction variable for a given loop. Previously, we only
generated a scalar induction variable if we knew all its users were going to be
scalar. Otherwise, we generated a vector induction variable. In the case of a
loop with both scalar and vector users of the induction variable, we would
generate the vector induction variable and extract scalar values from it for
the scalar users. With this patch, we now generate both versions of the
induction variable when there are both scalar and vector users and select which
version to use based on whether the user is scalar or vector.
Differential Revision: https://reviews.llvm.org/D22869
llvm-svn: 277474
This patch refactors the logic in collectLoopUniforms and
collectValuesToIgnore, untangling the concepts of "uniform" and "scalar". It
adds isScalarAfterVectorization along side isUniformAfterVectorization to
distinguish the two. Known scalar values include those that are uniform,
getelementptr instructions that won't be vectorized, and induction variables
and induction variable update instructions whose users are all known to be
scalar.
This patch includes the following functional changes:
- In collectLoopUniforms, we mark uniform the pointer operands of interleaved
accesses. Although non-consecutive, these pointers are treated like
consecutive pointers during vectorization.
- In collectValuesToIgnore, we insert a value into VecValuesToIgnore if it
isScalarAfterVectorization rather than isUniformAfterVectorization. This
differs from the previous functionaly in that we now add getelementptr
instructions that will not be vectorized into VecValuesToIgnore.
This patch also removes the ValuesNotWidened set used for induction variable
scalarization since, after the above changes, it is now equivalent to
isScalarAfterVectorization.
Differential Revision: https://reviews.llvm.org/D22867
llvm-svn: 277460
Summary:
TargetBaseAlign is no longer required since LSV checks if target allows misaligned accesses.
A constant defining a base alignment is still needed for stack accesses where alignment can be adjusted.
Reviewers: llvm-commits, jlebar
Subscribers: mzolotukhin, arsenm
Differential Revision: https://reviews.llvm.org/D22936
llvm-svn: 277038
Summary:
Given the crash in D22878, this patch converts the load/store vectorizer
to use explicit Instruction*s wherever possible. This is an overall
simplification and should be an improvement in safety, as we have fewer
naked cast<>s, and now where we use Value*, we really mean something
different from Instruction*.
This patch also gets rid of some cast<>s around Value*s returned by
Builder. Given that Builder constant-folds everything, we can't assume
much about what we get out of it.
One downside of this patch is that we have to copy our chain before
calling propagateMetadata. But I don't think this is a big deal, as our
chains are very small (usually 2 or 4 elems).
Reviewers: asbirlea
Subscribers: mzolotukhin, llvm-commits, arsenm
Differential Revision: https://reviews.llvm.org/D22887
llvm-svn: 276938
Summary:
When we ask the builder to create a bitcast on a constant, we get back a
constant, not an instruction.
Reviewers: asbirlea
Subscribers: jholewinski, mzolotukhin, llvm-commits, arsenm
Differential Revision: https://reviews.llvm.org/D22878
llvm-svn: 276922
Allowed loop vectorization with secondary FP IVs. Like this:
float *A;
float x = init;
for (int i=0; i < N; ++i) {
A[i] = x;
x -= fp_inc;
}
The auto-vectorization is possible when the induction binary operator is "fast" or the function has "unsafe" attribute.
Differential Revision: https://reviews.llvm.org/D21330
llvm-svn: 276554
When vectorizing a tree rooted at a store bundle, we currently try to sort the
stores before building the tree, so that the stores can be vectorized. For other
trees, the order of the root bundle - which determines the order of all other
bundles - is arbitrary. That is bad, since if a leaf bundle of consecutive loads
happens to appear in the wrong order, we will not vectorize it.
This is partially mitigated when the root is a binary operator, by trying to
build a "reversed" tree when that's considered profitable. This patch extends the
workaround we have for binops to trees rooted in a horizontal reduction.
This fixes PR28474.
Differential Revision: https://reviews.llvm.org/D22554
llvm-svn: 276477
This patch moves the update instruction for vectorized integer induction phi
nodes to the end of the latch block. This ensures consistent placement of all
induction updates across all the kinds of int inductions we create (scalar,
splat vector, or vector phi).
Differential Revision: https://reviews.llvm.org/D22416
llvm-svn: 276339
The earlier change added hotness attribute to missed-optimization
remarks. This follows up with the analysis remarks (the ones explaining
the reason for the missed optimization).
llvm-svn: 276192
Summary:
Previously we wouldn't move loads/stores across instructions that had
side-effects, where that was defined as may-write or may-throw. But
this is not sufficiently restrictive: Stores can't safely be moved
across instructions that may load.
This patch also adds a DEBUG check that all instructions in our chain
are either loads or stores.
Reviewers: asbirlea
Subscribers: llvm-commits, jholewinski, arsenm, mzolotukhin
Differential Revision: https://reviews.llvm.org/D22547
llvm-svn: 276171
Summary:
Previously if we had a chain that contained a side-effecting
instruction, we wouldn't vectorize it at all. Now we'll vectorize
everything that comes before the side-effecting instruction.
Reviewers: asbirlea
Subscribers: arsenm, jholewinski, llvm-commits, mzolotukhin
Differential Revision: https://reviews.llvm.org/D22536
llvm-svn: 276170
Summary:
getVectorizablePrefix previously didn't work properly in the face of
aliasing loads/stores. It unwittingly assumed that the loads/stores
appeared in the BB in address order. If they didn't, it would do the
wrong thing.
Reviewers: asbirlea, tstellarAMD
Subscribers: arsenm, llvm-commits, mzolotukhin
Differential Revision: https://reviews.llvm.org/D22535
llvm-svn: 276072
Summary:
Previously, the insertion point for stores was the last instruction in
Chain *before calling getVectorizablePrefixEndIdx*. Thus if
getVectorizablePrefixEndIdx didn't return Chain.size(), we still would
insert at the last instruction in Chain.
This patch changes our internal API a bit in an attempt to make it less
prone to this sort of error. As a result, we end up recalculating the
Chain's boundary instructions, but I think worrying about the speed hit
of this is a premature optimization right now.
Reviewers: asbirlea, tstellarAMD
Subscribers: mzolotukhin, arsenm, llvm-commits
Differential Revision: https://reviews.llvm.org/D22534
llvm-svn: 276056
Summary:
The DEBUG message was hard to read because two Values were being printed
on the same line with only the delimiter "aliases". This change makes
us print each Value on its own line.
Reviewers: asbirlea
Subscribers: llvm-commits, arsenm, mzolotukhin
Differential Revision: https://reviews.llvm.org/D22533
llvm-svn: 276055
For instructions in uniform set, they will not have vector versions so
add them to VecValuesToIgnore.
For induction vars, those only used in uniform instructions or consecutive
ptrs instructions have already been added to VecValuesToIgnore above. For
those induction vars which are only used in uniform instructions or
non-consecutive/non-gather scatter ptr instructions, the related phi and
update will also be added into VecValuesToIgnore set.
The change will make the vector RegUsages estimation less conservative.
Differential Revision: https://reviews.llvm.org/D20474
The recommit fixed the testcase global_alias.ll.
llvm-svn: 275936
For instructions in uniform set, they will not have vector versions so
add them to VecValuesToIgnore.
For induction vars, those only used in uniform instructions or consecutive
ptrs instructions have already been added to VecValuesToIgnore above. For
those induction vars which are only used in uniform instructions or
non-consecutive/non-gather scatter ptr instructions, the related phi and
update will also be added into VecValuesToIgnore set.
The change will make the vector RegUsages estimation less conservative.
Differential Revision: https://reviews.llvm.org/D20474
llvm-svn: 275912
This patch swaps A and B in the interleaved access analysis and clarifies
related comments. The algorithm is more intuitive if we let access A precede
access B in program order rather than the reverse. This change was requested in
the review of D19984.
llvm-svn: 275567
We now collect all accesses with a constant stride, not just the ones with a
stride greater than one. This change was requested in the review of D19984.
llvm-svn: 275473
This patch allows the formation of interleaved access groups in loops
containing predicated blocks. However, the predicated accesses are prevented
from forming groups.
Differential Revision: https://reviews.llvm.org/D19694
llvm-svn: 275471
This patch prevents increases in the number of instructions, pre-instcombine,
due to induction variable scalarization. An increase in instructions can lead
to an increase in the compile-time required to simplify the induction
variables. We now maintain a new map for scalarized induction variables to
prevent us from converting between the scalar and vector forms.
This patch should resolve compile-time regressions seen after r274627.
llvm-svn: 275419
Summary:
LSV used to abort vectorizing a chain for interleaved load/store accesses that alias.
Allow a valid prefix of the chain to be vectorized, mark just the prefix and retry vectorizing the remaining chain.
Reviewers: llvm-commits, jlebar, arsenm
Subscribers: mzolotukhin
Differential Revision: http://reviews.llvm.org/D22119
llvm-svn: 275317
The LCSSA pass itself will not generate several redundant PHI nodes in a single
exit block. However, such redundant PHI nodes don't violate LCSSA form, and may
be introduced by passes that preserve LCSSA, and/or preserved by the LCSSA pass
itself. So, assuming a single PHI node per exit block is not safe.
llvm-svn: 275217
Summary:
Aiming to correct the ordering of loads/stores. This patch changes the
insert point for loads to the position of the first load.
It updates the ordering method for loads to insert before, rather than after.
Before this patch the following sequence:
"load a[1], store a[1], store a[0], load a[2]"
Would incorrectly vectorize to "store a[0,1], load a[1,2]".
The correctness check was assuming the insertion point for loads is at
the position of the first load, when in practice it was at the last
load. An alternative fix would have been to invert the correctness check.
The current fix changes insert position but also requires reordering of
instructions before the vectorized load.
Updated testcases to reflect the changes.
Reviewers: tstellarAMD, llvm-commits, jlebar, arsenm
Subscribers: mzolotukhin
Differential Revision: http://reviews.llvm.org/D22071
llvm-svn: 275117
We currently always vectorize induction variables. However, if an induction
variable is only used for counting loop iterations or computing addresses with
getelementptr instructions, we don't need to do this. Vectorizing these trivial
induction variables can create vector code that is difficult to simplify later
on. This is especially true when the unroll factor is greater than one, and we
create vector arithmetic when computing step vectors. With this patch, we check
if an induction variable is only used for counting iterations or computing
addresses, and if so, scalarize the arithmetic when computing step vectors
instead. This allows for greater simplification.
This patch addresses the suboptimal pointer arithmetic sequence seen in
PR27881.
Reference: https://llvm.org/bugs/show_bug.cgi?id=27881
Differential Revision: http://reviews.llvm.org/D21620
llvm-svn: 274627
This patch also removes the SCEV variants of getStepVector() since they have no
uses after the refactoring.
Differential Revision: http://reviews.llvm.org/D21903
llvm-svn: 274558
Summary:
GetBoundryInstruction returns the last instruction as the instruction which follows or end(). Otherwise the last instruction in the boundry set is not being tested by isVectorizable().
Partially solve reordering of instructions. More extensive solution to follow.
Reviewers: tstellarAMD, llvm-commits, jlebar
Subscribers: escha, arsenm, mzolotukhin
Differential Revision: http://reviews.llvm.org/D21934
llvm-svn: 274389
integer.
Fixes issues on some architectures where we use arithmetic ops to build
vectors, which can cause bad things to happen for loads/stores of mixed
types.
Patch by Fiona Glaser
llvm-svn: 274307
Except the seed uniform instructions (conditional branch and consecutive ptr
instructions), dependencies to be added into uniform set should only be used
by existing uniform instructions or intructions outside of current loop.
Differential Revision: http://reviews.llvm.org/D21755
llvm-svn: 274262
For the new hotness attribute, the API will take the pass rather than
the pass name so we can no longer play the trick of AlwaysPrint being a
special pass name. This adds a getter to help the transition.
There is also a corresponding clang patch.
llvm-svn: 274100
It did not handle correctly cases without GEP.
The following loop wasn't vectorized:
for (int i=0; i<len; i++)
*to++ = *from++;
I use getPtrStride() to find Stride for memory access and return 0 is the Stride is not 1 or -1.
Re-commit rL273257 - revision: http://reviews.llvm.org/D20789
llvm-svn: 273864
The interleaved access analysis currently assumes that the inserted run-time
pointer aliasing checks ensure the absence of dependences that would prevent
its instruction reordering. However, this is not the case.
Issues can arise from how code generation is performed for interleaved groups.
For a load group, all loads in the group are essentially moved to the location
of the first load in program order, and for a store group, all stores in the
group are moved to the location of the last store. For groups having members
involved in a dependence relation with any other instruction in the loop, this
reordering can violate the dependence.
This patch teaches the interleaved access analysis how to avoid breaking such
dependences, and should fix PR27626.
An assumption of the original analysis was that the accesses had been collected
in "program order". The analysis was then simplified by visiting the accesses
bottom-up. However, this ordering was never guaranteed for anything other than
single basic block loops. Thus, this patch also enforces the desired ordering.
Reference: https://llvm.org/bugs/show_bug.cgi?id=27626
Differential Revision: http://reviews.llvm.org/D19984
llvm-svn: 273687
It did not handle correctly cases without GEP.
The following loop wasn't vectorized:
for (int i=0; i<len; i++)
*to++ = *from++;
I use getPtrStride() to find Stride for memory access and return 0 is the Stride is not 1 or -1.
Differential revision: http://reviews.llvm.org/D20789
llvm-svn: 273257
This is a functional change for LLE and LDist. The other clients (LV,
LVerLICM) already had this explicitly enabled.
The temporary boolean parameter to LAA is removed that allowed turning
off speculation of symbolic strides. This makes LAA's caching interface
LAA::getInfo only take the loop as the parameter. This makes the
interface more friendly to the new Pass Manager.
The flag -enable-mem-access-versioning is moved from LV to a LAA which
now allows turning off speculation globally.
llvm-svn: 273064
This is still NFCI, so the list of clients that allow symbolic stride
speculation does not change (yes: LV and LoopVersioningLICM, no: LLE,
LDist). However since the symbolic strides are now managed by LAA
rather than passed by client a new bool parameter is used to enable
symbolic stride speculation.
The existing test Transforms/LoopVectorize/version-mem-access.ll checks
that stride speculation is performed for LV.
The previously added test Transforms/LoopLoadElim/symbolic-stride.ll
ensures that no speculation is performed for LLE.
The next patch will change the functionality and turn on symbolic stride
speculation in all of LAA's clients and remove the bool parameter.
llvm-svn: 272970
Turns out SymbolicStrides is actually used in canVectorizeWithIfConvert
before it gets set up in canVectorizeMemory.
This works fine as long as SymbolicStrides resides in LV since we just
have an empty map. Based on this the conclusion is made that there are
no symbolic strides which is conservatively correct.
However once SymbolicStrides becomes part of LAI, LAI is nullptr at this
point so we need to differentiate the uninitialized state by returning a
nullptr for SymbolicStrides.
llvm-svn: 272966
LoopVectorizationLegality holds a constant reference to LAI, so this
will have to be const as well.
Also added missed function comment.
llvm-svn: 272851
The error on clang-x86-win2008-selfhost is:
C:\buildbot\slave-config\clang-x86-win2008-selfhost\llvm\lib\Transforms\Vectorize\SLPVectorizer.cpp(955) : error C2248: 'llvm::slpvectorizer::BoUpSLP::ScheduleData' : cannot access private struct declared in class 'llvm::slpvectorizer::BoUpSLP'
C:\buildbot\slave-config\clang-x86-win2008-selfhost\llvm\lib\Transforms\Vectorize\SLPVectorizer.cpp(608) : see declaration of 'llvm::slpvectorizer::BoUpSLP::ScheduleData'
C:\buildbot\slave-config\clang-x86-win2008-selfhost\llvm\lib\Transforms\Vectorize\SLPVectorizer.cpp(337) : see declaration of 'llvm::slpvectorizer::BoUpSLP'
I reproduced this locally with both MSVC 2013 and MSVC 2015.
llvm-svn: 272772
This wasn't failing for me with clang as the compiler. I think GCC may
disagree with clang about whether a friend declaration introduces a
declaration in the enclosing namespace (or something).
Example error:
/home/uweigand/sandbox/buildbot/clang-s390x-linux/llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp:950:77: error: ‘llvm::raw_ostream& llvm::slpvectorizer::operator<<(llvm::raw_ostream&, const llvm::slpvectorizer::BoUpSLP::ScheduleData&)’ should have been declared inside ‘llvm::slpvectorizer’
const BoUpSLP::ScheduleData &SD) {
^
llvm-svn: 272767
This uses the "runImpl" approach to share code with the old PM.
Porting to the new PM meant abandoning the anonymous namespace enclosing
most of SLPVectorizer.cpp which is a bit of a bummer (but not a big deal
compared to having to pull the pass class into a header which the new PM
requires since it calls the constructor directly).
llvm-svn: 272766
r272715 broke libcxx because it did not correctly handle cases where the
last iteration of one IV is the second-to-last iteration of another.
Original commit message:
Vectorizing loops with "escaping" IVs has been disabled since r190790, due to
PR17179. This re-enables it, with support for external use of both
"post-increment" (last iteration) and "pre-increment" (second-to-last iteration)
IVs.
llvm-svn: 272742
Vectorizing loops with "escaping" IVs has been disabled since r190790, due to
PR17179. This re-enables it, with support for external use of both
"post-increment" (last iteration) and "pre-increment" (second-to-last iteration)
IVs.
Differential Revision: http://reviews.llvm.org/D21048
llvm-svn: 272715
Previously, we materialized secondary vector IVs from the primary scalar IV,
by offseting the primary to match the correct start value, and then broadcasting
it - inside the loop body. Instead, we can use a real vector IV, like we do for
the primary.
This enables using vector IVs for secondary integer IVs whose type matches the
type of the primary.
Differential Revision: http://reviews.llvm.org/D20932
llvm-svn: 272283
Summary:
This fixes PR27617.
Bug description: The SLPVectorizer asserts on encountering GEPs with different index types, such as i8 and i64.
