This change clarifies that for Not-NonAffine-SubRegions we actually iterate over
the subnodes and for both NonAffine-SubRegions and BasicBlocks, we perform the
schedule construction. As a result, the tree traversal becomes trivial, the
special case for a scop consisting just of a single non-affine region
disappears and the indentation of the code is reduced.
No functional change intended.
llvm-svn: 256940
At code generation, scalar reads are generated before the other
statement's instructions, respectively scalar writes after them, in
contrast to array accesses which are "executed" with the instructions
they are linked to. Therefore it makes sense to not map the scalar
accesses to a place of execution. Follow-up patches will also remove
some of the directs links from a scalar access to a single instruction,
such that only having array accesses in InstructionToAccess ensures
consistency.
Differential Revision: http://reviews.llvm.org/D13676
llvm-svn: 256298
We clarify that certain code is only executed if LSchedule is != nullptr.
Previously some of these functions have been executed, but they only passed
a nullptr through. This caused some confusion when reading the code.
llvm-svn: 256209
Besides improving the documentation and the code we now assert in case the input
is invalid (N < 0) and also do not any more return a nullptr in case USet is
empty. This should make the code more readable.
llvm-svn: 256208
If a loop has a sufficiently large amount of compute instruction in its loop
body, it is unlikely that our rewrite of the loop iterators introduces large
performance changes. As Polly can also apply beneficical optimizations (such
as parallelization) to such loop nests, we mark them as profitable.
This option is currently "disabled" by default, but can be used to run
experiments. If enabled by setting it e.g. to 40 instructions, we currently
see some compile-time increases on LNT without any significant run-time
changes.
llvm-svn: 256199
.. and add some documentation. We also simplify the code by dropping an early
check that is also covered by the the later checks. This might have a small
compile time impact, but as the scops that are skipped are small we should
probably only add this back in the unlikely case that this has a notable
compile-time cost.
No functional change intended.
llvm-svn: 256149
As we already log an error when calling invalid, scops unprofitable scops are in
any case marked invalid, but returning immediately safes (a tiny bit of) compile
time and is consistent with our use of 'invalid' in the remainder of the file.
Found by inspection.
llvm-svn: 256140
Without this return we still log the incorrect array size (and do not detect
this scop), but we would unnecessarily continue to verify that access functions
are affine. As we do not need to do this, we can return right ahead and
consequently safe compile time.
This issue was found by inspection.
llvm-svn: 256139
Instead of counting all array memory accesses associated with a load
instruction, we now explicitly check that the single array access that could
(potentially) be associated with a load instruction does not exist. This helps
to document the current behavior of Polly where load instructions can indeed
have at most one associated array access. In the unlikely case this changes
in the future, we add an assert for the case where two load accesses would
prevent us to return a single memory access, but we still should communicate
that not all array memory accesses have been removed.
This addresses post-commit comments from Johannes Doerfert for commit 255776.
llvm-svn: 256136
Scops that contain many complex branches are likely to result in complex domain
conditions that consist of a large (> 100) number of conjucts. Transforming
such domains is expensive and unlikely to result in efficient code. To avoid
long compile times we detect this case and skip such scops. In the future we may
improve this by either using non-affine subregions to hide such complex
condition structures or by exploiting in certain cases properties (e.g.,
dominance) that allow us to construct the domains of a scop in a way that
results in a smaller number improving conjuncts.
Example of a code that results in complex iteration spaces:
loop.header
/ | \ \
A0 A2 A4 \
\ / \ / \
A1 A3 \
/ \ / \ |
B0 B2 B4 |
\ / \ / |
B1 B3 ^
/ \ / \ |
C0 C2 C4 |
\ / \ / /
C1 C3 /
\ / /
loop backedge
llvm-svn: 256123
The patch fixes Bug 25759 produced by inappropriate handling of unsigned
maximum SCEV expressions by SCEVRemoveMax. Without a fix, we get an infinite
loop and a segmentation fault, if we try to process, for example,
'((-1 + (-1 * %b1)) umax {(-1 + (-1 * %yStart)),+,-1}<%.preheader>)'.
It also fixes a potential issue related to signed maximum SCEV expressions.
Tested-by: Roman Gareev <gareevroman@gmail.com>
Fixed-by: Tobias Grosser <tobias@grosser.es>
Differential Revision: http://reviews.llvm.org/D15563
llvm-svn: 255922
When running 'clang -O3 -mllvm -polly -mllvm -polly-show' we now only show the
CFGs of functions with at least one detected scop. For larger files/projects
this reduces the number of graphs printed significantly and is likely what
developers want to see. The new option -polly-view-all enforces all graphs to be
printed and the exiting option -poll-view-only limites the graph printing to
functions that match a certain pattern.
This patch requires https://llvm.org/svn/llvm-project/llvm/trunk@255889 (and
vice versa) to compile correctly.
llvm-svn: 255891
Load instructions may possibly be related to multiple memory accesses, but we
are only interested in the array read access that describes the memory location
the load instructions loads from. By using getArrayAccessfor we ensure to always
obtain the right memory access.
This issue was found by inspection without having a failing test case.
llvm-svn: 255716
This reverts commit r255471.