The patch includes a simple relaxation of the assert to allow constants being of different types, along with a regression test that will provoke the unrelaxed assert.
Reviewers: nadav, mzolotukhin
Subscribers: JesperAntonsson, llvm-commits, mzolotukhin
Differential Revision: http://reviews.llvm.org/D20685
Patch by Jesper Antonsson!
llvm-svn: 272206
Previously, whenever we needed a vector IV, we would create it on the fly,
by splatting the scalar IV and adding a step vector. Instead, we can create a
real vector IV. This tends to save a couple of instructions per iteration.
This only changes the behavior for the most basic case - integer primary
IVs with a constant step.
Differential Revision: http://reviews.llvm.org/D20315
llvm-svn: 271410
This patch fixes bug https://llvm.org/bugs/show_bug.cgi?id=27897.
When query memory access cost, current SLP always passes in alignment value of 1 (unaligned), so it gets a very high cost of scalar memory access, and wrongly vectorize memory loads in the test case.
It can be fixed by simply giving correct alignment.
llvm-svn: 271333
In truncateToMinimalBitwidths() we were RAUW'ing an instruction then erasing it. However, that intruction could be cached in the map we're iterating over. The first check is "I->use_empty()" which in most cases would return true, as the (deleted) object was RAUW'd first so would have zero use count. However in some cases the object could have been polluted or written over and this wouldn't be the case. Also it makes valgrind, asan and traditionalists who don't like their compiler to crash sad.
No testcase as there are no externally visible symptoms apart from a crash if the stars align.
Fixes PR26509.
llvm-svn: 269908
The selection of the vectorization factor currently doesn't consider
interleaved accesses. The vectorization factor is based on the maximum safe
dependence distance computed by LAA. However, for loops with interleaved
groups, we should instead base the vectorization factor on the maximum safe
dependence distance divided by the maximum interleave factor of all the
interleaved groups. Interleaved accesses not in a group will be scalarized.
Differential Revision: http://reviews.llvm.org/D20241
llvm-svn: 269659
Split FCMP//ICMP/SEL from the basic arithmetic cost functions. They were not sharing any notable code path (just the return) and were repeatedly testing the opcode.
llvm-svn: 269348
LoopVectorBody was changed from a single pointer to a SmallVector when
store predication was introduced in r200270. Since r247139, store predication
no longer splits the vector loop body in-place, so we can go back to having
a single LoopVectorBody block.
This reverts the no-longer-needed changes from r200270.
llvm-svn: 269321
Allow vectorization when the step is a loop-invariant variable.
This is the loop example that is getting vectorized after the patch:
int int_inc;
int bar(int init, int *restrict A, int N) {
int x = init;
for (int i=0;i<N;i++){
A[i] = x;
x += int_inc;
}
return x;
}
"x" is an induction variable with *loop-invariant* step.
But it is not a primary induction. Primary induction variable with non-constant step is not handled yet.
Differential Revision: http://reviews.llvm.org/D19258
llvm-svn: 269023
Summary:
Some PHIs can have expressions that are not AddRecExprs due to the presence
of sext/zext instructions. In order to prevent the Loop Vectorizer from
bailing out when encountering these PHIs, we now coerce the SCEV
expressions to AddRecExprs using SCEV predicates (when possible).
We only do this when the alternative would be to not vectorize.
Reviewers: mzolotukhin, anemet
Subscribers: mssimpso, sanjoy, mzolotukhin, llvm-commits
Differential Revision: http://reviews.llvm.org/D17153
llvm-svn: 268633
We need to keep loop hints from the original loop on the new vector loop.
Failure to do this meant that, for example:
void foo(int *b) {
#pragma clang loop unroll(disable)
for (int i = 0; i < 16; ++i)
b[i] = 1;
}
this loop would be unrolled. Why? Because we'd vectorize it, thus dropping the
hints that unrolling should be disabled, and then we'd unroll it.
llvm-svn: 267970
We previously disallowed interleaved load groups that may cause us to
speculatively access memory out-of-bounds (r261331). We did this by ensuring
each load group had an access corresponding to the first and last member.
Instead of bailing out for these interleaved groups, this patch enables us to
peel off the last vector iteration, ensuring that we execute at least one
iteration of the scalar remainder loop. This solution was proposed in the
review of the previous patch.
Differential Revision: http://reviews.llvm.org/D19487
llvm-svn: 267751
This is the first of two commits for extending SLP Vectorizer to deal with aggregates.
This commit merely refactors existing logic.
http://reviews.llvm.org/D14185
llvm-svn: 267748
This change adds a new hook for estimating the cost of vector extracts followed
by zero- and sign-extensions. The motivating example for this change is the
SMOV and UMOV instructions on AArch64. These instructions move data from vector
to general purpose registers while performing the corresponding extension
(sign-extend for SMOV and zero-extend for UMOV) at the same time. For these
operations, TargetTransformInfo can assume the extensions are free and only
report the cost of the vector extract. The SLP vectorizer has been updated to
make use of the new hook.
Differential Revision: http://reviews.llvm.org/D18523
llvm-svn: 267725
I really thought we were doing this already, but we were not. Given this input:
void Test(int *res, int *c, int *d, int *p) {
for (int i = 0; i < 16; i++)
res[i] = (p[i] == 0) ? res[i] : res[i] + d[i];
}
we did not vectorize the loop. Even with "assume_safety" the check that we
don't if-convert conditionally-executed loads (to protect against
data-dependent deferenceability) was not elided.
One subtlety: As implemented, it will still prefer to use a masked-load
instrinsic (given target support) over the speculated load. The choice here
seems architecture specific; the best option depends on how expensive the
masked load is compared to a regular load. Ideally, using the masked load still
reduces unnecessary memory traffic, and so should be preferred. If we'd rather
do it the other way, flipping the order of the checks is easy.
The LangRef is updated to make explicit that llvm.mem.parallel_loop_access also
implies that if conversion is okay.
Differential Revision: http://reviews.llvm.org/D19512
llvm-svn: 267514
The original commit was reverted because of a buildbot problem with LazyCallGraph::SCC handling (not related to the OptBisect handling).
Differential Revision: http://reviews.llvm.org/D19172
llvm-svn: 267231
This patch implements a optimization bisect feature, which will allow optimizations to be selectively disabled at compile time in order to track down test failures that are caused by incorrect optimizations.
The bisection is enabled using a new command line option (-opt-bisect-limit). Individual passes that may be skipped call the OptBisect object (via an LLVMContext) to see if they should be skipped based on the bisect limit. A finer level of control (disabling individual transformations) can be managed through an addition OptBisect method, but this is not yet used.
The skip checking in this implementation is based on (and replaces) the skipOptnoneFunction check. Where that check was being called, a new call has been inserted in its place which checks the bisect limit and the optnone attribute. A new function call has been added for module and SCC passes that behaves in a similar way.
Differential Revision: http://reviews.llvm.org/D19172
llvm-svn: 267022
The functionality contained within getIntrinsicIDForCall is two-fold: it
checks if a CallInst's callee is a vectorizable intrinsic. If it isn't
an intrinsic, it attempts to map the call's target to a suitable
intrinsic.
Move the mapping functionality into getIntrinsicForCallSite and rename
getIntrinsicIDForCall to getVectorIntrinsicIDForCall while
reimplementing it in terms of getIntrinsicForCallSite.
llvm-svn: 266801
Removed some unused headers, replaced some headers with forward class declarations.
Found using simple scripts like this one:
clear && ack --cpp -l '#include "llvm/ADT/IndexedMap.h"' | xargs grep -L 'IndexedMap[<]' | xargs grep -n --color=auto 'IndexedMap'
Patch by Eugene Kosov <claprix@yandex.ru>
Differential Revision: http://reviews.llvm.org/D19219
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 266595
Some SIMD implementations are not IEEE-754 compliant, for example ARM's NEON.
This patch teaches the loop vectorizer to only allow transformations of loops
that either contain no floating-point operations or have enough allowance
flags supporting lack of precision (ex. -ffast-math, Darwin).
For that, the target description now has a method which tells us if the
vectorizer is allowed to handle FP math without falling into unsafe
representations, plus a check on every FP instruction in the candidate loop
to check for the safety flags.
This commit makes LLVM behave like GCC with respect to ARM NEON support, but
it stops short of fixing the underlying problem: sub-normals. Neither GCC
nor LLVM have a flag for allowing sub-normal operations. Before this patch,
GCC only allows it using unsafe-math flags and LLVM allows it by default with
no way to turn it off (short of not using NEON at all).
As a first step, we push this change to make it safe and in sync with GCC.
The second step is to discuss a new sub-normal's flag on both communitues
and come up with a common solution. The third step is to improve the FastMath
flags in LLVM to encode sub-normals and use those flags to restrict NEON FP.
Fixes PR16275.
llvm-svn: 266363
The behavior of {MIN,MAX}NAN differs from that of {MIN,MAX}NUM when only
one of the inputs is NaN: -NUM will return the non-NaN argument while
-NAN would return NaN.
It is desirable to lower to @llvm.{min,max}num to -NAN if they don't
have a native instruction for -NUM. Notably, ARMv7 NEON's vmin has the
-NAN semantics.
N.B. Of course, it is only safe to do this if the intrinsic call is
marked nnan.
llvm-svn: 266279
Vectorization cost of uniform load wasn't correctly calculated.
As a result, a simple loop that loads a uniform value wasn't vectorized.
Differential Revision: http://reviews.llvm.org/D18940
llvm-svn: 265901
InstCombine cannot effectively remove redundant assumptions without them
registered in the assumption cache. The vectorizer can create identical
assumptions but doesn't register them with the cache, resulting in
slower compile times because InstCombine tries to reason about a lot
more assumptions.
Fix this by registering the cloned assumptions.
llvm-svn: 265800
This re-commits r265535 which was reverted in r265541 because it
broke the windows bots. The problem was that we had a PointerIntPair
which took a pointer to a struct allocated with new. The problem
was that new doesn't provide sufficient alignment guarantees.
This pattern was already present before r265535 and it just happened
to work. To fix this, we now separate the PointerToIntPair from the
ExitNotTakenInfo struct into a pointer and a bool.
Original commit message:
Summary:
When the backedge taken codition is computed from an icmp, SCEV can
deduce the backedge taken count only if one of the sides of the icmp
is an AddRecExpr. However, due to sign/zero extensions, we sometimes
end up with something that is not an AddRecExpr.
However, we can use SCEV predicates to produce a 'guarded' expression.
This change adds a method to SCEV to get this expression, and the
SCEV predicate associated with it.
In HowManyGreaterThans and HowManyLessThans we will now add a SCEV
predicate associated with the guarded backedge taken count when the
analyzed SCEV expression is not an AddRecExpr. Note that we only do
this as an alternative to returning a 'CouldNotCompute'.
We use new feature in Loop Access Analysis and LoopVectorize to analyze
and transform more loops.
Reviewers: anemet, mzolotukhin, hfinkel, sanjoy
Subscribers: flyingforyou, mcrosier, atrick, mssimpso, sanjoy, mzolotukhin, llvm-commits
Differential Revision: http://reviews.llvm.org/D17201
llvm-svn: 265786
Summary:
When the backedge taken codition is computed from an icmp, SCEV can
deduce the backedge taken count only if one of the sides of the icmp
is an AddRecExpr. However, due to sign/zero extensions, we sometimes
end up with something that is not an AddRecExpr.
However, we can use SCEV predicates to produce a 'guarded' expression.
This change adds a method to SCEV to get this expression, and the
SCEV predicate associated with it.
In HowManyGreaterThans and HowManyLessThans we will now add a SCEV
predicate associated with the guarded backedge taken count when the
analyzed SCEV expression is not an AddRecExpr. Note that we only do
this as an alternative to returning a 'CouldNotCompute'.
We use new feature in Loop Access Analysis and LoopVectorize to analyze
and transform more loops.
Reviewers: anemet, mzolotukhin, hfinkel, sanjoy
Subscribers: flyingforyou, mcrosier, atrick, mssimpso, sanjoy, mzolotukhin, llvm-commits
Differential Revision: http://reviews.llvm.org/D17201
llvm-svn: 265535
A catchswitch cannot be preceded by another instruction in the same
basic block (other than a PHI node).
Instead, insert the extract element right after the materialization of
the vectorized value. This isn't optimal but is a reasonable compromise
given the constraints of WinEH.
This fixes PR27163.
llvm-svn: 265157
This change prevents the loop vectorizer from vectorizing when all of the vector
types it generates will be scalarized. I've run into this problem on the PPC's QPX
vector ISA, which only holds floating-point vector types. The loop vectorizer
will, however, happily vectorize loops with purely integer computation. Here's
an example:
LV: The Smallest and Widest types: 32 / 32 bits.
LV: The Widest register is: 256 bits.
LV: Found an estimated cost of 0 for VF 1 For instruction: %indvars.iv25 = phi i64 [ 0, %entry ], [ %indvars.iv.next26, %for.body ]
LV: Found an estimated cost of 0 for VF 1 For instruction: %arrayidx = getelementptr inbounds [1600 x i32], [1600 x i32]* %a, i64 0, i64 %indvars.iv25
LV: Found an estimated cost of 0 for VF 1 For instruction: %2 = trunc i64 %indvars.iv25 to i32
LV: Found an estimated cost of 1 for VF 1 For instruction: store i32 %2, i32* %arrayidx, align 4
LV: Found an estimated cost of 1 for VF 1 For instruction: %indvars.iv.next26 = add nuw nsw i64 %indvars.iv25, 1
LV: Found an estimated cost of 1 for VF 1 For instruction: %exitcond27 = icmp eq i64 %indvars.iv.next26, 1600
LV: Found an estimated cost of 0 for VF 1 For instruction: br i1 %exitcond27, label %for.cond.cleanup, label %for.body
LV: Scalar loop costs: 3.
LV: Found an estimated cost of 0 for VF 2 For instruction: %indvars.iv25 = phi i64 [ 0, %entry ], [ %indvars.iv.next26, %for.body ]
LV: Found an estimated cost of 0 for VF 2 For instruction: %arrayidx = getelementptr inbounds [1600 x i32], [1600 x i32]* %a, i64 0, i64 %indvars.iv25
LV: Found an estimated cost of 0 for VF 2 For instruction: %2 = trunc i64 %indvars.iv25 to i32
LV: Found an estimated cost of 2 for VF 2 For instruction: store i32 %2, i32* %arrayidx, align 4
LV: Found an estimated cost of 1 for VF 2 For instruction: %indvars.iv.next26 = add nuw nsw i64 %indvars.iv25, 1
LV: Found an estimated cost of 1 for VF 2 For instruction: %exitcond27 = icmp eq i64 %indvars.iv.next26, 1600
LV: Found an estimated cost of 0 for VF 2 For instruction: br i1 %exitcond27, label %for.cond.cleanup, label %for.body
LV: Vector loop of width 2 costs: 2.
LV: Found an estimated cost of 0 for VF 4 For instruction: %indvars.iv25 = phi i64 [ 0, %entry ], [ %indvars.iv.next26, %for.body ]
LV: Found an estimated cost of 0 for VF 4 For instruction: %arrayidx = getelementptr inbounds [1600 x i32], [1600 x i32]* %a, i64 0, i64 %indvars.iv25
LV: Found an estimated cost of 0 for VF 4 For instruction: %2 = trunc i64 %indvars.iv25 to i32
LV: Found an estimated cost of 4 for VF 4 For instruction: store i32 %2, i32* %arrayidx, align 4
LV: Found an estimated cost of 1 for VF 4 For instruction: %indvars.iv.next26 = add nuw nsw i64 %indvars.iv25, 1
LV: Found an estimated cost of 1 for VF 4 For instruction: %exitcond27 = icmp eq i64 %indvars.iv.next26, 1600
LV: Found an estimated cost of 0 for VF 4 For instruction: br i1 %exitcond27, label %for.cond.cleanup, label %for.body
LV: Vector loop of width 4 costs: 1.
...
LV: Selecting VF: 8.
LV: The target has 32 registers
LV(REG): Calculating max register usage:
LV(REG): At #0 Interval # 0
LV(REG): At #1 Interval # 1
LV(REG): At #2 Interval # 2
LV(REG): At #4 Interval # 1
LV(REG): At #5 Interval # 1
LV(REG): VF = 8
The problem is that the cost model here is not wrong, exactly. Since all of
these operations are scalarized, their cost (aside from the uniform ones) are
indeed VF*(scalar cost), just as the model suggests. In fact, the larger the VF
picked, the lower the relative overhead from the loop itself (and the
induction-variable update and check), and so in a sense, picking the largest VF
here is the right thing to do.
The problem is that vectorizing like this, where all of the vectors will be
scalarized in the backend, isn't really vectorizing, but rather interleaving.
By itself, this would be okay, but then the vectorizer itself also interleaves,
and that's where the problem manifests itself. There's aren't actually enough
scalar registers to support the normal interleave factor multiplied by a factor
of VF (8 in this example). In other words, the problem with this is that our
register-pressure heuristic does not account for scalarization.
While we might want to improve our register-pressure heuristic, I don't think
this is the right motivating case for that work. Here we have a more-basic
problem: The job of the vectorizer is to vectorize things (interleaving aside),
and if the IR it generates won't generate any actual vector code, then
something is wrong. Thus, if every type looks like it will be scalarized (i.e.
will be split into VF or more parts), then don't consider that VF.
This is not a problem specific to PPC/QPX, however. The problem comes up under
SSE on x86 too, and as such, this change fixes PR26837 too. I've added Sanjay's
reduced test case from PR26837 to this commit.
Differential Revision: http://reviews.llvm.org/D18537
llvm-svn: 264904
These checks are redundant and can be removed
Reviewers: hans
Subscribers: llvm-commits, mzolotukhin
Differential Revision: http://reviews.llvm.org/D18564
llvm-svn: 264872
Summary:
Use the new LoopVersioning facility (D16712) to add noalias metadata in
the vector loop if we versioned with memchecks. This can enable some
optimization opportunities further down the pipeline (see the included
test or the benchmark improvement quoted in D16712).
The test also covers the bug I had in the initial version in D16712.
The vectorizer did not previously use LoopVersioning. The reason is
that the vectorizer performs its transformations in single shot. It
creates an empty single-block vector loop that it then populates with
the widened, if-converted instructions. Thus creating an intermediate
versioned scalar loop seems wasteful.
So this patch (rather than bringing in LoopVersioning fully) adds a
special interface to LoopVersioning to allow the vectorizer to add
no-alias annotation while still performing its own versioning.
As the vectorizer propagates metadata from the instructions in the
original loop to the vector instructions we also check the pointer in
the original instruction and see if LoopVersioning can add no-alias
metadata based on the issued memchecks.
Reviewers: hfinkel, nadav, mzolotukhin
Subscribers: mzolotukhin, llvm-commits
Differential Revision: http://reviews.llvm.org/D17191
llvm-svn: 263744
This was a latent bug that got exposed by the change to add LoopSimplify
as a dependence to LoopLoadElimination. Since LoopInfo was corrupted
after LV, LoopSimplify mis-compiled nbench in the test-suite (more
details in the PR).
The problem was that when we create the blocks for predicated stores we
didn't add those to any loops.
The original testcase for store predication provides coverage for this
assuming we verify LI on the way out of LV.
Fixes PR26952.
llvm-svn: 263565
commit ae14bf6488e8441f0f6d74f00455555f6f3943ac
Author: Mehdi Amini <mehdi.amini@apple.com>
Date: Fri Mar 11 17:15:50 2016 +0000
Remove PreserveNames template parameter from IRBuilder
Summary:
Following r263086, we are now relying on a flag on the Context to
discard Value names in release builds.
Reviewers: chandlerc
Subscribers: mzolotukhin, llvm-commits
Differential Revision: http://reviews.llvm.org/D18023
From: Mehdi Amini <mehdi.amini@apple.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@263258
91177308-0d34-0410-b5e6-96231b3b80d8
until we can figure out what to do about clang and Release build testing.
This reverts commit 263258.
llvm-svn: 263321
Summary:
Following r263086, we are now relying on a flag on the Context to
discard Value names in release builds.
Reviewers: chandlerc
Subscribers: mzolotukhin, llvm-commits
Differential Revision: http://reviews.llvm.org/D18023
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 263258
MinVecRegSize is currently hardcoded to 128; this patch adds a cl::opt
to allow changing it. I tried not to change any existing behavior for the default
case.
Differential revision: http://reviews.llvm.org/D13278
llvm-svn: 263089
I missed == and != when I removed implicit conversions between iterators
and pointers in r252380 since they were defined outside ilist_iterator.
Since they depend on getNodePtrUnchecked(), they indirectly rely on UB.
This commit removes all uses of these operators. (I'll delete the
operators themselves in a separate commit so that it can be easily
reverted if necessary.)
There should be NFC here.
llvm-svn: 261498
This patch enables the vectorization of first-order recurrences. A first-order
recurrence is a non-reduction recurrence relation in which the value of the
recurrence in the current loop iteration equals a value defined in the previous
iteration. The load PRE of the GVN pass often creates these recurrences by
hoisting loads from within loops.
In this patch, we add a new recurrence kind for first-order phi nodes and
attempt to vectorize them if possible. Vectorization is performed by shuffling
the values for the current and previous iterations. The vectorization cost
estimate is updated to account for the added shuffle instruction.
Contributed-by: Matthew Simpson and Chad Rosier <mcrosier@codeaurora.org>
Differential Revision: http://reviews.llvm.org/D16197
llvm-svn: 261346
Summary:
If we don't have the first and last access of an interleaved load group,
the first and last wide load in the loop can do an out of bounds
access. Even though we discard results from speculative loads,
this can cause problems, since it can technically generate page faults
(or worse).
We now discard interleaved load groups that don't have the first and
load in the group.
Reviewers: hfinkel, rengolin
Subscribers: rengolin, llvm-commits, mzolotukhin, anemet
Differential Revision: http://reviews.llvm.org/D17332
llvm-svn: 261331
Commit r259357 was reverted because it caused PR26629. We were assuming all
roots of a vectorizable tree could be truncated to the same width, which is not
the case in general. This commit reapplies the patch along with a fix and a new
test case to ensure we don't regress because of this issue again. This should
fix PR26629.
llvm-svn: 261212
Loop vectorizer now knows to vectorize GEP and create masked gather and scatter intrinsics for random memory access.
The feature is enabled on AVX-512 target.
Differential Revision: http://reviews.llvm.org/D15690
llvm-svn: 261140
Summary:
While shrinking types according to the required bits, we can
encounter insert/extract element instructions. This will cause us to
reach an llvm_unreachable statement.
This change adds support for truncating insert/extract element
operations, and adds a regression test.
Reviewers: jmolloy
Subscribers: mzolotukhin, llvm-commits
Differential Revision: http://reviews.llvm.org/D17078
llvm-svn: 260893
sanitizer issue. The PredicatedScalarEvolution's copy constructor
wasn't copying the Generation value, and was leaving it un-initialized.
Original commit message:
[SCEV][LAA] Add no wrap SCEV predicates and use use them to improve strided pointer detection
Summary:
This change adds no wrap SCEV predicates with:
- support for runtime checking
- support for expression rewriting:
(sext ({x,+,y}) -> {sext(x),+,sext(y)}
(zext ({x,+,y}) -> {zext(x),+,sext(y)}
Note that we are sign extending the increment of the SCEV, even for
the zext case. This is needed to cover the fairly common case where y would
be a (small) negative integer. In order to do this, this change adds two new
flags: nusw and nssw that are applicable to AddRecExprs and permit the
transformations above.
We also change isStridedPtr in LAA to be able to make use of
these predicates. With this feature we should now always be able to
work around overflow issues in the dependence analysis.
Reviewers: mzolotukhin, sanjoy, anemet
Subscribers: mzolotukhin, sanjoy, llvm-commits, rengolin, jmolloy, hfinkel
Differential Revision: http://reviews.llvm.org/D15412
llvm-svn: 260112
Summary:
This change adds no wrap SCEV predicates with:
- support for runtime checking
- support for expression rewriting:
(sext ({x,+,y}) -> {sext(x),+,sext(y)}
(zext ({x,+,y}) -> {zext(x),+,sext(y)}
Note that we are sign extending the increment of the SCEV, even for
the zext case. This is needed to cover the fairly common case where y would
be a (small) negative integer. In order to do this, this change adds two new
flags: nusw and nssw that are applicable to AddRecExprs and permit the
transformations above.
We also change isStridedPtr in LAA to be able to make use of
these predicates. With this feature we should now always be able to
work around overflow issues in the dependence analysis.
Reviewers: mzolotukhin, sanjoy, anemet
Subscribers: mzolotukhin, sanjoy, llvm-commits, rengolin, jmolloy, hfinkel
Differential Revision: http://reviews.llvm.org/D15412
llvm-svn: 260085
Current SCEV expansion will expand SCEV as a sequence of operations
and doesn't utilize the value already existed. This will introduce
redundent computation which may not be cleaned up throughly by
following optimizations.
This patch introduces an ExprValueMap which is a map from SCEV to the
set of equal values with the same SCEV. When a SCEV is expanded, the
set of values is checked and reused whenever possible before generating
a sequence of operations.
The original commit triggered regressions in Polly tests. The regressions
exposed two problems which have been fixed in current version.
1. Polly will generate a new function based on the old one. To generate an
instruction for the new function, it builds SCEV for the old instruction,
applies some tranformation on the SCEV generated, then expands the transformed
SCEV and insert the expanded value into new function. Because SCEV expansion
may reuse value cached in ExprValueMap, the value in old function may be
inserted into new function, which is wrong.
In SCEVExpander::expand, there is a logic to check the cached value to
be used should dominate the insertion point. However, for the above
case, the check always passes. That is because the insertion point is
in a new function, which is unreachable from the old function. However
for unreachable node, DominatorTreeBase::dominates thinks it will be
dominated by any other node.
The fix is to simply add a check that the cached value to be used in
expansion should be in the same function as the insertion point instruction.
2. When the SCEV is of scConstant type, expanding it directly is cheaper than
reusing a normal value cached. Although in the cached value set in ExprValueMap,
there is a Constant type value, but it is not easy to find it out -- the cached
Value set is not sorted according to the potential cost. Existing reuse logic
in SCEVExpander::expand simply chooses the first legal element from the cached
value set.
The fix is that when the SCEV is of scConstant type, don't try the reuse
logic. simply expand it.
Differential Revision: http://reviews.llvm.org/D12090
llvm-svn: 259736
Current SCEV expansion will expand SCEV as a sequence of operations
and doesn't utilize the value already existed. This will introduce
redundent computation which may not be cleaned up throughly by
following optimizations.
This patch introduces an ExprValueMap which is a map from SCEV to the
set of equal values with the same SCEV. When a SCEV is expanded, the
set of values is checked and reused whenever possible before generating
a sequence of operations.
Differential Revision: http://reviews.llvm.org/D12090
llvm-svn: 259662
In the future, we will vectorize recurrences other than reductions. This patch
renames a few variables and updates their associated comments to enable them to
be reused for non-reduction PHI nodes.
This change was requested in the review for D16197.
llvm-svn: 259364
This patch is the second attempt to reapply commit r258404. There was bug in
the initial patch and subsequent fix (mentioned below).
The initial patch caused an assertion because we were computing smaller type
sizes for instructions that cannot be demoted. The fix first determines the
instructions that will be demoted, and then applies the smaller type size to
only those instructions.
This should fix PR26239 and PR26307.
llvm-svn: 258929
Summary:
This patch is provided in preparation for removing autoconf on 1/26. The proposal to remove autoconf on 1/26 was discussed on the llvm-dev thread here: http://lists.llvm.org/pipermail/llvm-dev/2016-January/093875.html
"I felt a great disturbance in the [build system], as if millions of [makefiles] suddenly cried out in terror and were suddenly silenced. I fear something [amazing] has happened."
- Obi Wan Kenobi
Reviewers: chandlerc, grosbach, bob.wilson, tstellarAMD, echristo, whitequark
Subscribers: chfast, simoncook, emaste, jholewinski, tberghammer, jfb, danalbert, srhines, arsenm, dschuff, jyknight, dsanders, joker.eph, llvm-commits
Differential Revision: http://reviews.llvm.org/D16471
llvm-svn: 258861
This commit exposes a crash in computeKnownBits on the Chromium buildbots.
Reverting to investigate.
Reference: https://llvm.org/bugs/show_bug.cgi?id=26307
llvm-svn: 258812
This is a recommit of r258620 which causes PR26293.
The original message:
Now LIR can turn following codes into memset:
typedef struct foo {
int a;
int b;
} foo_t;
void bar(foo_t *f, unsigned n) {
for (unsigned i = 0; i < n; ++i) {
f[i].a = 0;
f[i].b = 0;
}
}
void test(foo_t *f, unsigned n) {
for (unsigned i = 0; i < n; i += 2) {
f[i] = 0;
f[i+1] = 0;
}
}
llvm-svn: 258777
We were hitting an assertion because we were computing smaller type sizes for
instructions that cannot be demoted. The fix first determines the instructions
that will be demoted, and then applies the smaller type size to only those
instructions.
This should fix PR26239.
llvm-svn: 258705
This change attempts to produce vectorized integer expressions in bit widths
that are narrower than their scalar counterparts. The need for demotion arises
especially on architectures in which the small integer types (e.g., i8 and i16)
are not legal for scalar operations but can still be used in vectors. Like
similar work done within the loop vectorizer, we rely on InstCombine to perform
the actual type-shrinking. We use the DemandedBits analysis and
ComputeNumSignBits from ValueTracking to determine the minimum required bit
width of an expression.
Differential revision: http://reviews.llvm.org/D15815
llvm-svn: 258404
The fix uniques the bundle of getelementptr indices we are about to vectorize
since it's possible for the same index to be used by multiple instructions.
The original commit message is below.
[SLP] Vectorize the index computations of getelementptr instructions.
This patch seeds the SLP vectorizer with getelementptr indices. The primary
motivation in doing so is to vectorize gather-like idioms beginning with
consecutive loads (e.g., g[a[0] - b[0]] + g[a[1] - b[1]] + ...). While these
cases could be vectorized with a top-down phase, seeding the existing bottom-up
phase with the index computations avoids the complexity, compile-time, and
phase ordering issues associated with a full top-down pass. Only bundles of
single-index getelementptrs with non-constant differences are considered for
vectorization.
llvm-svn: 257918
This patch seeds the SLP vectorizer with getelementptr indices. The primary
motivation in doing so is to vectorize gather-like idioms beginning with
consecutive loads (e.g., g[a[0] - b[0]] + g[a[1] - b[1]] + ...). While these
cases could be vectorized with a top-down phase, seeding the existing bottom-up
phase with the index computations avoids the complexity, compile-time, and
phase ordering issues associated with a full top-down pass. Only bundles of
single-index getelementptrs with non-constant differences are considered for
vectorization.
Differential Revision: http://reviews.llvm.org/D14829
llvm-svn: 257800
This patch prevents us from unintentionally creating entries in the reductions
map for PHIs that are not actually reductions. This is currently not an issue
since we bail out if we encounter PHIs other than inductions or reductions.
However the behavior could become problematic as we add support for additional
recurrence types.
llvm-svn: 256930
When considering incoming values as part of a reduction phi, ensure the
incoming value is dominated by said phi.
Failing to ensure this property causes miscompiles.
Fixes PR25787.
Many thanks to Mattias Eriksson for reporting, reducing and analyzing the
problem for me.
Differential Revision: http://reviews.llvm.org/D15580
llvm-svn: 255792
(This is the third attempt to check in this patch, and the first two are r255454
and r255460. The once failed test file reg-usage.ll is now moved to
test/Transform/LoopVectorize/X86 directory with target datalayout and target
triple indicated.)
LoopVectorizationCostModel::calculateRegisterUsage() is used to estimate the
register usage for specific VFs. However, it takes into account many
instructions that won't be vectorized, such as induction variables,
GetElementPtr instruction, etc.. This makes the loop vectorizer too conservative
when choosing VF. In this patch, the induction variables that won't be
vectorized plus GetElementPtr instruction will be added to ValuesToIgnore set
so that their register usage won't be considered any more.
Differential revision: http://reviews.llvm.org/D15177
llvm-svn: 255691
(This is the second attempt to check in this patch: REQUIRES: asserts is added
to reg-usage.ll now.)
LoopVectorizationCostModel::calculateRegisterUsage() is used to estimate the
register usage for specific VFs. However, it takes into account many
instructions that won't be vectorized, such as induction variables,
GetElementPtr instruction, etc.. This makes the loop vectorizer too conservative
when choosing VF. In this patch, the induction variables that won't be
vectorized plus GetElementPtr instruction will be added to ValuesToIgnore set
so that their register usage won't be considered any more.
Differential revision: http://reviews.llvm.org/D15177
llvm-svn: 255460
LoopVectorizationCostModel::calculateRegisterUsage() is used to estimate the
register usage for specific VFs. However, it takes into account many
instructions that won't be vectorized, such as induction variables,
GetElementPtr instruction, etc.. This makes the loop vectorizer too conservative
when choosing VF. In this patch, the induction variables that won't be
vectorized plus GetElementPtr instruction will be added to ValuesToIgnore set
so that their register usage won't be considered any more.
Differential revision: http://reviews.llvm.org/D15177
llvm-svn: 255454
GlobalsAA's assumptions that passes do not escape globals not previously
escaped is not violated by AlignmentFromAssumptions and SLPVectorizer. Marking
them as such allows GlobalsAA to be preserved until GVN in the LTO pipeline.
http://lists.llvm.org/pipermail/llvm-dev/2015-December/092972.html
Patch by Vaivaswatha Nagaraj!
llvm-svn: 255348
ScalarEvolution.h, in order to avoid cyclic dependencies between the Transform
and Analysis modules:
[LV][LAA] Add a layer over SCEV to apply run-time checked knowledge on SCEV expressions
Summary:
This change creates a layer over ScalarEvolution for LAA and LV, and centralizes the
usage of SCEV predicates. The SCEVPredicatedLayer takes the statically deduced knowledge
by ScalarEvolution and applies the knowledge from the SCEV predicates. The end goal is
that both LAA and LV should use this interface everywhere.
This also solves a problem involving the result of SCEV expression rewritting when
the predicate changes. Suppose we have the expression (sext {a,+,b}) and two predicates
P1: {a,+,b} has nsw
P2: b = 1.
Applying P1 and then P2 gives us {a,+,1}, while applying P2 and the P1 gives us
sext({a,+,1}) (the AddRec expression was changed by P2 so P1 no longer applies).
The SCEVPredicatedLayer maintains the order of transformations by feeding back
the results of previous transformations into new transformations, and therefore
avoiding this issue.
The SCEVPredicatedLayer maintains a cache to remember the results of previous
SCEV rewritting results. This also has the benefit of reducing the overall number
of expression rewrites.
Reviewers: mzolotukhin, anemet
Subscribers: jmolloy, sanjoy, llvm-commits
Differential Revision: http://reviews.llvm.org/D14296
llvm-svn: 255122
Summary:
This change creates a layer over ScalarEvolution for LAA and LV, and centralizes the
usage of SCEV predicates. The SCEVPredicatedLayer takes the statically deduced knowledge
by ScalarEvolution and applies the knowledge from the SCEV predicates. The end goal is
that both LAA and LV should use this interface everywhere.
This also solves a problem involving the result of SCEV expression rewritting when
the predicate changes. Suppose we have the expression (sext {a,+,b}) and two predicates
P1: {a,+,b} has nsw
P2: b = 1.
Applying P1 and then P2 gives us {a,+,1}, while applying P2 and the P1 gives us
sext({a,+,1}) (the AddRec expression was changed by P2 so P1 no longer applies).