Johannes raised in the post-commit review of r255471 the concern that PHI
writes in non-affine regions with two exiting blocks are not really MUST_WRITE,
but we just know that at least one out of the set of all possible PHI writes
will be executed. Modeling all PHI nodes as MUST_WRITEs is probably save, but
adding the needed documentation for such a special case is probably not worth
the effort. Michael will be proposing a new patch that ensures only a single
PHI_WRITE is created for non-affine regions, which - besides other benefits -
should also allow us to use a single well-defined MUST_WRITE for such PHI
writes.
(This is not a full revert, but the condition and documentation have been
slightly extended)
llvm-svn: 255503
Before this commit, only the region's entry block was assumed to always
execute in a non-affine subregion. We replace this by a test whether it
dominates the exit block (this necessarily includes the entry block)
which should be more accurate.
llvm-svn: 255473
LLVM's IR guarantees that a value definition occurs before any use, and
also the value of a PHI must be one of the incoming values, "written"
in one of the incoming blocks. Hence, such writes are never conditional
in the context of a non-affine subregion.
llvm-svn: 255471
Over time different vocabulary has been introduced to describe the different
memory objects in Polly, resulting in different - often inconsistent - naming
schemes in different parts of Polly. We now standartize this to the following
scheme:
KindArray, KindValue, KindPHI, KindExitPHI
| ------- isScalar -----------|
In most cases this naming scheme has already been used previously (this
minimizes changes and ensures we remain consistent with previous publications).
The main change is that we remove KindScalar to clearify the difference between
a scalar as a memory object of kind Value, PHI or ExitPHI and a value (former
KindScalar) which is a memory object modeling a llvm::Value.
We also move all documentation to the Kind* enum in the ScopArrayInfo class,
remove the second enum in the MemoryAccess class and update documentation to be
formulated from the perspective of the memory object, rather than the memory
access. The terms "Implicit"/"Explicit", formerly used to describe memory
accesses, have been dropped. From the perspective of memory accesses they
described the different memory kinds well - especially from the perspective of
code generation - but just from the perspective of a memory object it seems more
straightforward to talk about scalars and arrays, rather than explicit and
implicit arrays. The last comment is clearly subjective, though. A less
subjective reason to go for these terms is the historic use both in mailing list
discussions and publications.
llvm-svn: 255467
Use it to print "null" if a MemoryAccess's access relation is not
available instead of printing nothing.
Suggested-by: Johannes Doerfert
llvm-svn: 255466
Introduce a function getStmtForRegionNode() to the corresponding
ScopStmt of a RegionNode. We can use it to call the existing
ScopStmt::isEmpty() function instead of searching for accesses.
llvm-svn: 255465
Acc==MA implies Acc->getAccessInstruction() == MA->getAccessInstruction().
Suggested as post-commit review for 254305 by Michael Kruse.
llvm-svn: 254327
The use of C++'s high-level iterator functionality instead of two while loops
and explicit iterator handling improves readability of this code.
Proposed-by: Michael Kruse <llvm@meinersbur.de>
Differential Revision: http://reviews.llvm.org/D15068
llvm-svn: 254305
Previously, accesses that originate from PHI nodes in the exit block
were registered as SCALAR. In some context they are treated as scalars,
but it makes a difference in others. We used to check whether the
AccessInstruction is a terminator to differentiate the cases.
This patch introduces an MemoryAccess origin EXIT_PHI and a
ScopArrayInfo kind KIND_EXIT_PHI to make this case more explicit. No
behavioural change intended.
Differential Revision: http://reviews.llvm.org/D14688
llvm-svn: 254149
gfortran (and fortran in general?) does not compute the address of an array
element directly from the array sizes (e.g., %s0, %s1), but takes first the
maximum of the sizes and 0 (e.g., max(0, %s0)) before multiplying the resulting
value with the per-dimension array subscript expressions. To successfully
delinearize index expressions as we see them in fortran, we first filter 'smax'
expressions out of the SCEV expression, use them to guess array size parameters
and only then continue with the existing delinearization.
llvm-svn: 253995
Trying to build up access functions for any of these blocks is likely to fail,
as error blocks may contain invalid/non-representable instructions, and blocks
dominated by error blocks may reference such instructions, which wil also cause
failures. As all of these blocks are anyhow assumed to not be executed, we can
just remove them early on.
This fixes http://llvm.org/PR25596
llvm-svn: 253818
At some point we enforced lcssa for the loop surrounding the entry block.
This is not only questionable as it does not check any other loop but also
not needed any more.
llvm-svn: 253789
In case the original parameter instruction does not have a name, but it comes
from a load instruction where the base pointer has a name we used the name of
the load instruction to give some more intuition of where the parameter came
from. To ensure this works also through GEPs which may have complex offsets,
we originally just dropped the offsets and _only_ used the base pointer name.
As this can result in multiple parameters to get the same name, we now prefix
the parameter ID to ensure parameter names are unique. This will make it easier
to understand debug output.
This change does not affect correctness, as parameter IDs (even of the same
name) can always be distinguished through the SCEV pointer stored inside them.
llvm-svn: 253330
Without this change we may start to refuse scops in larger compilation units
just because a lot of code has already been compiled earlier.
Found by inspection. I do not yet have a good test case for this.
llvm-svn: 253050
Only when we check for wrapping we want to use the store size, for all
other cases we use the alloc size now.
Suggested by: Tobias Grosser <tobias@grosser.es>
llvm-svn: 252941
If an llvm.assume dominates the SCoP entry block and the assumed condition
can be expressed as an affine inequality we will now add it to the context.