The SCEVPredicatedLayer maintains the order of transformations by feeding back
the results of previous transformations into new transformations, and therefore
avoiding this issue.
The SCEVPredicatedLayer maintains a cache to remember the results of previous
SCEV rewritting results. This also has the benefit of reducing the overall number
of expression rewrites.
Reviewers: mzolotukhin, anemet
Subscribers: jmolloy, sanjoy, llvm-commits
Differential Revision: http://reviews.llvm.org/D14296
llvm-svn: 255115
The order in which instructions are truncated in truncateToMinimalBitwidths
effects code generation. Switch to a map with a determinisic order, since the
iteration order over a DenseMap is not defined.
This code is not hot, so the difference in container performance isn't
interesting.
Many thanks to David Blaikie for making me aware of MapVector!
Fixes PR25490.
Differential Revision: http://reviews.llvm.org/D14981
llvm-svn: 254179
Measurements primarily on AArch64 have shown this feature does not
significantly effect compile-time. The are no significant perf changes in LNT,
but for AArch64 at least, there are wins in third party benchmarks.
As discussed on llvm-dev, we're going to try turning this on by default and see
how other targets react to the change.
llvm-svn: 252733
Implemented as many of Michael's suggestions as were possible:
* clang-format the added code while it is still fresh.
* tried to change Value* to Instruction* in many places in computeMinimumValueSizes - unfortunately there are several places where Constants need to be handled so this wasn't possible.
* Reduce the pass list on loop-vectorization-factors.ll.
* Fix a bug where we were querying MinBWs for I->getOperand(0) but using MinBWs[I].
llvm-svn: 252469
The SLPVectorizer had a very crude way of trying to benefit
from associativity: it tried to optimize for splat/broadcast
or in order to have the same operator on the same side.
This is benefitial to the cost model and allows more vectorization
to occur.
This patch improve the logic and make the detection optimal (locally,
we don't look at the full tree but only at the immediate children).
Should fix https://llvm.org/bugs/show_bug.cgi?id=25247
Reviewers: mzolotukhin
Differential Revision: http://reviews.llvm.org/D13996
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 252337
To be able to maximize the bandwidth during vectorization, this patch provides a new flag vectorizer-maximize-bandwidth. When it is turned on, the vectorizer will determine the vectorization factor (VF) using the smallest instead of widest type in the loop. To avoid increasing register pressure too much, estimates of the register usage for different VFs are calculated so that we only choose a VF when its register usage doesn't exceed the number of available registers.
This is the second attempt to submit this patch. The first attempt got a test failure on ARM. This patch is updated to try to fix the failure (more specifically, by handling the case when VF=1).
Differential revision: http://reviews.llvm.org/D8943
llvm-svn: 251850
Summary:
SCEV Predicates represent conditions that typically cannot be derived from
static analysis, but can be used to reduce SCEV expressions to forms which are
usable for different optimizers.
ScalarEvolution now has the rewriteUsingPredicate method which can simplify a
SCEV expression using a SCEVPredicateSet. The normal workflow of a pass using
SCEVPredicates would be to hold a SCEVPredicateSet and every time assumptions
need to be made a new SCEV Predicate would be created and added to the set.
Each time after calling getSCEV, the user will call the rewriteUsingPredicate
method.
We add two types of predicates
SCEVPredicateSet - implements a set of predicates
SCEVEqualPredicate - tests for equality between two SCEV expressions
We use the SCEVEqualPredicate to re-implement stride versioning. Every time we
version a stride, we will add a SCEVEqualPredicate to the context.
Instead of adding specific stride checks, LoopVectorize now adds a more
generic SCEV check.
We only need to add support for this in the LoopVectorizer since this is the
only pass that will do stride versioning.
Reviewers: mzolotukhin, anemet, hfinkel, sanjoy
Subscribers: sanjoy, hfinkel, rengolin, jmolloy, llvm-commits
Differential Revision: http://reviews.llvm.org/D13595
llvm-svn: 251800
To be able to maximize the bandwidth during vectorization, this patch provides a new flag vectorizer-maximize-bandwidth. When it is turned on, the vectorizer will determine the vectorization factor (VF) using the smallest instead of widest type in the loop. To avoid increasing register pressure too much, estimates of the register usage for different VFs are calculated so that we only choose a VF when its register usage doesn't exceed the number of available registers.
llvm-svn: 251592
Summary:
This change could be way off-piste, I'm looking for any feedback on whether it's an acceptable approach.
It never seems to be a problem to gobble up as many reduction values as can be found, and then to attempt to reduce the resulting tree. Some of the workloads I'm looking at have been aggressively unrolled by hand, and by selecting reduction widths that are not constrained by a vector register size, it becomes possible to profitably vectorize. My test case shows such an unrolling which SLP was not vectorizing (on neither ARM nor X86) before this patch, but with it does vectorize.
I measure no significant compile time impact of this change when combined with D13949 and D14063. There are also no significant performance regressions on ARM/AArch64 in SPEC or LNT.
The more principled approach I thought of was to generate several candidate tree's and use the cost model to pick the cheapest one. That seemed like quite a big design change (the algorithms seem very much one-shot), and would likely be a costly thing for compile time. This seemed to do the job at very little cost, but I'm worried I've misunderstood something!
Reviewers: nadav, jmolloy
Subscribers: mssimpso, llvm-commits, aemerson
Differential Revision: http://reviews.llvm.org/D14116
llvm-svn: 251428
Summary:
Currently, when the SLP vectorizer considers whether a phi is part of a reduction, it dismisses phi's whose incoming blocks are not the same as the block containing the phi. For the patterns I'm looking at, extending this rule to allow phis whose incoming block is a containing loop latch allows me to vectorize certain workloads.
There is no significant compile-time impact, and combined with D13949, no performance improvement measured in ARM/AArch64 in any of SPEC2000, SPEC2006 or LNT.
Reviewers: jmolloy, mcrosier, nadav
Subscribers: mssimpso, nadav, aemerson, llvm-commits
Differential Revision: http://reviews.llvm.org/D14063
llvm-svn: 251425
Summary:
Certain workloads, in particular sum-of-absdiff loops, can be vectorized using SLP if it can treat select instructions as reduction values.
The test case is a bit awkward. The AArch64 cost model needs some tuning to not be so pessimistic about selects. I've had to tweak the SLP threshold here.
Reviewers: jmolloy, mzolotukhin, spatel, nadav
Subscribers: nadav, mssimpso, aemerson, llvm-commits
Differential Revision: http://reviews.llvm.org/D13949
llvm-svn: 251424
Vectorization of memory instruction (Load/Store) is possible when the pointer is coming from GEP. The GEP analysis allows to estimate the profit.
In some cases we have a "bitcast" between GEP and memory instruction.
I added code that skips the "bitcast".
http://reviews.llvm.org/D13886
llvm-svn: 251291
Besides the usual, I finally added an overload to
`BasicBlock::splitBasicBlock()` that accepts an `Instruction*` instead
of `BasicBlock::iterator`. Someone can go back and remove this overload
later (after updating the callers I'm going to skip going forward), but
the most common call seems to be
`BB->splitBasicBlock(BB->getTerminator(), ...)` and I'm not sure it's
better to add `->getIterator()` to every one than have the overload.
It's pretty hard to get the usage wrong.
llvm-svn: 250745
Originally I planned to use the same interface for masked gather/scatter and set isConsecutive to "false" in this case.
Now I'm implementing masked gather/scatter and see that the interface is inconvenient. I want to add interfaces isLegalMaskedGather() / isLegalMaskedScatter() instead of using the "Consecutive" parameter in the existing interfaces.
Differential Revision: http://reviews.llvm.org/D13850
llvm-svn: 250686
C semantics force sub-int-sized values (e.g. i8, i16) to be promoted to int
type (e.g. i32) whenever arithmetic is performed on them.
For targets with native i8 or i16 operations, usually InstCombine can shrink
the arithmetic type down again. However InstCombine refuses to create illegal
types, so for targets without i8 or i16 registers, the lengthening and
shrinking remains.
Most SIMD ISAs (e.g. NEON) however support vectors of i8 or i16 even when
their scalar equivalents do not, so during vectorization it is important to
remove these lengthens and truncates when deciding the profitability of
vectorization.
The algorithm this uses starts at truncs and icmps, trawling their use-def
chains until they terminate or instructions outside the loop are found (or
unsafe instructions like inttoptr casts are found). If the use-def chains
starting from different root instructions (truncs/icmps) meet, they are
unioned. The demanded bits of each node in the graph are ORed together to form
an overall mask of the demanded bits in the entire graph. The minimum bitwidth
that graph can be truncated to is the bitwidth minus the number of leading
zeroes in the overall mask.
The intention is that this algorithm should "first do no harm", so it will
never insert extra cast instructions. This is why the use-def graphs are
unioned, so that subgraphs with different minimum bitwidths do not need casts
inserted between them.
This algorithm works hard to reduce compile time impact. DemandedBits are only
queried if there are extends of illegal types and if a truncate to an illegal
type is seen. In the general case, this results in a simple linear scan of the
instructions in the loop.
No non-noise compile time impact was seen on a clang bootstrap build.
llvm-svn: 250032
The most important part required to make clang
devirtualization works ( ͡°͜ʖ ͡°).
The code is able to find non local dependencies, but unfortunatelly
because the caller can only handle local dependencies, I had to add
some restrictions to look for dependencies only in the same BB.
http://reviews.llvm.org/D12992
llvm-svn: 249196
Summary:
Given an array of i2 elements, 4 consecutive scalar loads will be lowered to
i8-sized loads and thus will access 4 consecutive bytes in memory. If we
vectorize these loads into a single <4 x i2> load, it'll access only 1 byte in
memory. Hence, we should prohibit vectorization in such cases.
PS: Initial patch was proposed by Arnold.
Reviewers: aschwaighofer, nadav, hfinkel
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D13277
llvm-svn: 248943
Usually large blocks are not a problem. But if a large block (> 10k instructions)
contains many (potential) chains of vector instructions, and those chains are
spread over a wide range of instructions, then scheduling becomes a compile time problem.
This change introduces a limit for the accumulate scheduling region size of a block.
For real-world functions this limit will never be exceeded (it's about 10x larger than
the maximum value seen in the test-suite and external test suite).
llvm-svn: 248917
Summary:
It is fairly common to call SE->getConstant(Ty, 0) or
SE->getConstant(Ty, 1); this change makes such uses a little bit
briefer.
I've refactored the call sites I could find easily to use getZero /
getOne.
Reviewers: hfinkel, majnemer, reames
Subscribers: sanjoy, llvm-commits
Differential Revision: http://reviews.llvm.org/D12947
llvm-svn: 248362
We're currently losing any fast-math flags when synthesizing fcmps for
min/max reductions. In LV, make sure we copy over the scalar inst's
flags. In LoopUtils, we know we only ever match patterns with
hasUnsafeAlgebra, so apply that to any synthesized ops.
llvm-svn: 248201
with the new pass manager, and no longer relying on analysis groups.
This builds essentially a ground-up new AA infrastructure stack for
LLVM. The core ideas are the same that are used throughout the new pass
manager: type erased polymorphism and direct composition. The design is
as follows:
- FunctionAAResults is a type-erasing alias analysis results aggregation
interface to walk a single query across a range of results from
different alias analyses. Currently this is function-specific as we
always assume that aliasing queries are *within* a function.
- AAResultBase is a CRTP utility providing stub implementations of
various parts of the alias analysis result concept, notably in several
cases in terms of other more general parts of the interface. This can
be used to implement only a narrow part of the interface rather than
the entire interface. This isn't really ideal, this logic should be
hoisted into FunctionAAResults as currently it will cause
a significant amount of redundant work, but it faithfully models the
behavior of the prior infrastructure.
- All the alias analysis passes are ported to be wrapper passes for the
legacy PM and new-style analysis passes for the new PM with a shared
result object. In some cases (most notably CFL), this is an extremely
naive approach that we should revisit when we can specialize for the
new pass manager.
- BasicAA has been restructured to reflect that it is much more
fundamentally a function analysis because it uses dominator trees and
loop info that need to be constructed for each function.
All of the references to getting alias analysis results have been
updated to use the new aggregation interface. All the preservation and
other pass management code has been updated accordingly.
The way the FunctionAAResultsWrapperPass works is to detect the
available alias analyses when run, and add them to the results object.
This means that we should be able to continue to respect when various
passes are added to the pipeline, for example adding CFL or adding TBAA
passes should just cause their results to be available and to get folded
into this. The exception to this rule is BasicAA which really needs to
be a function pass due to using dominator trees and loop info. As
a consequence, the FunctionAAResultsWrapperPass directly depends on
BasicAA and always includes it in the aggregation.
This has significant implications for preserving analyses. Generally,
most passes shouldn't bother preserving FunctionAAResultsWrapperPass
because rebuilding the results just updates the set of known AA passes.
The exception to this rule are LoopPass instances which need to preserve
all the function analyses that the loop pass manager will end up
needing. This means preserving both BasicAAWrapperPass and the
aggregating FunctionAAResultsWrapperPass.
Now, when preserving an alias analysis, you do so by directly preserving
that analysis. This is only necessary for non-immutable-pass-provided
alias analyses though, and there are only three of interest: BasicAA,
GlobalsAA (formerly GlobalsModRef), and SCEVAA. Usually BasicAA is
preserved when needed because it (like DominatorTree and LoopInfo) is
marked as a CFG-only pass. I've expanded GlobalsAA into the preserved
set everywhere we previously were preserving all of AliasAnalysis, and
I've added SCEVAA in the intersection of that with where we preserve
SCEV itself.
One significant challenge to all of this is that the CGSCC passes were
actually using the alias analysis implementations by taking advantage of
a pretty amazing set of loop holes in the old pass manager's analysis
management code which allowed analysis groups to slide through in many
cases. Moving away from analysis groups makes this problem much more
obvious. To fix it, I've leveraged the flexibility the design of the new
PM components provides to just directly construct the relevant alias
analyses for the relevant functions in the IPO passes that need them.
This is a bit hacky, but should go away with the new pass manager, and
is already in many ways cleaner than the prior state.
Another significant challenge is that various facilities of the old
alias analysis infrastructure just don't fit any more. The most
significant of these is the alias analysis 'counter' pass. That pass
relied on the ability to snoop on AA queries at different points in the
analysis group chain. Instead, I'm planning to build printing
functionality directly into the aggregation layer. I've not included
that in this patch merely to keep it smaller.
Note that all of this needs a nearly complete rewrite of the AA
documentation. I'm planning to do that, but I'd like to make sure the
new design settles, and to flesh out a bit more of what it looks like in
the new pass manager first.
Differential Revision: http://reviews.llvm.org/D12080
llvm-svn: 247167
Predicating stores requires creating extra blocks. It's much cleaner if we do this in one pass instead of mutating the CFG while writing vector instructions.
Besides which we can make use of helper functions to update domtree for us, reducing the work we need to do.
llvm-svn: 247139
We were bailing to two places if our runtime checks failed. If the initial overflow check failed, we'd go to ScalarPH. If any other check failed, we'd go to MiddleBlock. This caused us to have to have an extra PHI per induction and reduction as the vector loop's exit block was not dominated by its latch.
There's no need to have this behavior - if we just always go to ScalarPH we can get rid of a bunch of complexity.
llvm-svn: 246637
This reduces the complexity of createEmptyBlock() and will open the door to further refactoring.
The test change is simply because we're now constant folding a trivial test.
llvm-svn: 246634
It makes things easier to understand if this is in a helper method. This is part of my ongoing spaghetti-removal operation on createEmptyLoop.
llvm-svn: 246632
There's no need to widen canonical induction variables. It's just as efficient to create a *new*, wide, induction variable.
Consider, if we widen an indvar, then we'll have to truncate it before its uses anyway (1 trunc). If we create a new indvar instead, we'll have to truncate that instead (1 trunc) [besides which IndVars should go and clean up our mess after us anyway on principle].
This lets us remove a ton of special-casing code.
llvm-svn: 246631
Vectorized loops only ever have one induction variable. All induction PHIs from the scalar loop are rewritten to be in terms of this single indvar.
We were trying very hard to pick an indvar that already existed, even if that indvar wasn't canonical (didn't start at zero). But trying so hard is really fruitless - creating a new, canonical, indvar only results in one extra add in the worst case and that add is trivially easy to push through the PHI out of the loop by instcombine.
If we try and be less clever here and instead let instcombine clean up our mess (as we do in many other places in LV), we can remove unneeded complexity.
llvm-svn: 246630
This patch changes the analysis diagnostics produced when loops with
floating-point recurrences or memory operations are identified. The new messages
say "cannot prove it is safe to reorder * operations; allow reordering by
specifying #pragma clang loop vectorize(enable)". Depending on the type of
diagnostic the message will include additional options such as ffast-math or
__restrict__.
This patch also allows the vectorize(enable) pragma to override the low pointer
memory check threshold. When the hint is given a higher threshold is used.
See the clang patch for the options produced for each diagnostic.
llvm-svn: 246187
Unlike scalar operations, we can perform vector operations on element types that
are smaller than the native integer types. We type-promote scalar operations if
they are smaller than a native type (e.g., i8 arithmetic is promoted to i32
arithmetic on Arm targets). This patch detects and removes type-promotions
within the reduction detection framework, enabling the vectorization of small
size reductions.
In the legality phase, we look through the ANDs and extensions that InstCombine
creates during promotion, keeping track of the smaller type. In the
profitability phase, we use the smaller type and ignore the ANDs and extensions
in the cost model. Finally, in the code generation phase, we truncate the result
of the reduction to allow InstCombine to rewrite the entire expression in the
smaller type.
This fixes PR21369.
http://reviews.llvm.org/D12202
Patch by Matt Simpson <mssimpso@codeaurora.org>!
llvm-svn: 246149
... and move it into LoopUtils where it can be used by other passes, just like ReductionDescriptor. The API is very similar to ReductionDescriptor - that is, not very nice at all. Sorting these both out will come in a followup.
NFC
llvm-svn: 246145
This patch ensures that every analysis diagnostic produced by the vectorizer
will be printed if the loop has a vectorization hint on it. The condition has
also been improved to prevent printing when a disabling hint is specified.
llvm-svn: 246132
The loop minimum iterations check below ensures the loop has enough trip count so the generated
vector loop will likely be executed, and it covers the overflow check.
Differential Revision: http://reviews.llvm.org/D12107.
llvm-svn: 245952
and make it always preserve debug locations, since all callers wanted this
behavior anyway.
This is addressing a post-commit review feedback for r245589.
NFC (inside the LLVM tree).
llvm-svn: 245622
Instruction::dropUnknownMetadata(KnownSet) is supposed to preserve all
metadata in KnownSet, but the condition for DebugLocs was inverted.
Most users of dropUnknownMetadata() actually worked around this by not
adding LLVMContext::MD_dbg to their list of KnowIDs.
This is now made explicit.
llvm-svn: 245589
This change makes ScalarEvolution a stand-alone object and just produces
one from a pass as needed. Making this work well requires making the
object movable, using references instead of overwritten pointers in
a number of places, and other refactorings.
I've also wired it up to the new pass manager and added a RUN line to
a test to exercise it under the new pass manager. This includes basic
printing support much like with other analyses.
But there is a big and somewhat scary change here. Prior to this patch
ScalarEvolution was never *actually* invalidated!!! Re-running the pass
just re-wired up the various other analyses and didn't remove any of the
existing entries in the SCEV caches or clear out anything at all. This
might seem OK as everything in SCEV that can uses ValueHandles to track
updates to the values that serve as SCEV keys. However, this still means
that as we ran SCEV over each function in the module, we kept
accumulating more and more SCEVs into the cache. At the end, we would
have a SCEV cache with every value that we ever needed a SCEV for in the
entire module!!! Yowzers. The releaseMemory routine would dump all of
this, but that isn't realy called during normal runs of the pipeline as
far as I can see.
To make matters worse, there *is* actually a key that we don't update
with value handles -- there is a map keyed off of Loop*s. Because
LoopInfo *does* release its memory from run to run, it is entirely
possible to run SCEV over one function, then over another function, and
then lookup a Loop* from the second function but find an entry inserted
for the first function! Ouch.
To make matters still worse, there are plenty of updates that *don't*
trip a value handle. It seems incredibly unlikely that today GVN or
another pass that invalidates SCEV can update values in *just* such
a way that a subsequent run of SCEV will incorrectly find lookups in
a cache, but it is theoretically possible and would be a nightmare to
debug.
With this refactoring, I've fixed all this by actually destroying and
recreating the ScalarEvolution object from run to run. Technically, this
could increase the amount of malloc traffic we see, but then again it is
also technically correct. ;] I don't actually think we're suffering from
tons of malloc traffic from SCEV because if we were, the fact that we
never clear the memory would seem more likely to have come up as an
actual problem before now. So, I've made the simple fix here. If in fact
there are serious issues with too much allocation and deallocation,
I can work on a clever fix that preserves the allocations (while
clearing the data) between each run, but I'd prefer to do that kind of
optimization with a test case / benchmark that shows why we need such
cleverness (and that can test that we actually make it faster). It's
possible that this will make some things faster by making the SCEV
caches have higher locality (due to being significantly smaller) so
until there is a clear benchmark, I think the simple change is best.
Differential Revision: http://reviews.llvm.org/D12063
llvm-svn: 245193
AliasAnalysis.
Same as the other commits, the TLI access from an alias analysis is
going away and isn't very clean -- it is better to explicitly mark the
dependencies.
llvm-svn: 244785
This patch and a relatec clang patch solve the problem of having to explicitly enable analysis when specifying a loop hint pragma to get the diagnostics. Passing AlwasyPrint as the pass name (see below) causes the front-end to print the diagnostic if the user has specified '-Rpass-analysis' without an '=<target-pass>’. Users of loop hints can pass that compiler option without having to specify the pass and they will get diagnostics for only those loops with loop hints.
llvm-svn: 244555
This patch moves checking the threshold of runtime pointer checks to the vectorization requirements (late diagnostics) and emits a diagnostic that infroms the user the loop would be vectorized if not for exceeding the pointer-check threshold. Clang will also append the options that can be used to allow vectorization.
llvm-svn: 244523
This patch moves the verification of fast-math to just before vectorization is done. This way we can tell clang to append the command line options would that allow floating-point commutativity. Specifically those are enableing fast-math or specifying a loop hint.
llvm-svn: 244489
Sometimes interleaving is not beneficial, as determined by the cost-model and sometimes it is disabled by a loop hint (by the user). This patch modifies the diagnostic messages to make it clear why interleaving wasn't done.
llvm-svn: 244485
Summary:
This adds a hook to TTI which enables us to selectively turn on by default
interleaved access vectorization for targets on which we have have performed
the required benchmarking.
Reviewers: rengolin
Subscribers: rengolin, llvm-commits
Differential Revision: http://reviews.llvm.org/D11901
llvm-svn: 244449
Create wrapper methods in the Function class for the OptimizeForSize and MinSize
attributes. We want to hide the logic of "or'ing" them together when optimizing
just for size (-Os).
Currently, we are not consistent about this and rely on a front-end to always set
OptimizeForSize (-Os) if MinSize (-Oz) is on. Thus, there are 18 FIXME changes here
that should be added as follow-on patches with regression tests.
This patch is NFC-intended: it just replaces existing direct accesses of the attributes
by the equivalent wrapper call.
Differential Revision: http://reviews.llvm.org/D11734
llvm-svn: 243994
The patch changes the SLPVectorizer::vectorizeStores to choose the immediate
succeeding or preceding candidate for a store instruction when it has multiple
consecutive candidates. In this way it has better chance to find more slp
vectorization opportunities.
Differential Revision: http://reviews.llvm.org/D10445
llvm-svn: 243666
part of simplifying its interface and usage in preparation for porting
to work with the new pass manager.
Note that this will likely expose that we have dead arguments, members,
and maybe even pass requirements for AA. I'll be cleaning those up in
seperate patches. This just zaps the actual update API.
Differential Revision: http://reviews.llvm.org/D11325
llvm-svn: 242881
change because the diff is *useless*. I assure you, I just switched to
early-return in this function.
Cleanup in preparation for my next commit, as requested in code review!
llvm-svn: 242880
This is useful when we want to do block frequency analysis
conditionally (e.g. only in PGO mode) but don't want to add
one more pass dependence.
Patch by congh.
Approved by dexonsmith.
Differential Revision: http://reviews.llvm.org/D11196
llvm-svn: 242248
I am planning to add more nested classes inside RuntimePointerCheck so
all these triple-nesting would be hard to follow.
Also rename it to RuntimePointerChecking (i.e. append 'ing').
llvm-svn: 242218
Passes should never modify it, just use the const version. While there
reduce copying in LoopInterchange. No functional change intended.
llvm-svn: 242041
The following functions are moved from the LoopVectorizer to VectorUtils:
- getGEPInductionOperand
- stripGetElementPtr
- getUniqueCastUse
- getStrideFromPointer
These used to be static functions in LoopVectorize, but will also be used by
the upcoming loop versioning LICM transformation.
Patch by Ashutosh Nema!
llvm-svn: 241980
Summary:
Following the discussion on r241884, it's more reasonable to assume that a
target has no vector registers by default instead of letting every such
target overrides getNumberOfRegisters.
Therefore, this patch modifies BasicTTIImpl::getNumberOfRegisters to
return 0 when Vector is true, and partially reverts r241884 which
modifies NVPTXTTIImpl::getNumberOfRegisters.
It also fixes a performance bug in LoopVectorizer. Even if a target has
no vector registers, vectorization may still help ILP. So, we need both
checks to be false before disabling loop vectorization all together.
Reviewers: hfinkel
Subscribers: llvm-commits, jholewinski
Differential Revision: http://reviews.llvm.org/D11108
llvm-svn: 241942
Place all code corresponding to a run-time check in one place.
Previously we generated some code, then proceeded to a next check, then
finished the code for the first check (like splitting blocks and
generating branches). Now the code for generating a check is
self-contained.
llvm-svn: 241741
This is mostly an NFC, which increases code readability (instead of
saving old terminator, generating new one in front of old, and deleting
old, we just call a function). However, it would additionaly copy
the debug location from old instruction to replacement, which
would help PR23837.
llvm-svn: 241197
If we are dealing with a pointer induction variable, isInductionPHI
gives back a step value of Stride / size of pointer. However, we might
be indexing with a legal type wider than the pointer width.
Handle this by inserting casts where appropriate instead of crashing.
This fixes PR23954.
llvm-svn: 240877
With option OptForSize enabled, the Loop Vectorizer is not supposed to
create tail loop. The condition checking that was invalid and was not
matching to the comment above.
Patch by Marianne Mailhot-Sarrasin.
llvm-svn: 240556
The patch is generated using this command:
tools/clang/tools/extra/clang-tidy/tool/run-clang-tidy.py -fix \
-checks=-*,llvm-namespace-comment -header-filter='llvm/.*|clang/.*' \
llvm/lib/
Thanks to Eugene Kosov for the original patch!
llvm-svn: 240137
that it is its own entity in the form of MemoryLocation, and update all
the callers.
This is an entirely mechanical change. References to "Location" within
AA subclases become "MemoryLocation", and elsewhere
"AliasAnalysis::Location" becomes "MemoryLocation". Hope that helps
out-of-tree folks update.
llvm-svn: 239885
A reduction is a special kind of recurrence. In the loop vectorizer we currently
identify basic reductions. Future patches will extend this to identifying basic
recurrences.
llvm-svn: 239835
Interleaved memory accesses are grouped and vectorized into vector load/store and shufflevector.
E.g. for (i = 0; i < N; i+=2) {
a = A[i]; // load of even element
b = A[i+1]; // load of odd element
... // operations on a, b, c, d
A[i] = c; // store of even element
A[i+1] = d; // store of odd element
}
The loads of even and odd elements are identified as an interleave load group, which will be transfered into vectorized IRs like:
%wide.vec = load <8 x i32>, <8 x i32>* %ptr
%vec.even = shufflevector <8 x i32> %wide.vec, <8 x i32> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
%vec.odd = shufflevector <8 x i32> %wide.vec, <8 x i32> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
The stores of even and odd elements are identified as an interleave store group, which will be transfered into vectorized IRs like:
%interleaved.vec = shufflevector <4 x i32> %vec.even, %vec.odd, <8 x i32> <i32 0, i32 4, i32 1, i32 5, i32 2, i32 6, i32 3, i32 7>
store <8 x i32> %interleaved.vec, <8 x i32>* %ptr
This optimization is currently disabled by defaut. To try it by adding '-enable-interleaved-mem-accesses=true'.
llvm-svn: 239291
port it to the new pass manager.
All this does is extract the inner "location" class used by AA into its
own full fledged type. This seems *much* cleaner as MemoryDependence and
soon MemorySSA also use this heavily, and it doesn't make much sense
being inside the AA infrastructure.
This will also make it much easier to break apart the AA infrastructure
into something that stands on its own rather than using the analysis
group design.
There are a few places where this makes APIs not make sense -- they were
taking an AliasAnalysis pointer just to build locations. I'll try to
clean those up in follow-up commits.
Differential Revision: http://reviews.llvm.org/D10228
llvm-svn: 239003
If the type isn't trivially moveable emplace can skip a potentially
expensive move. It also saves a couple of characters.
Call sites were found with the ASTMatcher + some semi-automated cleanup.
memberCallExpr(
argumentCountIs(1), callee(methodDecl(hasName("push_back"))),
on(hasType(recordDecl(has(namedDecl(hasName("emplace_back")))))),
hasArgument(0, bindTemporaryExpr(
hasType(recordDecl(hasNonTrivialDestructor())),
has(constructExpr()))),
unless(isInTemplateInstantiation()))
No functional change intended.
llvm-svn: 238602
Now that Intrinsic::ID is a typed enum, we can forward declare it and so return it from this method.
This updates all users which were either using an unsigned to store it, or had a now unnecessary cast.
llvm-svn: 237810
The patch disabled unrolling in loop vectorization pass when VF==1 on x86 architecture,
by setting MaxInterleaveFactor to 1. Unrolling in loop vectorization pass may introduce
the cost of overflow check, memory boundary check and extra prologue/epilogue code when
regular unroller will unroll the loop another time. Disable it when VF==1 remove the
unnecessary cost on x86. The same can be done for other platforms after verifying
interleaving/memory bound checking to be not perf critical on those platforms.
Differential Revision: http://reviews.llvm.org/D9515
llvm-svn: 236613
(reverted in r235533)
Original commit message:
"Calls to llvm::Value::mutateType are becoming extra-sensitive now that
instructions have extra type information that will not be derived from
operands or result type (alloca, gep, load, call/invoke, etc... ). The
special-handling for mutateType will get more complicated as this work
continues - it might be worth making mutateType virtual & pushing the
complexity down into the classes that need special handling. But with
only two significant uses of mutateType (vectorization and linking) this
seems OK for now.
Totally open to ideas/suggestions/improvements, of course.
With this, and a bunch of exceptions, we can roundtrip an indirect call
site through bitcode and IR. (a direct call site is actually trickier...
I haven't figured out how to deal with the IR deserializer's lazy
construction of Function/GlobalVariable decl's based on the type of the
entity which means looking through the "pointer to T" type referring to
the global)"
The remapping done in ValueMapper for LTO was insufficient as the types
weren't correctly mapped (though I was using the post-mapped operands,
some of those operands might not have been mapped yet so the type
wouldn't be post-mapped yet). Instead use the pre-mapped type and
explicitly map all the types.
llvm-svn: 235651
This patch refactors the definition of common utility function "isInductionPHI" to LoopUtils.cpp.
This fixes compilation error when configured with -DBUILD_SHARED_LIBS=ON
llvm-svn: 235577
This reverts commit r235458.
It looks like this might be breaking something LTO-ish. Looking into it
& will recommit with a fix/test case/etc once I've got more to go on.
llvm-svn: 235533
Calls to llvm::Value::mutateType are becoming extra-sensitive now that
instructions have extra type information that will not be derived from
operands or result type (alloca, gep, load, call/invoke, etc... ). The
special-handling for mutateType will get more complicated as this work
continues - it might be worth making mutateType virtual & pushing the
complexity down into the classes that need special handling. But with
only two significant uses of mutateType (vectorization and linking) this
seems OK for now.
Totally open to ideas/suggestions/improvements, of course.
With this, and a bunch of exceptions, we can roundtrip an indirect call
site through bitcode and IR. (a direct call site is actually trickier...
I haven't figured out how to deal with the IR deserializer's lazy
construction of Function/GlobalVariable decl's based on the type of the
entity which means looking through the "pointer to T" type referring to
the global)
llvm-svn: 235458
This patch refactors reduction identification code out of LoopVectorizer and
exposes them as common utilities.
No functional change.
Review: http://reviews.llvm.org/D9046
llvm-svn: 235284
(Re-apply r234361 with a fix and a testcase for PR23157)
Both run-time pointer checking and the dependence analysis are capable
of dealing with uniform addresses. I.e. it's really just an orthogonal
property of the loop that the analysis computes.
Run-time pointer checking will only try to reason about SCEVAddRec
pointers or else gives up. If the uniform pointer turns out the be a
SCEVAddRec in an outer loop, the run-time checks generated will be
correct (start and end bounds would be equal).
In case of the dependence analysis, we work again with SCEVs. When
compared against a loop-dependent address of the same underlying object,
the difference of the two SCEVs won't be constant. This will result in
returning an Unknown dependence for the pair.
When compared against another uniform access, the difference would be
constant and we should return the right type of dependence
(forward/backward/etc).
The changes also adds support to query this property of the loop and
modify the vectorizer to use this.
Patch by Ashutosh Nema!
llvm-svn: 234424
Both run-time pointer checking and the dependence analysis are capable
of dealing with uniform addresses. I.e. it's really just an orthogonal
property of the loop that the analysis computes.
Run-time pointer checking will only try to reason about SCEVAddRec
pointers or else gives up. If the uniform pointer turns out the be a
SCEVAddRec in an outer loop, the run-time checks generated will be
correct (start and end bounds would be equal).
In case of the dependence analysis, we work again with SCEVs. When
compared against a loop-dependent address of the same underlying object,
the difference of the two SCEVs won't be constant. This will result in
returning an Unknown dependence for the pair.
When compared against another uniform access, the difference would be
constant and we should return the right type of dependence
(forward/backward/etc).
The changes also adds support to query this property of the loop and
modify the vectorizer to use this.
Patch by Ashutosh Nema!
llvm-svn: 234361
The plan here is to push the API changes out from the common components
(like Constant::getGetElementPtr and IRBuilder::CreateGEP related
functions) and just update callers to either pass the type if it's
obvious, or pass null.
Do this with LoadInst as well and anything else that comes up, then to
start porting specific uses to not pass null anymore - this may require
some refactoring in each case.
llvm-svn: 234042
The changes to InstCombine do seem a bit silly - it doesn't make
anything obviously better to have the caller access the pointers element
type (the thing I'm trying to remove) than the GEP itself, but it's a
helpful migration step. This will allow me to more obviously lock down
GEP (& Load, etc) API usage, then fix all the code that accesses pointer
element types except the places that need to be removed (most of the
InstCombines) anyway - at which point I'll need to just remove all that
code because it won't be meaningful anymore (there will be no pointer
types, so no bitcasts to combine)
llvm-svn: 233126
Now the analysis won't "fail" if the memchecks exceed the threshold. It
is the transform pass' responsibility to perform the check.
This allows the transform pass to further analyze/eliminate the
memchecks. E.g. in Loop distribution we only need to check pointers
that end up in different partitions.
Note that there is a slight change of functionality here. The logic in
analyzeLoop is that if dependence checking fails due to non-constant
distance between the pointers, another attempt is made to prove safety
of the dependences purely using run-time checks.