Differential Revision: http://reviews.llvm.org/D14413
llvm-svn: 252851
Error blocks may contain arbitrary instructions, among them some which we can
not modeled correctly. As we do not generate ScopStmts for error blocks anyhow
there is no point in trying to generate access functions for them.
This fixes llvm.org/PR25494
llvm-svn: 252794
In certain cases isl will not free the return values of operations for which
a computeout has been triggered. Hence, make sure we free it explicitly.
No test, as I did not manage to reduce one yet.
llvm-svn: 252766
For complex inputs our current approach of construction the boundary context
may in rare cases become computationally so expensive that it is better to
abort. This change adds a compute out check that bounds the compuations we
spend on boundary context construction and bails out if this limit is reached.
We can probably make our boundary construction algorithm more efficient, but
this requires some more investigation and probably also some additional changes
to isl. Until these have been added, we bound the compile time to ensure our
buildbots are green.
llvm-svn: 252758
In certain rare cases (mostly -polly-process-unprofitable on large sequences
of conditions - often without any loop), we see some compile-time timeouts due
to the construction of an overly complex assumption context. This change limits
the number of disjuncts to 150 (adjustable), to prevent us from creating
assumptions contexts that are too large for even the compilation to finish.
The limit has been choosen as large as possible to make sure we do not
unnecessarily drop test coverage. If such cases also appear in
-polly-process-unprofitable=false mode we may need to think about this again,
as the current limitations may still allow assumptions that are way to complex
to be checked profitably at run-time.
There is also certainly room for improvement regarding how (and how efficient)
we construct an assumed context, but this requires some more thinking.
This completes llvm.org/PR25458
llvm-svn: 252750
r252713 introduced a couple of regressions due to later basic blocks refering
to instructions defined in error blocks which have not yet been modeled.
This commit is currently just encoding limitations of our modeling and code
generation backends to ensure correctness. In theory, we should be able to
generate and optimize such regions, as everything that is dominated by an error
region is assumed to not be executed anyhow. We currently just lack the code
to make this happen in practice.
llvm-svn: 252725
Previously, we just skipped error blocks during scop construction. With
this change we make sure we can construct domains for error blocks such that
these domains can be forwarded to subsequent basic blocks.
This change ensures that basic blocks that post-dominate and are dominated by
a basic block that branches to an error condition have the very same iteration
domain as the branching basic block. Before, this change we would construct
a domain that excludes all error conditions. Such domains could become _very_
complex and were undesirable to build.
Another solution would have been to drop these constraints using a
dominance/post-dominance check instead of modeling the error blocks. Such
a solution could also work in case of unreachable statements or infinite
loops in the scop. However, as we currently (to my believe incorrectly) model
unreachable basic blocks in the post-dominance tree, such a solution is not
yet feasible and requires first a change to LLVM's post-dominance tree
construction.
This commit addresses the most sever compile time issue reported in:
http://llvm.org/PR25458
llvm-svn: 252713
We now create all invariant equivalence classes for required invariant loads
instead of creating them on-demand. This way we can check if a parameter
references an invariant load that is actually not executed and was therefor
not materialized. If that happens the parameter is not materialized either.
This fixes bug 25469.
llvm-svn: 252701
Since 252422 we do not only distinguish two ScopArrayInfo kinds, PHI nodes
and others, but work with three kind of ScopArrayInfo objects. SCALAR, PHI and
ARRAY objects. Instead of keeping two boolean flags isPHI and isScalar and
wonder what an ScopArrayInfo object of kind (!isScalar && isPHI) is, we
list now explicitly the three different possible types of memory objects.
This change also allows us to remove the confusing nested pairs that have
been used in ArrayInfoMapTy.
llvm-svn: 252620
In polly the first dimensions of an array as well as all scalars do not carry
any size information. This commit makes this explicit in the interface of
getDimensionSize. Before this commit getDimensionSize(0) returned the size of
the first dimension that carried a size. After this commit getDimensionSize(i)
will either return the size of dimension 'i' or assert in case 'i' does not
carry a size or does not exist at all.
This very same behaviour was already present in getDimensionSizePw(). This
commit also adds assertions that ensure getDimensionSizePw() is called
appropriately.
llvm-svn: 252607
Memory references are now printed as follows:
Old New
Scalars: i64 MemRef_val[*] i64 MemRef_val;
Arrays: i64 MemRef_A[*][%m][%o][8] i64 MemRef_A[*][%m][%o];
We do not print any more information about the element size in the type. Such
information has already been available in a comment after the scalar/array
declaration. It was redundant and did not match well with what people were used
from C.
llvm-svn: 252602
If a SCoP contains error blocks we cannot use the domain constraints
to simplify the assumptions as the domain is already influenced by the
assumptions we took. Before this patch we did that and some assumptions
became self-fulfilling as they were implied by the domain constraints.
llvm-svn: 252424
Even if a scalar and memory access have the same base pointer, we cannot use
one SAI object as the type but also the number of dimensions are wrong. For
the attached test case this caused a crash in the invariant load hoisting,
though it could cause various other problems too.
This fixes bug 25428 and a execution time bug in MallocBench/cfrac.