Before this patch we could fail the loop due to exceeding the memcheck
threshold after the first step, now we only check the threshold in the
client after the full analysis. There is no measurable compile-time
effect but I wanted to record this here.
llvm-svn: 231817
Summary:
Now that the DataLayout is a mandatory part of the module, let's start
cleaning the codebase. This patch is a first attempt at doing that.
This patch is not exactly NFC as for instance some places were passing
a nullptr instead of the DataLayout, possibly just because there was a
default value on the DataLayout argument to many functions in the API.
Even though it is not purely NFC, there is no change in the
validation.
I turned as many pointer to DataLayout to references, this helped
figuring out all the places where a nullptr could come up.
I had initially a local version of this patch broken into over 30
independant, commits but some later commit were cleaning the API and
touching part of the code modified in the previous commits, so it
seemed cleaner without the intermediate state.
Test Plan:
Reviewers: echristo
Subscribers: llvm-commits
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 231740
Runtime unrolling is an expensive optimization which can bring benefit
only if the loop is hot and iteration number is relatively large enough.
For some loops, we know they are not worth to be runtime unrolled.
The scalar loop from vectorization is one of the cases.
llvm-svn: 231631
Summary:
DataLayout keeps the string used for its creation.
As a side effect it is no longer needed in the Module.
This is "almost" NFC, the string is no longer
canonicalized, you can't rely on two "equals" DataLayout
having the same string returned by getStringRepresentation().
Get rid of DataLayoutPass: the DataLayout is in the Module
The DataLayout is "per-module", let's enforce this by not
duplicating it more than necessary.
One more step toward non-optionality of the DataLayout in the
module.
Make DataLayout Non-Optional in the Module
Module->getDataLayout() will never returns nullptr anymore.
Reviewers: echristo
Subscribers: resistor, llvm-commits, jholewinski
Differential Revision: http://reviews.llvm.org/D7992
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 231270
The only difference between these two is that VectorizerReport adds a
vectorizer-specific prefix to its messages. When LAA is used in the
vectorizer context the prefix is added when we promote the
LoopAccessReport into a VectorizerReport via one of the constructors.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229897
When I split out LoopAccessReport from this, I need to create some temps
so constness becomes necessary.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229896
Also add pass name as an argument to VectorizationReport::emitAnalysis.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229894
This is a function pass that runs the analysis on demand. The analysis
can be initiated by querying the loop access info via LAA::getInfo. It
either returns the cached info or runs the analysis.
Symbolic stride information continues to reside outside of this analysis
pass. We may move it inside later but it's not a priority for me right
now. The idea is that Loop Distribution won't support run-time stride
checking at least initially.
This means that when querying the analysis, symbolic stride information
can be provided optionally. Whether stride information is used can
invalidate the cache entry and rerun the analysis. Note that if the
loop does not have any symbolic stride, the entry should be preserved
across Loop Distribution and LV.
Since currently the only user of the pass is LV, I just check that the
symbolic stride information didn't change when using a cached result.
On the LV side, LoopVectorizationLegality requests the info object
corresponding to the loop from the analysis pass. A large chunk of the
diff is due to LAI becoming a pointer from a reference.
A test will be added as part of the -analyze patch.
Also tested that with AVX, we generate identical assembly output for the
testsuite (including the external testsuite) before and after.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229893
LAA will be an on-demand analysis pass, so we need to cache the result
of the analysis. canVectorizeMemory is renamed to analyzeLoop which
computes the result. canVectorizeMemory becomes the query function for
the cached result.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229892
The transformation passes will query this and then emit them as part of
their own report. The currently only user LV is modified to do just
that.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229891
As LAA is becoming a pass, we can no longer pass the params to its
constructor. This changes the command line flags to have external
storage. These can now be accessed both from LV and LAA.
VectorizerParams is moved out of LoopAccessInfo in order to shorten the
code to access it.
This commits also has the fix (D7731) to the break dependence cycle
between the analysis and vector libraries.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229890
This reverts commit r229651.
I'd like to ultimately revert r229650 but this reformat stands in the
way. I'll reformat the affected files once the the loop-access pass is
fully committed.
llvm-svn: 229889
r229622: "[LoopAccesses] Make VectorizerParams global"
r229623: "[LoopAccesses] Stash the report from the analysis rather than emitting it"
r229624: "[LoopAccesses] Cache the result of canVectorizeMemory"
r229626: "[LoopAccesses] Create the analysis pass"
r229628: "[LoopAccesses] Change debug messages from LV to LAA"
r229630: "[LoopAccesses] Add canAnalyzeLoop"
r229631: "[LoopAccesses] Add missing const to APIs in VectorizationReport"
r229632: "[LoopAccesses] Split out LoopAccessReport from VectorizerReport"
r229633: "[LoopAccesses] Add -analyze support"
r229634: "[LoopAccesses] Change LAA:getInfo to return a constant reference"
r229638: "Analysis: fix buildbots"
llvm-svn: 229650
The only difference between these two is that VectorizerReport adds a
vectorizer-specific prefix to its messages. When LAA is used in the
vectorizer context the prefix is added when we promote the
LoopAccessReport into a VectorizerReport via one of the constructors.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229632
When I split out LoopAccessReport from this, I need to create some temps
so constness becomes necessary.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229631
Also add pass name as an argument to VectorizationReport::emitAnalysis.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229628
This is a function pass that runs the analysis on demand. The analysis
can be initiated by querying the loop access info via LAA::getInfo. It
either returns the cached info or runs the analysis.
Symbolic stride information continues to reside outside of this analysis
pass. We may move it inside later but it's not a priority for me right
now. The idea is that Loop Distribution won't support run-time stride
checking at least initially.
This means that when querying the analysis, symbolic stride information
can be provided optionally. Whether stride information is used can
invalidate the cache entry and rerun the analysis. Note that if the
loop does not have any symbolic stride, the entry should be preserved
across Loop Distribution and LV.
Since currently the only user of the pass is LV, I just check that the
symbolic stride information didn't change when using a cached result.
On the LV side, LoopVectorizationLegality requests the info object
corresponding to the loop from the analysis pass. A large chunk of the
diff is due to LAI becoming a pointer from a reference.
A test will be added as part of the -analyze patch.
Also tested that with AVX, we generate identical assembly output for the
testsuite (including the external testsuite) before and after.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229626
blockNeedsPredication is in LoopAccess in order to share it with the
vectorizer. It's a utility needed by LoopAccess not strictly provided
by it but it's a good place to share it. This makes the function static
so that it no longer required to create an LoopAccessInfo instance in
order to access it from LV.
This was actually causing problems because it would have required
creating LAI much earlier that LV::canVectorizeMemory().
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229625
LAA will be an on-demand analysis pass, so we need to cache the result
of the analysis. canVectorizeMemory is renamed to analyzeLoop which
computes the result. canVectorizeMemory becomes the query function for
the cached result.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229624
The transformation passes will query this and then emit them as part of
their own report. The currently only user LV is modified to do just
that.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229623
As LAA is becoming a pass, we can no longer pass the params to its
constructor. This changes the command line flags to have external
storage. These can now be accessed both from LV and LAA.
VectorizerParams is moved out of LoopAccessInfo in order to shorten the
code to access it.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229622
LoopAccessAnalysis will be used as the name of the pass.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229621
Canonicalize access to function attributes to use the simpler API.
getAttributes().getAttribute(AttributeSet::FunctionIndex, Kind)
=> getFnAttribute(Kind)
getAttributes().hasAttribute(AttributeSet::FunctionIndex, Kind)
=> hasFnAttribute(Kind)
llvm-svn: 229202
LLVM's include tree and the use of using declarations to hide the
'legacy' namespace for the old pass manager.
This undoes the primary modules-hostile change I made to keep
out-of-tree targets building. I sent an email inquiring about whether
this would be reasonable to do at this phase and people seemed fine with
it, so making it a reality. This should allow us to start bootstrapping
with modules to a certain extent along with making it easier to mix and
match headers in general.
The updates to any code for users of LLVM are very mechanical. Switch
from including "llvm/PassManager.h" to "llvm/IR/LegacyPassManager.h".
Qualify the types which now produce compile errors with "legacy::". The
most common ones are "PassManager", "PassManagerBase", and
"FunctionPassManager".
llvm-svn: 229094
Apparently some code finally started to tickle this after my
canonicalization changes to instcombine.
The bug stems from trying to form a vector type out of scalars that
aren't compatible at all. In this example, from x86_mmx values. The code
in the vectorizer that checks for reasonable types whas checking for
aggregates or vectors, but there are lots of other types that should
just never reach the vectorizer.
Debugging this was made more confusing by the lie in an assert in
VectorType::get() -- it isn't that the types are *primitive*. The types
must be integer, pointer, or floating point types. No other types are
allowed.
I've improved the assert and added a helper to the vectorizer to handle
the element type validity checks. It now re-uses the VectorType static
function and then further excludes weird target-specific types that we
probably shouldn't be touching here (x86_fp80 and ppc_fp128). Neither of
these are really reachable anyways (neither 80-bit nor 128-bit things
will get vectorized) but it seems better to just eagerly exclude such
nonesense.
I've added a test case, but while it definitely covers two of the paths
through this code there may be more paths that would benefit from test
coverage. I'm not familiar enough with the SLP vectorizer to synthesize
test cases for all of these, but was able to update the code itself by
inspection.
llvm-svn: 228899
This allows IDEs to recognize the entire set of header files for
each of the core LLVM projects.
Differential Revision: http://reviews.llvm.org/D7526
Reviewed By: Chris Bieneman
llvm-svn: 228798
Summary:
The alias.scope metadata represents sets of things an instruction might
alias with. When generically combining the metadata from two
instructions the result must be the union of the original sets, because
the new instruction might alias with anything any of the original
instructions aliased with.
Reviewers: hfinkel
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D7490
llvm-svn: 228525
This will allow it to be shared with the new Loop Distribution pass.
getFirstInst is currently duplicated across LoopVectorize.cpp and
LoopAccessAnalysis.cpp. This is a short-term work-around until we figure out
a better solution.
NFC. (The code moved is adjusted a bit for the name of the Loop member and
that PtrRtCheck is now a reference rather than a pointer.)
llvm-svn: 228418
I've noticed this while trying to move addRuntimeCheck to LoopAccessAnalysis.
I think that the intention was to early exit from the overflow checking before
the code for the memchecks. This is the entire reason why we compute
FirstCheckInst but then we don't use that as the splitting instruction but the
final check. Looks like an oversight.
llvm-svn: 228056
The commit r225977 uncovered this bug. The problem was that the vectorizer tried to
read the second operand of an already deleted instruction.
The bug didn't show up before r225977 because the freed memory still contained a non-null pointer.
With r225977 deletion of instructions is delayed and the read operand pointer is always null.
llvm-svn: 227800
Other than moving code and adding the boilerplate for the new files, the code
being moved is unchanged.
There are a few global functions that are shared with the rest of the
LoopVectorizer. I moved these to the new module as well (emitLoopAnalysis,
stripIntegerCast, replaceSymbolicStrideSCEV) along with the Report class used
by emitLoopAnalysis. There is probably room for further improvement in this
area.
I kept DEBUG_TYPE "loop-vectorize" because it's used as the PassName with
emitOptimizationRemarkAnalysis. This will obviously have to change.
NFC. This is part of the patchset that splits out the memory dependence logic
from LoopVectorizationLegality into a new class LoopAccessAnalysis.
LoopAccessAnalysis will be used by the new Loop Distribution pass.
llvm-svn: 227756
This class needs to remain public because it's used by
LoopVectorizationLegality::addRuntimeCheck.
NFC. This is part of the patchset that splits out the memory dependence logic
from LoopVectorizationLegality into a new class LoopAccessAnalysis.
LoopAccessAnalysis will be used by the new Loop Distribution pass.
llvm-svn: 227755
Rather than using globals use a structure to pass parameters from the
vectorizer. This prepares the class to be moved outside the LoopVectorizer.
It's not great how all this is passed through in LoopAccessAnalysis but this
is all expected to change once the class start servicing the Loop Distribution
pass as well where some of these parameters make no sense.
NFC. This is part of the patchset that splits out the memory dependence logic
from LoopVectorizationLegality into a new class LoopAccessAnalysis.
LoopAccessAnalysis will be used by the new Loop Distribution pass.
llvm-svn: 227754
Move the canVectorizeMemory functionality from LoopVectorizationLegality to a
new class LoopAccessAnalysis and forward users.
Currently the collection of the symbolic stride information is kept with
LoopVectorizationLegality and it becomes an input to LoopAccessAnalysis.
NFC. This is part of the patchset that splits out the memory dependence logic
from LoopVectorizationLegality into a new class LoopAccessAnalysis.
LoopAccessAnalysis will be used by the new Loop Distribution pass.
llvm-svn: 227751
These members are moving to LoopAccessAnalysis. The accessors help to hide
this.
NFC. This is part of the patchset that splits out the memory dependence logic
from LoopVectorizationLegality into a new class LoopAccessAnalysis.
LoopAccessAnalysis will be used by the new Loop Distribution pass.
llvm-svn: 227750
This class will become public in the new LoopAccessAnalysis header so the name
needs to be more global.
NFC. This is part of the patchset that splits out the memory dependence logic
from LoopVectorizationLegality into a new class LoopAccessAnalysis.
LoopAccessAnalysis will be used by the new Loop Distribution pass.
llvm-svn: 227749
The logic in emitAnalysis is duplicated across multiple functions. This
splits it into a function. Another use will be added by the patchset.
NFC. This is part of the patchset that splits out the memory dependence logic
from LoopVectorizationLegality into a new class LoopAccessAnalysis.
LoopAccessAnalysis will be used by the new Loop Distribution pass.
llvm-svn: 227748
RuntimePointerCheck will be used through LoopAccessAnalysis in
LoopVectorizationLegality. Later in the patchset it will become a local class
of LoopAccessAnalysis.
NFC. This is part of the patchset that splits out the memory dependence logic
from LoopVectorizationLegality into a new class LoopAccessAnalysis.
LoopAccessAnalysis will be used by the new Loop Distribution pass.
llvm-svn: 227747
getTTI method used to get an actual TTI object.
No functionality changed. This just threads the argument and ensures
code like the inliner can correctly look up the callee's TTI rather than
using a fixed one.
The next change will use this to implement per-function subtarget usage
by TTI. The changes after that should eliminate the need for FTTI as that
will have become the default.
llvm-svn: 227730
type erased interface and a single analysis pass rather than an
extremely complex analysis group.
The end result is that the TTI analysis can contain a type erased
implementation that supports the polymorphic TTI interface. We can build
one from a target-specific implementation or from a dummy one in the IR.
I've also factored all of the code into "mix-in"-able base classes,
including CRTP base classes to facilitate calling back up to the most
specialized form when delegating horizontally across the surface. These
aren't as clean as I would like and I'm planning to work on cleaning
some of this up, but I wanted to start by putting into the right form.
There are a number of reasons for this change, and this particular
design. The first and foremost reason is that an analysis group is
complete overkill, and the chaining delegation strategy was so opaque,
confusing, and high overhead that TTI was suffering greatly for it.
Several of the TTI functions had failed to be implemented in all places
because of the chaining-based delegation making there be no checking of
this. A few other functions were implemented with incorrect delegation.
The message to me was very clear working on this -- the delegation and
analysis group structure was too confusing to be useful here.
The other reason of course is that this is *much* more natural fit for
the new pass manager. This will lay the ground work for a type-erased
per-function info object that can look up the correct subtarget and even
cache it.
Yet another benefit is that this will significantly simplify the
interaction of the pass managers and the TargetMachine. See the future
work below.
The downside of this change is that it is very, very verbose. I'm going
to work to improve that, but it is somewhat an implementation necessity
in C++ to do type erasure. =/ I discussed this design really extensively
with Eric and Hal prior to going down this path, and afterward showed
them the result. No one was really thrilled with it, but there doesn't
seem to be a substantially better alternative. Using a base class and
virtual method dispatch would make the code much shorter, but as
discussed in the update to the programmer's manual and elsewhere,
a polymorphic interface feels like the more principled approach even if
this is perhaps the least compelling example of it. ;]
Ultimately, there is still a lot more to be done here, but this was the
huge chunk that I couldn't really split things out of because this was
the interface change to TTI. I've tried to minimize all the other parts
of this. The follow up work should include at least:
1) Improving the TargetMachine interface by having it directly return
a TTI object. Because we have a non-pass object with value semantics
and an internal type erasure mechanism, we can narrow the interface
of the TargetMachine to *just* do what we need: build and return
a TTI object that we can then insert into the pass pipeline.
2) Make the TTI object be fully specialized for a particular function.
This will include splitting off a minimal form of it which is
sufficient for the inliner and the old pass manager.
3) Add a new pass manager analysis which produces TTI objects from the
target machine for each function. This may actually be done as part
of #2 in order to use the new analysis to implement #2.
4) Work on narrowing the API between TTI and the targets so that it is
easier to understand and less verbose to type erase.
5) Work on narrowing the API between TTI and its clients so that it is
easier to understand and less verbose to forward.
6) Try to improve the CRTP-based delegation. I feel like this code is
just a bit messy and exacerbating the complexity of implementing
the TTI in each target.
Many thanks to Eric and Hal for their help here. I ended up blocked on
this somewhat more abruptly than I expected, and so I appreciate getting
it sorted out very quickly.
Differential Revision: http://reviews.llvm.org/D7293
llvm-svn: 227669
Previously, only -1 and +1 step values are supported for induction variables. This patch extends LV to support
arbitrary constant steps.
Initial patch by Alexey Volkov. Some bug fixes are added in the following version.
Differential Revision: http://reviews.llvm.org/D6051 and http://reviews.llvm.org/D7193
llvm-svn: 227557
Even with the current limit on the number of alias checks, the containing loop has quadratic complexity.
This begins to hurt for blocks containing > 1K load/store instructions.