Reported-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
llvm-svn: 252422
Remove all the implicit ilist iterator conversions from polly, in
preparation for making them illegal in ADT. There was one oddity I came
across: at line 95 of lib/CodeGen/LoopGenerators.cpp, there was a
post-increment `Builder.GetInsertPoint()++`.
Since it was a no-op, I removed it, but I admit I wonder if it might be
a bug (both before and after this change)? Perhaps it should be a
pre-increment?
llvm-svn: 252357
Before this commit memory reference identifiers have only been unique per
basic block, but not per (non-affine) ScopStmt. This commit now uses the
MemoryAccess base pointer to uniquely identify each Memory access.
llvm-svn: 252200
An incoming value from a block the is not inside the scop is an
external use, even if the phi is inside the scop. A previous fix in
r251208 did not apply if the phi is inside a non-affine subregion. We
move the check for this phi case before the non-affine subregion check.
llvm-svn: 252157
We do not need to model read-only statements in the SCoP as they will
not cause any side effects that are visible to the outside anyway.
Removing them should safe us time and might even simplify the ASTs we
generate.
Differential Revision: http://reviews.llvm.org/D14272
llvm-svn: 251948
ScalarEvolution doesn't allow the operands of an AddRec to be variant in the
loop of the AddRec. When we rewrite parameter SCEVs it might seem like the
new SCEV violates this property and ScalarEvolution will trigger an
assertion. To avoid this we move the start part out of an AddRec when we
rewrite it, thus avoid the operands to be possibly variant completely.
llvm-svn: 251945
In some cases different memory accesses access the very same array using a
different multi-dimensional array layout where the same dimensions have
different sizes. Instead of asserting when encountering this issue, we
gracefully bail out for this scop.
This fixes llvm.org/PR25252
llvm-svn: 251791
We remove -polly-detect-unprofitable and -polly-no-early-exit. Both have been
superseeded by -polly-process-unprofitable and were only kept as aliases for
our buildbots to continue to work. As all buildbots have been moved to the new
options, we can now remove the old ones for good.
llvm-svn: 251787
Volatile or atomic memory accesses are currently not supported. Neither did
we think about any special handling needed nor do we support the unknown
instructions the alias set tracker turns them into sometimes. Before this
patch, us not supporting unkown instructions in an alias set caused the
following assertion failures:
Assertion `AG.size() > 1 && "Alias groups should contain at least two accesses"'
failed
llvm-svn: 251234
When verifying if a scop is still valid we rerun all analysis, but did not
update DetectionContextMap. This change ensures that information, e.g. about
non-affine regions, is correctly updated
llvm-svn: 251227
the size expression.
We previously only checked if the size expression is 'undef', but allowed size
expressions of the form 'undef * undef' by accident. After this change we now
require size expressions to be affine which implies no 'undef' appears anywhere
in the expression.
llvm-svn: 251225
of the Region are external.
During code generation we split off the parts of the PHI nodes in the entry
block, which have incoming blocks that are not part of the region. As these
split-off PHI nodes then are external uses, we consequently also need to model
these uses in ScopInfo.
llvm-svn: 251208
There are several different kinds of constants that could occur in a
branch condition, however we can only handle the most interesting one
namely constant integers. To this end we have to treat others as
non-affine.
This fixes bug 25244.
llvm-svn: 250669
We build the schedule based on a traversal of the region and accumulate
information for each loop in it. The total schedule is associated with the
loop surrounding the SCoP, though it can happen that there are blocks in the
SCoP which are part of loops that are only partially in the SCoP. Instead of
associating information with them (they are not part of the SCoP and
consequently are not modeled) we have to associate the schedule information
with the surrounding loop if any.
This fixes bug 25240.
llvm-svn: 250668
Accesses that have a relative offset (in bytes) that is not divisible
by the type size (in bytes) will be represented as empty in the SCoP
description. This is on its own not good but it also crashed the
invariant load hoisting. This patch will fix the latter problem while
the former should be addressed too.
This fixes bug 25236.
llvm-svn: 250664
If the base pointer of a load is invariant and defined in the SCoP but
not loaded we cannot hoist the load as we would not hoist the base
pointer definition.
This fixes bug 25237.
llvm-svn: 250663
Sorting is replaced by a demand driven code generation that will pre-load a
value when it is needed or, if it was not needed before, at some point
determined by the order of invariant accesses in the program. Only in very
little cases this demand driven pre-loading will kick in, though it will
prevent us from generating faulty code. An example where it is needed is
shown in:
test/ScopInfo/invariant_loads_complicated_dependences.ll
Invariant loads that appear in parameters but are not on the top-level (e.g.,
the parameter is not a SCEVUnknown) will now be treated correctly.
Differential Revision: http://reviews.llvm.org/D13831
llvm-svn: 250655
Polly can now be used as a analysis only tool as long as the code
generation is disabled. However, we do not have an alternative to the
independent blocks pass in place yet, though in the relevant cases
this does not seem to impact the performance much. Nevertheless, a
virtual alternative that allows the same transformations without
changing the input region will follow shortly.
llvm-svn: 250652
While clang-format takes care that the line-length is not surpassed, the
resulting comments sometimes look not optimal. We re-flow the text in the
comment to avoid these ugly single-word lines.
llvm-svn: 250626
Instead of generating implicit loads within basic blocks, put them
before the instructions of the statment itself, including non-affine
subregions. The region's entry node is dominating all blocks in the
region and therefore the loaded value will be available there.
Implicit writes in block-stmts were already stored back at the end of
the block. Now, also generate the stores of non-affine subregions when
leaving the statement, i.e. in the exiting block.
This change is required for array-mapped implicits ("De-LICM") to
ensure that there are no dependencies of demoted scalars within
statments. Statement load all required values, operator on copied in
registers, and then write back the changed value to the demoted memory.
Lifetimes analysis within statements becomes unecessary.
Differential Revision: http://reviews.llvm.org/D13487
llvm-svn: 250625
Accesses for exit node phis will be handled separately by
buildPHIAccesses if there is more than one exiting edge,
buildScalarDependences does not need to create additional SCALAR
accesses.
This is a corrected version of r250517, which was reverted in r250607.
Differential Revision: http://reviews.llvm.org/D13848
llvm-svn: 250622
When pulling a llvm::Value to be written as a PHI write, the former
code did only check whether it is within the same basic block, but it
could also be the same non-affine subregion. In that case some
unecessary pair of MemoryAccesses would have been created.
Two unit test were explicitely checking for the unecessary writes,
including the comments that the writes are unecessary.
llvm-svn: 250411
We also allow such products for cases where 'Parameter' is loaded within the
scop, but where we can dynamically verify that the value of 'Parameter' remains
unchanged during the execution of the scop.
This change relies on Polly's new RequiredILS tracking infrastructure recently
contributed by Johannes.
llvm-svn: 250019
The domain generation can handle lazy && and || by default but eager
evaluated expressions were dismissed as non-affine. With this patch we
will allow arbitrary combinations of and/or bit-operations in the
conditions of branches.
Differential Revision: http://reviews.llvm.org/D13624
llvm-svn: 249971
Helper functions in the BlockGenerators.h/cpp introduce dependences
from the frontend to the backend of Polly. As they are used in
ScopDetection, ScopInfo, etc. we move them to the ScopHelper file.
llvm-svn: 249919
If a (assumed) invariant location is loaded multiple times we
generated a parameter for each location. However, this caused compile
time problems for several benchmarks (e.g., 445_gobmk in SPEC2006 and
BT in the NAS benchmarks). Additionally, the code we generate is
suboptimal as we preload the same location multiple times and perform
the same checks on all the parameters that refere to the same value.
With this patch we consolidate the invariant loads in three steps:
1) During SCoP initialization required invariant loads are put in
equivalence classes based on their pointer operand. One
representing load is used to generate a parameter for the whole
class, thus we never generate multiple parameters for the same
location.
2) During the SCoP simplification we remove invariant memory
accesses that are in the same equivalence class. While doing so
we build the union of all execution domains as it is only
important that the location is at least accessed once.
3) During code generation we only preload one element of each
equivalence class with the unified execution domain. All others
are mapped to that preloaded value.
Differential Revision: http://reviews.llvm.org/D13338
llvm-svn: 249853
This was left out from the original patch proposed in
http://reviews.llvm.org/D13195
even though it is needed to define an order invariant loads
are hoisted.
llvm-svn: 249680
Drop an unused flag polly-allow-non-scev-backedge-taken-count and also
its occurrences from the tests.
Contributed-by: Chris Jenneisch <chrisj@codeaurora.org>
Differential Revision: http://reviews.llvm.org/D13400
llvm-svn: 249675
ScopDetection users are interested in the detection context and access
these via different get-methods. However, not all information was
exposed though the number of maps to hold it was increasing steadily.
With this change only the detection contexts the rejection log and the
ValidRegions set are mapped. The former is needed, the second could be
integrated in the first and the ValidRegions set is only needed for the
deterministic order of the regions.
llvm-svn: 249614
This replaces the support for user defined error functions by a
heuristic that tries to determine if a call to a non-pure function
should be considered "an error". If so the block is assumed not to be
executed at runtime. While treating all non-pure function calls as
errors will allow a lot more regions to be analyzed, it will also
cause us to dismiss a lot again due to an infeasible runtime context.
This patch tries to limit that effect. A non-pure function call is
considered an error if it is executed only in conditionally with
regards to a cheap but simple heuristic.
llvm-svn: 249611
This patch allows invariant loads to be used in the SCoP description,
e.g., as loop bounds, conditions or in memory access functions.
First we collect "required invariant loads" during SCoP detection that
would otherwise make an expression we care about non-affine. To this
end a new level of abstraction was introduced before
SCEVValidator::isAffineExpr() namely ScopDetection::isAffine() and
ScopDetection::onlyValidRequiredInvariantLoads(). Here we can decide
if we want a load inside the region to be optimistically assumed
invariant or not. If we do, it will be marked as required and in the
SCoP generation we bail if it is actually not invariant. If we don't
it will be a non-affine expression as before. At the moment we
optimistically assume all "hoistable" (namely non-loop-carried) loads
to be invariant. This causes us to expand some SCoPs and dismiss them
later but it also allows us to detect a lot we would dismiss directly
if we would ask e.g., AliasAnalysis::canBasicBlockModify(). We also
allow potential aliases between optimistically assumed invariant loads
and other pointers as our runtime alias checks are sound in case the
loads are actually invariant. Together with the invariant checks this
combination allows to handle a lot more than LICM can.