This commit introduces a limit for the loop count. It reduces the runtime for such very large blocks.
llvm-svn: 226792
This patch fixes 2 issues in reorderInputsAccordingToOpcode
1) AllSameOpcodeLeft and AllSameOpcodeRight was being calculated incorrectly resulting in code not being vectorized in few cases.
2) Adds logic to reorder operands if we get longer chain of consecutive loads enabling vectorization. Handled the same for cases were we have AltOpcode.
Thanks Michael for inputs and review.
Review: http://reviews.llvm.org/D6677
llvm-svn: 226547
In case of blocks with many memory-accessing instructions, alias checking can take lot of time
(because calculating the memory dependencies has quadratic complexity).
I chose a limit which resulted in no changes when running the benchmarks.
llvm-svn: 226439
cleaner to derive from the generic base.
Thise removes a ton of boiler plate code and somewhat strange and
pointless indirections. It also remove a bunch of the previously needed
friend declarations. To fully remove these, I also lifted the verify
logic into the generic LoopInfoBase, which seems good anyways -- it is
generic and useful logic even for the machine side.
llvm-svn: 226385
a LoopInfoWrapperPass to wire the object up to the legacy pass manager.
This switches all the clients of LoopInfo over and paves the way to port
LoopInfo to the new pass manager. No functionality change is intended
with this iteration.
llvm-svn: 226373
This patch was generated by a clang tidy checker that is being open sourced.
The documentation of that checker is the following:
/// The emptiness of a container should be checked using the empty method
/// instead of the size method. It is not guaranteed that size is a
/// constant-time function, and it is generally more efficient and also shows
/// clearer intent to use empty. Furthermore some containers may implement the
/// empty method but not implement the size method. Using empty whenever
/// possible makes it easier to switch to another container in the future.
Patch by Gábor Horváth!
llvm-svn: 226161
The pass is really just a means of accessing a cached instance of the
TargetLibraryInfo object, and this way we can re-use that object for the
new pass manager as its result.
Lots of delta, but nothing interesting happening here. This is the
common pattern that is developing to allow analyses to live in both the
old and new pass manager -- a wrapper pass in the old pass manager
emulates the separation intrinsic to the new pass manager between the
result and pass for analyses.
llvm-svn: 226157
This speeds up the dependency calculations for blocks with many load/store/call instructions.
Beside the improved runtime, there is no functional change.
Compared to the original commit, this re-applied commit contains a bug fix which ensures that there are
no incorrect collisions in the alias cache.
llvm-svn: 225977
The issue was introduced in r214638:
+ for (auto &BSIter : BlocksSchedules) {
+ scheduleBlock(BSIter.second.get());
+ }
Because BlocksSchedules is a DenseMap with BasicBlock* keys, blocks are
scheduled in non-deterministic order, resulting in unpredictable IR.
Patch by Daniel Reynaud!
llvm-svn: 225821
The alias cache has a problem of incorrect collisions in case a new instruction is allocated at the same address as a previously deleted instruction.
llvm-svn: 225790
This speeds up the dependency calculations for blocks with many load/store/call instructions.
Beside the improved runtime, there is no functional change.
llvm-svn: 225786
{code}
// loop body
... = a[i] (1)
... = a[i+1] (2)
.......
a[i+1] = .... (3)
a[i] = ... (4)
{code}
The algorithm tries to collect memory access candidates from AliasSetTracker, and then check memory dependences one another. The memory accesses are unique in AliasSetTracker, and a single memory access in AliasSetTracker may map to multiple entries in AccessAnalysis, which could cover both 'read' and 'write'. Originally the algorithm only checked 'write' entry in Accesses if only 'write' exists. This is incorrect and the consequence is it ignored all read access, and finally some RAW and WAR dependence are missed.
For the case given above, if we ignore two reads, the dependence between (1) and (3) would not be able to be captured, and finally this loop will be incorrectly vectorized.
The fix simply inserts a new loop to find all entries in Accesses. Since it will skip most of all other memory accesses by checking the Value pointer at the very beginning of the loop, it should not increase compile-time visibly.
llvm-svn: 225159
a cache of assumptions for a single function, and an immutable pass that
manages those caches.
The motivation for this change is two fold. Immutable analyses are
really hacks around the current pass manager design and don't exist in
the new design. This is usually OK, but it requires that the core logic
of an immutable pass be reasonably partitioned off from the pass logic.
This change does precisely that. As a consequence it also paves the way
for the *many* utility functions that deal in the assumptions to live in
both pass manager worlds by creating an separate non-pass object with
its own independent API that they all rely on. Now, the only bits of the
system that deal with the actual pass mechanics are those that actually
need to deal with the pass mechanics.
Once this separation is made, several simplifications become pretty
obvious in the assumption cache itself. Rather than using a set and
callback value handles, it can just be a vector of weak value handles.
The callers can easily skip the handles that are null, and eventually we
can wrap all of this up behind a filter iterator.
For now, this adds boiler plate to the various passes, but this kind of
boiler plate will end up making it possible to port these passes to the
new pass manager, and so it will end up factored away pretty reasonably.
llvm-svn: 225131
The loop vectorizer optimizes loops containing conditional memory
accesses by generating masked load and store intrinsics.
This decision is target dependent.
http://reviews.llvm.org/D6527
llvm-svn: 224334
Split `Metadata` away from the `Value` class hierarchy, as part of
PR21532. Assembly and bitcode changes are in the wings, but this is the
bulk of the change for the IR C++ API.
I have a follow-up patch prepared for `clang`. If this breaks other
sub-projects, I apologize in advance :(. Help me compile it on Darwin
I'll try to fix it. FWIW, the errors should be easy to fix, so it may
be simpler to just fix it yourself.
This breaks the build for all metadata-related code that's out-of-tree.
Rest assured the transition is mechanical and the compiler should catch
almost all of the problems.
Here's a quick guide for updating your code:
- `Metadata` is the root of a class hierarchy with three main classes:
`MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from
the `Value` class hierarchy. It is typeless -- i.e., instances do
*not* have a `Type`.
- `MDNode`'s operands are all `Metadata *` (instead of `Value *`).
- `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be
replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively.
If you're referring solely to resolved `MDNode`s -- post graph
construction -- just use `MDNode*`.
- `MDNode` (and the rest of `Metadata`) have only limited support for
`replaceAllUsesWith()`.
As long as an `MDNode` is pointing at a forward declaration -- the
result of `MDNode::getTemporary()` -- it maintains a side map of its
uses and can RAUW itself. Once the forward declarations are fully
resolved RAUW support is dropped on the ground. This means that
uniquing collisions on changing operands cause nodes to become
"distinct". (This already happened fairly commonly, whenever an
operand went to null.)
If you're constructing complex (non self-reference) `MDNode` cycles,
you need to call `MDNode::resolveCycles()` on each node (or on a
top-level node that somehow references all of the nodes). Also,
don't do that. Metadata cycles (and the RAUW machinery needed to
construct them) are expensive.
- An `MDNode` can only refer to a `Constant` through a bridge called
`ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`).
As a side effect, accessing an operand of an `MDNode` that is known
to be, e.g., `ConstantInt`, takes three steps: first, cast from
`Metadata` to `ConstantAsMetadata`; second, extract the `Constant`;
third, cast down to `ConstantInt`.
The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have
metadata schema owners transition away from using `Constant`s when
the type isn't important (and they don't care about referring to
`GlobalValue`s).
In the meantime, I've added transitional API to the `mdconst`
namespace that matches semantics with the old code, in order to
avoid adding the error-prone three-step equivalent to every call
site. If your old code was:
MDNode *N = foo();
bar(isa <ConstantInt>(N->getOperand(0)));
baz(cast <ConstantInt>(N->getOperand(1)));
bak(cast_or_null <ConstantInt>(N->getOperand(2)));
bat(dyn_cast <ConstantInt>(N->getOperand(3)));
bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4)));
you can trivially match its semantics with:
MDNode *N = foo();
bar(mdconst::hasa <ConstantInt>(N->getOperand(0)));
baz(mdconst::extract <ConstantInt>(N->getOperand(1)));
bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2)));
bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3)));
bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4)));
and when you transition your metadata schema to `MDInt`:
MDNode *N = foo();
bar(isa <MDInt>(N->getOperand(0)));
baz(cast <MDInt>(N->getOperand(1)));
bak(cast_or_null <MDInt>(N->getOperand(2)));
bat(dyn_cast <MDInt>(N->getOperand(3)));
bay(dyn_cast_or_null<MDInt>(N->getOperand(4)));
- A `CallInst` -- specifically, intrinsic instructions -- can refer to
metadata through a bridge called `MetadataAsValue`. This is a
subclass of `Value` where `getType()->isMetadataTy()`.
`MetadataAsValue` is the *only* class that can legally refer to a
`LocalAsMetadata`, which is a bridged form of non-`Constant` values
like `Argument` and `Instruction`. It can also refer to any other
`Metadata` subclass.
(I'll break all your testcases in a follow-up commit, when I propagate
this change to assembly.)
llvm-svn: 223802
Remove an unnecessary `MDNode::replaceAllUsesWith()`. In the preceding
line, `TheLoop->setLoopID()` visits all backedges and sets the new loop
ID. This sufficiently updates the loop metadata.
Metadata RAUW is going away as part of PR21532.
llvm-svn: 223210
This reverts commit r222632 (and follow-up r222636), which caused a host
of LNT failures on an internal bot. I'll respond to the commit on the
list with a reproduction of one of the failures.
Conflicts:
lib/Target/X86/X86TargetTransformInfo.cpp
llvm-svn: 222936
Introduced new target-independent intrinsics in order to support masked vector loads and stores. The loop vectorizer optimizes loops containing conditional memory accesses by generating these intrinsics for existing targets AVX2 and AVX-512. The vectorizer asks the target about availability of masked vector loads and stores.
Added SDNodes for masked operations and lowering patterns for X86 code generator.
Examples:
<16 x i32> @llvm.masked.load.v16i32(i8* %addr, <16 x i32> %passthru, i32 4 /* align */, <16 x i1> %mask)
declare void @llvm.masked.store.v8f64(i8* %addr, <8 x double> %value, i32 4, <8 x i1> %mask)
Scalarizer for other targets (not AVX2/AVX-512) will be done in a separate patch.
http://reviews.llvm.org/D6191
llvm-svn: 222632
This is to be consistent with StringSet and ultimately with the standard
library's associative container insert function.
This lead to updating SmallSet::insert to return pair<iterator, bool>,
and then to update SmallPtrSet::insert to return pair<iterator, bool>,
and then to update all the existing users of those functions...
llvm-svn: 222334
Hide the fact that `MDString`'s string is stored in `Value::Name` --
that's going to change soon. Update the only in-tree client that was
using it instead of `Value::getString()`.
Part of PR21532.
llvm-svn: 221951
Instead, we're going to separate metadata from the Value hierarchy. See
PR21532.
This reverts commit r221375.
This reverts commit r221373.
This reverts commit r221359.
This reverts commit r221167.
This reverts commit r221027.
This reverts commit r221024.
This reverts commit r221023.
This reverts commit r220995.
This reverts commit r220994.
llvm-svn: 221711
A pointer's pointee might not be sized: the pointee could be a function.
Report this as IK_NoInduction when calculating isInductionVariable.
This fixes PR21508.
llvm-svn: 221501
Change `Instruction::getMetadata()` to return `Value` as part of
PR21433.
Update most callers to use `Instruction::getMDNode()`, which wraps the
result in a `cast_or_null<MDNode>`.
llvm-svn: 221024
These are named following the IEEE-754 names for these
functions, rather than the libm fmin / fmax to avoid
possible ambiguities. Some languages may implement something
resembling fmin / fmax which return NaN if either operand is
to propagate errors. These implement the IEEE-754 semantics
of returning the other operand if either is a NaN representing
missing data.
llvm-svn: 220341
The SLP vectorizer should not vectorize ephemeral values. These are used to
express information to the optimizer, and vectorizing them does not lead to
faster code (because the ephemeral values are dropped prior to code generation,
vectorized or not), and obscures the information the instructions are
attempting to communicate (the logic that interprets the arguments to
@llvm.assume generically does not understand vectorized conditions).
Also, uses by ephemeral values are free (because they, and the necessary
extractelement instructions, will be dropped prior to code generation).
llvm-svn: 219816
A few minor changes to prevent @llvm.assume from interfering with loop
vectorization. First, treat @llvm.assume like the lifetime intrinsics, which
are scalarized (but don't otherwise interfere with the legality checking).
Second, ignore the cost of ephemeral instructions in the loop (these will go
away anyway during CodeGen).
Alignment assumptions and other uses of @llvm.assume can often end up inside of
loops that should be vectorized (this is not uncommon for assumptions generated
by __attribute__((align_value(n))), for example).
llvm-svn: 219741
getSmallConstantTripCount even when it isn't the exiting block.
I missed this in my first audit, very sorry. This was found in LNT and
elsewhere. I don't have a test case, but it was completely obvious from
inspection that this was the problem. I'll see if I can reduce a test
case, but I'm not really hopeful, and the value seems quite low.
llvm-svn: 219562
routines and fix all of the bugs they expose.
I hit a test case that crashed even without these asserts due to passing
a non-exiting latch to the ExitingBlock parameter of the trip count
computation machinery. However, when I add the nice asserts, it turns
out we have plenty of coverage of these bugs, they just didn't manifest
in crashers.
The core problem seems to stem from an assumption that the latch *is*
the exiting block. While this is often true, and somewhat the "normal"
way to think about loops, it isn't necessarily true. The correct way to
call the trip count routines in a *generic* fashion (that is, without
a particular exit in mind) is to just use the loop's single exiting
block if it has one. The trip count can't be computed generically unless
it does. This works great for the loop vectorizer. The loop unroller
actually *wants* to select the latch when it has to chose between
multiple exits because for unrolling it is the latch trips that matter.
But if this is the desire, it needs to explicitly guard for non-exiting
latches and check for the generic trip count in that case.
I've added the asserts, and added convenience APIs for querying the trip
count generically that check for a single exit block. I've kept the APIs
consistent between computing trip count and trip multiples.
Thansk to Mark for the help debugging and tracking down the *right* fix
here!
llvm-svn: 219550
"Unroll" is not the appropriate name for this variable. Clang already uses
the term "interleave" in pragmas and metadata for this.
Differential Revision: http://reviews.llvm.org/D5066
llvm-svn: 217528
The SLP vectorizer should propagate IR-level optimization hints/flags (nsw, nuw, exact, fast-math)
when converting scalar instructions into vectors. But this isn't a simple copy - we need to take
the intersection (the logical 'and') of the sets of flags on the scalars.
The solution is further complicated because we can have non-uniform (non-SIMD) vector ops after:
http://reviews.llvm.org/D4015http://llvm.org/viewvc/llvm-project?view=revision&revision=211339
The vast majority of changed files are existing tests that were not propagating IR flags, but I've
also added a new test file for focused testing of IR flag possibilities.
Differential Revision: http://reviews.llvm.org/D5172
llvm-svn: 217051
Adding 'IR' to the names in an attempt to be less ambiguous about the flags we're dealing with here.
The 'and' method is needed by the SLPVectorizer (PR20802) and possibly other passes.
llvm-svn: 217004
The loop vectorizer preserves wrapping, exact, and fast-math properties of scalar instructions.
This patch adds a convenience method to make that operation easier because we need to do this
in the loop vectorizer, SLP vectorizer, and possibly other places.
Although this is a 'no functional change' patch, I've added a testcase to verify that the exact
flag is preserved by the loop vectorizer. The wrapping and fast-math flags are already checked
in existing testcases.
Differential Revision: http://reviews.llvm.org/D5138
llvm-svn: 216886
Previously, the hint mechanism relied on clean up passes to remove redundant
metadata, which still showed up if running opt at low levels of optimization.
That also has shown that multiple nodes of the same type, but with different
values could still coexist, even if temporary, and cause confusion if the
next pass got the wrong value.
This patch makes sure that, if metadata already exists in a loop, the hint
mechanism will never append a new node, but always replace the existing one.
It also enhances the algorithm to cope with more metadata types in the future
by just adding a new type, not a lot of code.
Re-applying again due to MSVC 2013 being minimum requirement, and this patch
having C++11 that MSVC 2012 didn't support.
Fixes PR20655.
llvm-svn: 216870
For a detailed description of the problem see the comment in the test file.
The problematic moveBefore() calls are not required anymore because the new
scheduling algorithm ensures a correct ordering anyway.
llvm-svn: 216656
This patch adds support to recognize division by uniform power of 2 and modifies the cost table to vectorize division by uniform power of 2 whenever possible.
Updates Cost model for Loop and SLP Vectorizer.The cost table is currently only updated for X86 backend.
Thanks to Hal, Andrea, Sanjay for the review. (http://reviews.llvm.org/D4971)
llvm-svn: 216371
In unreachable blocks it's legal to have instructions like "%x = op %x".
Such instuctions are not schedulable. Therefore the SLPVectorizer has to check for
unreachable blocks and ignore them.
Fixes bug 20646.
llvm-svn: 216256
If we have a scalar reduction, we can increase the critical path length if the loop we're unrolling is inside another loop. Limit, by default to 2, so the critical path only gets increased by one reduction operation.
llvm-svn: 216140
Previously, the hint mechanism relied on clean up passes to remove redundant
metadata, which still showed up if running opt at low levels of optimization.
That also has shown that multiple nodes of the same type, but with different
values could still coexist, even if temporary, and cause confusion if the
next pass got the wrong value.
This patch makes sure that, if metadata already exists in a loop, the hint
mechanism will never append a new node, but always replace the existing one.
It also enhances the algorithm to cope with more metadata types in the future
by just adding a new type, not a lot of code.
llvm-svn: 215994
Some types, such as 128-bit vector types on AArch64, don't have any callee-saved registers. So if a value needs to stay live over a callsite, it must be spilled and refilled. This cost is now taken into account.
llvm-svn: 214859
When the cost model determines vectorization is not possible/profitable these remarks print an analysis of that decision.
Note that in selectVectorizationFactor() we can assume that OptForSize and ForceVectorization are mutually exclusive.