The code generation of the invariant loads had to be extended as we
can now have dependences between parameters and invariant (hoisted)
loads as well as the other way around, e.g.,
test/Isl/CodeGen/invariant_load_parameters_cyclic_dependence.ll
First, it is important to note that we cannot have real cycles but
only dependences from a hoisted load to a parameter and from another
parameter to that hoisted load (and so on). To handle such cases we
materialize llvm::Values for parameters that are referred by a hoisted
load on demand and then materialize the remaining parameters. Second,
there are new kinds of dependences between hoisted loads caused by the
constraints on their execution. If a hoisted load is conditionally
executed it might depend on the value of another hoisted load. To deal
with such situations we sort them already in the ScopInfo such that
they can be generated in the order they are listed in the
Scop::InvariantAccesses list (see compareInvariantAccesses). The
dependences between hoisted loads caused by indirect accesses are
handled the same way as before.
llvm-svn: 249607
This single option replaces -polly-detect-unprofitable and -polly-no-early-exit
and is supposed to be the only option that disables compile-time heuristics that
aim to bail out early on scops that are believed to not benefit from Polly
optimizations.
Suggested-by: Johannes Doerfert
llvm-svn: 249426
A statement with an empty domain complicates the invariant load
hoisting and does not help any subsequent analysis or transformation.
In fact it might introduce parameter dimensions or increase the
schedule dimensionality. To this end, we remove statements with an
empty domain early in the SCoP simplification.
llvm-svn: 249276
Before isValidCFG() could hide the fact that a loop is non-affine by
over-approximation. This is problematic if a subregion of the loop contains
an exit/latch block and is over-approximated. Now we do not over-approximate
in the isValidCFG function if we check loop control. If such control is
non-affine the whole loop is over-approximated, not only a subregion.
llvm-svn: 249273
This patch cannot be tested on its own as the isValidCFG currently
hides the fact that control is actually non-affine with
over-approximation. This will be corrected in the next patch and a
test for non-affine latches will be added.
llvm-svn: 249272
We have to skip accesses in non-affine subregions during hoisting as
they might not be executed under the same condition as the entry of
the non-affine subregion.
llvm-svn: 249139
This moves the construction of ScopStmt to the beginning of the
ScopInfo pass. The late creation was a result of the earlier separation
of ScopInfo and TempScopInfo. This will avoid introducing more
ScopStmt-like maps in future commits. The AccFuncMap will also be
removed in some future commit. DomainMap might also be included into
ScopStmt.
The order in which ScopStmt are created changes and initially creates
empty statements that are removed in a simplification.
Differential Revision: http://reviews.llvm.org/D13341
llvm-svn: 249132
When error blocks are not terminated by an unreachable they have successors
that might only be reachable via error blocks. Additionally, branches in
error blocks are not checked during SCoP detection, thus we might not be able
to handle them. With this patch we do not try to model error block exit
conditions. Anything that is only reachable via error blocks is ignored too,
as it will not be executed in the optimized version of the SCoP anyway.
llvm-svn: 249099
With this patch we erase cached results for regions that are invalid
as early as possible. If we do not (as before), it is possible that
two expanded regions will have the same address and the tracked
results for both are mixed. Currently this would "only" cause
pessimism in later passes but that will change when we allow invariant
loads in the SCoP. Additionally, it triggers non-deterministic
results as we might dismiss a later region because of results cached
for an earlier one.
There is no test case as the problem occurs only non-deterministically.
llvm-svn: 249000
This makes ScopInfo's scop member available earlier to other methods which will make some planned changes simpler.
No behavioral change intended
llvm-svn: 248879
As a first step in the direction of assumed invariant loads (loads
that are not written in some context) we now detect and hoist
definitively invariant loads. These invariant loads will be preloaded
in the code generation and used in the optimized version of the SCoP.
If the load is only conditionally executed the preloaded version will
also only be executed under the same condition, hence we will never
access memory that wouldn't have been accessed otherwise. This is also
the most distinguishing feature to licm.
As hoisting can make statements empty we will simplify the SCoP and
remove empty statements that would otherwise cause artifacts in the
code generation.
Differential Revision: http://reviews.llvm.org/D13194
llvm-svn: 248861
This patch allows switch instructions with affine conditions in the
SCoP. Also switch instructions in non-affine subregions are allowed.
Both did not require much changes to the code, though there was some
refactoring needed to integrate them without code duplication.
In the llvm-test suite the number of profitable SCoPs increased from
135 to 139 but more importantly we can handle more benchmarks and user
inputs without preprocessing.
Differential Revision: http://reviews.llvm.org/D13200
llvm-svn: 248701
This check was needed at some point but seems not useful anymore. Only
one adjustment in the domain generation was needed to cope with the
cases this check prevented from happening before.
llvm-svn: 248695
When the whole SCoP is a non-affine region we need to use the
surrounding loop in the construction of the schedule as that is
the one that will be looked up after the schedule generation.
This fixes bug 24947
llvm-svn: 248667
When recovering multi-dimensional memory accesses, it may happen that different
accesses to the same base array are recovered with different dimensionality.
This patch ensures that the dimensionalities are unified by adding zero valued
dimensions to acesses with lower dimensionality. When starting to model
fixed-size arrays as multi-dimensional in 247906, this has not been taken
care of.
llvm-svn: 248662
There are three possible reasons to add a memory memory access: For explicit load and stores, for llvm::Value defs/uses, and to emulate PHI nodes (the latter two called implicit accesses). Previously MemoryAccess only stored IsPHI. Register accesses could be identified through the isScalar() method if it was no IsPHI. isScalar() determined the number of dimensions of the underlaying array, scalars represented by zero dimensions.
For the work on de-LICM, implicit accesses can have more than zero dimensions, making the distinction of isScalars() useless, hence now stored explicitly in the MemoryAccess. Instead, we replace it by isImplicit() and avoid the term scalar for zero-dimensional arrays as it might be confused with llvm::Value which are also often referred to as scalars (or alternatively, as registers).
No behavioral change intended, under the condition that it was impossible to create explicit accesses to zero-dimensional "arrays".
llvm-svn: 248616
This makes the intent of each created object clearer and allows to add more specific asserts. The bug fixed in r248535 has been discovered this way.
No functional change intended; everything should behave as before.
llvm-svn: 248603
This change addresses three issues:
- Read only scalars that enter a PHI node through an edge that comes from
outside the scop are not modeled any more, as such PHI nodes will always
be initialized to this initial value right before the SCoP is entered.
- For PHI nodes that depend on a scalar value that is defined outside the
scop, but where the scalar values is passed through an edge that itself
comes from a BB that is part of the region, we introduce in this basic
block a read of the out-of-scop value to ensure it's value is available
to write it into the PHI alloc location.
- Read only uses of scalars by PHI nodes are ignored in the general read only
handling code, as they are taken care of by the general PHI node modeling
code.
llvm-svn: 248535
After the merge of TempScopInfo into ScopInfo the analysis output
remained because of the existing unit tests. These remains are removed
and the units tests converted to match the equivalent output of
ScopInfo's analysis output. The unit tests are also moved into the
directory of ScopInfo tests.
Differential Revision: http://reviews.llvm.org/D13116
llvm-svn: 248485
A missing return statement that previously did not have a visibly negative
effect caused after some data-structure changes in r248024 multi-dimensional
accesses to be modeled both multi-dimensional as well as linearized. This
commit adds the missing return to avoid the incorrect double modeling as
well as the compile time increases it caused.
llvm-svn: 248171
If we encounter a <nsw> tagged AddRec for a loop we know the trip count of
that loop has to be bounded or the semantics is undefined anyway. Hence, we
only need to add unbounded assumptions if no such AddRec is known.
llvm-svn: 248128
So far we ignored the unbounded parts of the iteration domain, however
we need to assume they do not occure at all to remain sound if they do.
llvm-svn: 248126
We now add loop carried information during the second traversal of the
region instead of in a intermediate step in-between. This makes the
generation simpler, removes code and should even be faster.
llvm-svn: 248125
In order to allow multiple back edges we:
- compute the conditions under which each back edge is taken
- build the union over all these conditions, thus the condition that
any back edge is taken
- apply the same logic to the union we applied to a single back edge
llvm-svn: 248120
As we currently do not perform any optimizations that targets (or is
even aware) small trip counts we will skip them when we count the
loops in a region.
llvm-svn: 248119
All MemoryAccess objects will be owned by ScopInfo::AccFuncMap which
previously stored the IRAccess objects. Instead of creating new
MemoryAccess objects, the already created ones are reused, but their
order might be different now. Some fields of IRAccess and MemoryAccess
had the same meaning and are merged.
This is the last step of fusioning TempScopInfo.{h|cpp} and
ScopInfo.{h.cpp}. Some refactoring might still make sense.
Differential Revision: http://reviews.llvm.org/D12843
llvm-svn: 248024
In some cases instcombine introduces bitcasts that slightly obfuscate the
multi-dimensionality of an array. This patch teaches our fixed-size
delinearization how to look through bitcasts.
llvm-svn: 247928
This information is implicitly available through the multi-dimensionality of
memory accesses. This reduces compile time for 3mm from 430ms to 400ms and
should generally benefit compile time for cases where the assumed context
is complex.
llvm-svn: 247907
If the GEP instructions give us enough insights, model scalar accesses as
multi-dimensional (and generate the relevant run-time checks to ensure
correctness). This will allow us to simplify the dependence computation in
a subsequent commit.
llvm-svn: 247906
This will allow to generate non-wrap assumptions for integer expressions
that are part of the SCoP. We compare the common isl representation of
the expression with one computed with modulo semantic. For all parameter
combinations they are not equal we can have integer overflows.
The nsw flags are respected when the modulo representation is computed,
nuw and nw flags are ignored for now.
In order to not increase compile time to much, the non-wrap assumptions
are collected in a separate boundary context instead of the assumed
context. This helps compile time as the boundary context can become
complex and it is therefor not advised to use it in other operations
except runtime check generation. However, the assumed context is e.g.,
used to tighten dependences. While the boundary context might help to
tighten the assumed context it is doubtful that it will help in practice
(it does not effect lnt much) as the boundary (or no-wrap assumptions)
only restrict the very end of the possible value range of parameters.