Reviewed by Arnold Schwaighofer
llvm-svn: 214599
The current remark is ambiguous and makes it sounds like explicitly specifying vectorization will allow the loop to be vectorized. This is not the case. The improved remark directs the user to -Rpass-analysis=loop-vectorize to determine the cause of the pass-miss.
Reviewed by Arnold Schwaighofer`
llvm-svn: 214445
This commit adds scoped noalias metadata. The primary motivations for this
feature are:
1. To preserve noalias function attribute information when inlining
2. To provide the ability to model block-scope C99 restrict pointers
Neither of these two abilities are added here, only the necessary
infrastructure. In fact, there should be no change to existing functionality,
only the addition of new features. The logic that converts noalias function
parameters into this metadata during inlining will come in a follow-up commit.
What is added here is the ability to generally specify noalias memory-access
sets. Regarding the metadata, alias-analysis scopes are defined similar to TBAA
nodes:
!scope0 = metadata !{ metadata !"scope of foo()" }
!scope1 = metadata !{ metadata !"scope 1", metadata !scope0 }
!scope2 = metadata !{ metadata !"scope 2", metadata !scope0 }
!scope3 = metadata !{ metadata !"scope 2.1", metadata !scope2 }
!scope4 = metadata !{ metadata !"scope 2.2", metadata !scope2 }
Loads and stores can be tagged with an alias-analysis scope, and also, with a
noalias tag for a specific scope:
... = load %ptr1, !alias.scope !{ !scope1 }
... = load %ptr2, !alias.scope !{ !scope1, !scope2 }, !noalias !{ !scope1 }
When evaluating an aliasing query, if one of the instructions is associated
with an alias.scope id that is identical to the noalias scope associated with
the other instruction, or is a descendant (in the scope hierarchy) of the
noalias scope associated with the other instruction, then the two memory
accesses are assumed not to alias.
Note that is the first element of the scope metadata is a string, then it can
be combined accross functions and translation units. The string can be replaced
by a self-reference to create globally unqiue scope identifiers.
[Note: This overview is slightly stylized, since the metadata nodes really need
to just be numbers (!0 instead of !scope0), and the scope lists are also global
unnamed metadata.]
Existing noalias metadata in a callee is "cloned" for use by the inlined code.
This is necessary because the aliasing scopes are unique to each call site
(because of possible control dependencies on the aliasing properties). For
example, consider a function: foo(noalias a, noalias b) { *a = *b; } that gets
inlined into bar() { ... if (...) foo(a1, b1); ... if (...) foo(a2, b2); } --
now just because we know that a1 does not alias with b1 at the first call site,
and a2 does not alias with b2 at the second call site, we cannot let inlining
these functons have the metadata imply that a1 does not alias with b2.
llvm-svn: 213864
In order to enable the preservation of noalias function parameter information
after inlining, and the representation of block-level __restrict__ pointer
information (etc.), additional kinds of aliasing metadata will be introduced.
This metadata needs to be carried around in AliasAnalysis::Location objects
(and MMOs at the SDAG level), and so we need to generalize the current scheme
(which is hard-coded to just one TBAA MDNode*).
This commit introduces only the necessary refactoring to allow for the
introduction of other aliasing metadata types, but does not actually introduce
any (that will come in a follow-up commit). What it does introduce is a new
AAMDNodes structure to hold all of the aliasing metadata nodes associated with
a particular memory-accessing instruction, and uses that structure instead of
the raw MDNode* in AliasAnalysis::Location, etc.
No functionality change intended.
llvm-svn: 213859
Prior to this change, the loop vectorizer did not make use of the alias
analysis infrastructure. Instead, it performed memory dependence analysis using
ScalarEvolution-based linear dependence checks within equivalence classes
derived from the results of ValueTracking's GetUnderlyingObjects.
Unfortunately, this meant that:
1. The loop vectorizer had logic that essentially duplicated that in BasicAA
for aliasing based on identified objects.
2. The loop vectorizer could not partition the space of dependency checks
based on information only easily available from within AA (TBAA metadata is
currently the prime example).
This means, for example, regardless of whether -fno-strict-aliasing was
provided, the vectorizer would only vectorize this loop with a runtime
memory-overlap check:
void foo(int *a, float *b) {
for (int i = 0; i < 1600; ++i)
a[i] = b[i];
}
This is suboptimal because the TBAA metadata already provides the information
necessary to show that this check unnecessary. Of course, the vectorizer has a
limit on the number of such checks it will insert, so in practice, ignoring
TBAA means not vectorizing more-complicated loops that we should.
This change causes the vectorizer to use an AliasSetTracker to keep track of
the pointers in the loop. The resulting alias sets are then used to partition
the space of dependency checks, and potential runtime checks; this results in
more-efficient vectorizations.
When pointer locations are added to the AliasSetTracker, two things are done:
1. The location size is set to UnknownSize (otherwise you'd not catch
inter-iteration dependencies)
2. For instructions in blocks that would need to be predicated, TBAA is
removed (because the metadata might have a control dependency on the condition
being speculated).
For non-predicated blocks, you can leave the TBAA metadata. This is safe
because you can't have an iteration dependency on the TBAA metadata (if you
did, and you unrolled sufficiently, you'd end up with the same pointer value
used by two accesses that TBAA says should not alias, and that would yield
undefined behavior).
llvm-svn: 213486
Summary: This patch introduces two new iterator ranges and updates existing code to use it. No functional change intended.
Test Plan: All tests (make check-all) still pass.
Reviewers: dblaikie
Reviewed By: dblaikie
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D4481
llvm-svn: 213474
IRBuilder has CreateAligned(Load|Store) functions; use them and we don't need
to make a second call to setAlignment.
No functionality change intended.
llvm-svn: 213453
There are some kinds of metadata that are safe to propagate from the scalar
instructions to the vector instructions (fpmath and tbaa currently).
Regarding TBAA, one might worry about propagating it on if-converted loads and
stores, because the metadata might have had a control dependency on the
condition, and thus actually aliased with some other non-speculated memory
access when the condition was false. However, this would be caught by the
runtime overlap checks.
llvm-svn: 213452
This patch modifies the existing DiagnosticInfo system to create a generic base
class that is inherited to produce diagnostic-based warnings. This is used by
the loop vectorizer to trigger a warning when vectorization is forced and
fails. Several tests have been added to verify this behavior.
Reviewed by: Arnold Schwaighofer
llvm-svn: 213110
string_ostream is a safe and efficient string builder that combines opaque
stack storage with a built-in ostream interface.
small_string_ostream<bytes> additionally permits an explicit stack storage size
other than the default 128 bytes to be provided. Beyond that, storage is
transferred to the heap.
This convenient class can be used in most places an
std::string+raw_string_ostream pair or SmallString<>+raw_svector_ostream pair
would previously have been used, in order to guarantee consistent access
without byte truncation.
The patch also converts much of LLVM to use the new facility. These changes
include several probable bug fixes for truncated output, a programming error
that's no longer possible with the new interface.
llvm-svn: 211749
[LLVM part]
These patches rename the loop unrolling and loop vectorizer metadata
such that they have a common 'llvm.loop.' prefix. Metadata name
changes:
llvm.vectorizer.* => llvm.loop.vectorizer.*
llvm.loopunroll.* => llvm.loop.unroll.*
This was a suggestion from an earlier review
(http://reviews.llvm.org/D4090) which added the loop unrolling
metadata.
Patch by Mark Heffernan.
llvm-svn: 211710
This patch adds support to recognize patterns such as fadd,fsub,fadd,fsub.../add,sub,add,sub... and
vectorizes them as vector shuffles if they are profitable.
These patterns of vector shuffle can later be converted to instructions such as addsubpd etc on X86.
Thanks to Arnold and Hal for the reviews. http://reviews.llvm.org/D4015
llvm-svn: 211339
If we have common uses on separate paths in the tree; process the one with greater common depth first.
This makes sure that we do not assume we need to extract a load when it is actually going to be part of a vectorized tree.
Review: http://reviews.llvm.org/D3800
llvm-svn: 210310
This patch adds support to vectorize intrinsics such as powi, cttz and ctlz in Vectorizer. These intrinsics are different from other
intrinsics as second argument to these function must be same in order to vectorize them and it should be represented as a scalar.
Review: http://reviews.llvm.org/D3851#inline-32769 and http://reviews.llvm.org/D3937#inline-32857
llvm-svn: 209873
The loop vectorizer instantiates be-taken-count + 1 as the loop iteration count.
If this expression overflows the generated code was invalid.
In case of overflow the code now jumps to the scalar loop.
Fixes PR17288.
llvm-svn: 209854
Summary:
This adds two new diagnostics: -pass-remarks-missed and
-pass-remarks-analysis. They take the same values as -pass-remarks but
are intended to be triggered in different contexts.
-pass-remarks-missed is used by LLVMContext::emitOptimizationRemarkMissed,
which passes call when they tried to apply a transformation but
couldn't.
-pass-remarks-analysis is used by LLVMContext::emitOptimizationRemarkAnalysis,
which passes call when they want to inform the user about analysis
results.
The patch also:
1- Adds support in the inliner for the two new remarks and a
test case.
2- Moves emitOptimizationRemark* functions to the llvm namespace.
3- Adds an LLVMContext argument instead of making them member functions
of LLVMContext.
Reviewers: qcolombet
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D3682
llvm-svn: 209442
Turns out that there is a very cheap way of testing whether a block is dead,
just look it up in the DomTree. We have to do this anyways so just ignore
unreachable blocks before sorting by domination. This restores a proper
ordering for std::stable_sort when dead code is present.
Covered by existing tests & buildbots running in STL debug mode (MSVC).
llvm-svn: 208492
There is no total ordering if the CFG is disconnected. We don't care if we
catch all CSE opportunities in dead code either so just exclude ignore them in
the assert.
PR19646
llvm-svn: 208461
1) Fix for printing debug locations for absolute paths.
2) Location printing is moved into public method DebugLoc::print() to avoid re-inventing the wheel.
Differential Revision: http://reviews.llvm.org/D3513
llvm-svn: 208177
When can't assume a vectorized tree is rooted in an instruction. The IRBuilder
could have constant folded it. When we rebuild the build_vector (the series of
InsertElement instructions) use the last original InsertElement instruction. The
vectorized tree root is guaranteed to be before it.
Also, we can't assume that the n-th InsertElement inserts the n-th element into
a vector.
This reverts r207746 which reverted the revert of the revert of r205018 or so.
Fixes the test case in PR19621.
llvm-svn: 207939
There is no point in creating it if we're not going to vectorize
anything. Creating the map is expensive as it creates large values.
No functionality change.
llvm-svn: 207916
There are public functions that mutate various members as well as
another private member already, so make all the members private to
avoid the discontinuity and add accessors for the values. Should
be no functional change.
llvm-svn: 207868
=[
Turns out that this was the root cause of PR19621. We found a crasher
only recently (likely due to improvements elsewhere in the SLP
vectorizer) but the reduced test case failed all the way back to here.
I've confirmed that reverting this patch both fixes the reduced test
case in PR19621 and the actual source file that led to it, so it seems
to really be rooted here. I've replied to the commit thread with
discussion of my (feeble) attempts to debug this. Didn't make it very
far, so reverting now that we have a good test case so that things can
get back to healthy while the debugging carries on.
llvm-svn: 207746
This patch changes the vectorization remarks to also inform when
vectorization is possible but not beneficial.
Added tests to exercise some loop remarks.
llvm-svn: 207574
Summary:
This calls emitOptimizationRemark from the loop unroller and vectorizer
at the point where they make a positive transformation. For the
vectorizer, it reports vectorization and interleave factors. For the
loop unroller, it reports all the different supported types of
unrolling.
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D3456
llvm-svn: 207528
Use -stats to see how many loops were analyzed for possible vectorization and how many of them were actually vectorized.
Patch by Zinovy Nis
Differential Revision: http://reviews.llvm.org/D3438
llvm-svn: 206956
definition below all of the header #include lines, lib/Transforms/...
edition.
This one is tricky for two reasons. We again have a couple of passes
that define something else before the includes as well. I've sunk their
name macros with the DEBUG_TYPE.
Also, InstCombine contains headers that need DEBUG_TYPE, so now those
headers #define and #undef DEBUG_TYPE around their code, leaving them
well formed modular headers. Fixing these headers was a large motivation
for all of these changes, as "leaky" macros of this form are hard on the
modules implementation.
llvm-svn: 206844
Some Intrinsics are overloaded to the extent that return type equality (all
that's been checked up to now) does not guarantee that the arguments are the
same. In these cases SLP vectorizer should not recurse into the operands, which
can be achieved by comparing them as "Function *" rather than simply the ID.
llvm-svn: 205424
For the purpose of calculating the cost of the loop at various vectorization
factors, we need to count dependencies of consecutive pointers as uniforms
(which means that the VF = 1 cost is used for all overall VF values).
For example, the TSVC benchmark function s173 has:
...
%3 = add nsw i64 %indvars.iv, 16000
%arrayidx8 = getelementptr inbounds %struct.GlobalData* @global_data, i64 0, i32 0, i64 %3
...
and we must realize that the add will be a scalar in order to correctly deduce
it to be profitable to vectorize this on PowerPC with VSX enabled. In fact, all
dependencies of a consecutive pointer must be a scalar (uniform), and so we
simply need to add all consecutive pointers to the worklist that currently
detects collects uniforms.
Fixes PR19296.
llvm-svn: 205387
This reverts commit r205018.
Conflicts:
lib/Transforms/Vectorize/SLPVectorizer.cpp
test/Transforms/SLPVectorizer/X86/insert-element-build-vector.ll
This is breaking libclc build.
llvm-svn: 205260
Extracts coming from phis were being hoisted, while all others were
sunk to their uses. This was inconsistent and didn't seem to serve a
purpose. Changing all extracts to be sunk to uses is a prerequisite
for adding block frequency to the SLP vectorizer's cost model.
I benchmarked the change in isolation (without block frequency). I
only saw noise on x86 and some potentially significant improvements on
ARM. No major regressions is good enough for me.
llvm-svn: 204699
noise.
Original commit log:
Replace some dead code with an assert. When I first ported this pass
from a loop pass to a function pass I did so in the naive, recursive
way. It doesn't actually work, we need a worklist instead. When
I switched to the worklist I didn't delete the naive recursion. That
recursion was also buggy because it was dead and never really exercised.
llvm-svn: 204187
pass from a loop pass to a function pass I did so in the naive,
recursive way. It doesn't actually work, we need a worklist instead.
When I switched to the worklist I didn't delete the naive recursion.
That recursion was also buggy because it was dead and never really
exercised.
llvm-svn: 204184
This requires a number of steps.
1) Move value_use_iterator into the Value class as an implementation
detail
2) Change it to actually be a *Use* iterator rather than a *User*
iterator.
3) Add an adaptor which is a User iterator that always looks through the
Use to the User.
4) Wrap these in Value::use_iterator and Value::user_iterator typedefs.
5) Add the range adaptors as Value::uses() and Value::users().
6) Update *all* of the callers to correctly distinguish between whether
they wanted a use_iterator (and to explicitly dig out the User when
needed), or a user_iterator which makes the Use itself totally
opaque.
Because #6 requires churning essentially everything that walked the
Use-Def chains, I went ahead and added all of the range adaptors and
switched them to range-based loops where appropriate. Also because the
renaming requires at least churning every line of code, it didn't make
any sense to split these up into multiple commits -- all of which would
touch all of the same lies of code.
The result is still not quite optimal. The Value::use_iterator is a nice
regular iterator, but Value::user_iterator is an iterator over User*s
rather than over the User objects themselves. As a consequence, it fits
a bit awkwardly into the range-based world and it has the weird
extra-dereferencing 'operator->' that so many of our iterators have.
I think this could be fixed by providing something which transforms
a range of T&s into a range of T*s, but that *can* be separated into
another patch, and it isn't yet 100% clear whether this is the right
move.
However, this change gets us most of the benefit and cleans up
a substantial amount of code around Use and User. =]
llvm-svn: 203364
Move the test for this class into the IR unittests as well.
This uncovers that ValueMap too is in the IR library. Ironically, the
unittest for ValueMap is useless in the Support library (honestly, so
was the ValueHandle test) and so it already lives in the IR unittests.
Mmmm, tasty layering.
llvm-svn: 202821
I am really sorry for the noise, but the current state where some parts of the
code use TD (from the old name: TargetData) and other parts use DL makes it
hard to write a patch that changes where those variables come from and how
they are passed along.
llvm-svn: 201827
'OK_NonUniformConstValue' to identify operands which are constants but
not constant splats.
The cost model now allows returning 'OK_NonUniformConstValue'
for non splat operands that are instances of ConstantVector or
ConstantDataVector.
With this change, targets are now able to compute different costs
for instructions with non-uniform constant operands.
For example, On X86 the cost of a vector shift may vary depending on whether
the second operand is a uniform or non-uniform constant.
This patch applies the following changes:
- The cost model computation now takes into account non-uniform constants;
- The cost of vector shift instructions has been improved in
X86TargetTransformInfo analysis pass;
- BBVectorize, SLPVectorizer and LoopVectorize now know how to distinguish
between non-uniform and uniform constant operands.
Added a new test to verify that the output of opt
'-cost-model -analyze' is valid in the following configurations: SSE2,
SSE4.1, AVX, AVX2.
llvm-svn: 201272
Before conditional store vectorization/unrolling we had only one
vectorized/unrolled basic block. After adding support for conditional store
vectorization this will not only be one block but multiple basic blocks. The
last block would have the back-edge. I updated the code to use a vector of basic
blocks instead of a single basic block and fixed the users to use the last entry
in this vector. But, I forgot to add the basic blocks to this vector!
Fixes PR18724.
llvm-svn: 201028
Ideally only those transform passes that run at -O0 remain enabled,
in reality we get as close as we reasonably can.
Passes are responsible for disabling themselves, it's not the job of
the pass manager to do it for them.
llvm-svn: 200892
David Majnemer