PET uses a different approach to compute the no-wrap context, though lnt runs
have shown that this version performs slightly better for us.
llvm-svn: 247732
Due to the new domain generation, the SCoP keeps track of the domain
for all blocks, thus the SCEVAffinator can now work with blocks to avoid
duplication of the domains.
llvm-svn: 247731
At some point we build loop trip counts using this method. It was replaced by
a simpler trick that works only for affine (e.g., not modulo) constraints and
relies on the removal of unbounded parts. In order to allow modulo constrains
again we go back to the former, more accurate method.
llvm-svn: 247540
Summary:
TempScop is basically a holder for AccFuncMap, the dictionary from BasicBlocks to IRAccess lists. We move the list into polly::Scop and remove the polly::TempScop class.
There is one small change in behavior: If ScopInfo finds that its AssumedContext is impossible, it bails out by deleting the Scop object. The TempScop::print (invoked with opt -polly-scops -analyze) cannot print the AccFuncMap anymore as it would with a separate TempScop.
Differential Revision: http://reviews.llvm.org/D12803
llvm-svn: 247480
Hoist runtime checks in the loop nest if they guard an "error" like event.
Such events are recognized as blocks with an unreachable terminator or a call
to the ubsan function that deals with out of bound accesses. Other "error"
events can be added easily.
We will ignore these blocks when we detect/model/optmize and code generate SCoPs
but we will make sure that they would not have been executed using the assumption
framework.
llvm-svn: 247310
As we do not rely on ScalarEvolution any more we do not need to get
the backedge taken count. Additionally, our domain generation handles
everything that is affine and has one latch and our ScopDetection will
over-approximate everything else.
This change will therefor allow loops with:
- one latch
- exiting conditions that are affine
Additionally, it will not check for structured control flow anymore.
Hence, loops and conditionals are not necessarily single entry single
exit regions any more.
Differential Version: http://reviews.llvm.org/D12758
llvm-svn: 247289
The TempScopInfo (-polly-analyze-ir) pass is removed and its work taken
over by ScopInfo (-polly-scops). Several tests depend on
-polly-analyze-ir and use -polly-scops instead which for the moment
prints the output of both passes. This again is not expected by some
other tests, especially those with negative searches, which have been
adapted.
Differential Version: http://reviews.llvm.org/D12694
llvm-svn: 247288
This patch replaces the last legacy part of the domain generation, namely the
ScalarEvolution part that was used to obtain loop bounds. We now iterate over
the loops in the region and propagate the back edge condition to the header
blocks. Afterwards we propagate the new information once through the whole
region. In this process we simply ignore unbounded parts of the domain and
thereby assume the absence of infinite loops.
+ This patch already identified a couple of broken unit tests we had for
years.
+ We allow more loops already and the step to multiple exit and multiple back
edges is minimal.
+ It allows to model the overflow checks properly as we actually visit
every block in the SCoP and know where which condition is evaluated.
- It is currently not compatible with modulo constraints in the
domain.
Differential Revision: http://reviews.llvm.org/D12499
llvm-svn: 247279
This prepares for a series of patches that merges TempScopInfo into ScopInfo to
reduce Polly's code complexity. Only ScopInfo.{cpp|h} will be left thereafter.
Moving the code of TempScopInfo in one commit makes the mains diffs simpler to
understand.
In detail, merging the following classes is planned:
TempScopInfo into ScopInfo
TempScop into Scop
IRAccess into MemoryAccess
Only moving code, no functional changes intended.
Differential Version: http://reviews.llvm.org/D12693
llvm-svn: 247274
The support for pointer expressions is broken as it can only handle
some patterns in the IslExprBuilder. We should to treat pointers in
expressions the same as integers at some point and revert this patch.
llvm-svn: 247147
While we do not need to model PHI nodes in the region exit (as it is not part
of the SCoP), we need to prepare for the case that the exit block is split in
code generation to create a single exiting block. If this will happen, hence
if the region did not have a single exiting block before, we will model the
operands of the PHI nodes as escaping scalars in the SCoP.
Differential Revision: http://reviews.llvm.org/D12051
llvm-svn: 247078
Instead of having two separate options
-polly-detect-scops-in-functions-without-loops and
-polly-detect-scops-in-regions-without-loops we now just use
-polly-detect-unprofitable to force the detection of scops ignoring any compile
time saving bailout heuristics.
llvm-svn: 247057
Originally, we disallowed the import of multi-dimensional access functions due
to our code generation not supporting the generation of new address expressions
for multi-dimensional memory accesses. When building our run-time alias check
infrastructure we added code generation support for multi-dimensional address
calculations. Hence, we can now savely allow the import of new
multi-dimensional access functions.
llvm-svn: 246917
By just removing dimensions (and the constraints they are involved in) we
may loose information about the dimensions we do not remove. By instead
using project_out, we are sure all constraints on the outer dimensions are
preserved.
No test case, as this error condition is very unlikely to be triggered by
isl's current code. We still 'fix' this, as isl gives little guarantees
regarding the behavior of remove_divs.
llvm-svn: 246567
Code generation currently does not expect unbounded loops. When
using ISL to compute the loop trip count, if we find that the
iteration domain remains unbounded, we invalidate the Scop by
creating an infeasible context.
Contributed-by: Matthew Simpson <mssimpso@codeaurora.org>
This fixes PR24634.
Differential Revision: http://reviews.llvm.org/D12493
llvm-svn: 246477
Instead of building domains with MaxLoopDepth dimensions, we now build
the domains such that they have the right amount of dimensions all the
time.
llvm-svn: 246443
Before this commit we did this only for Arguments or Constants, but indeed
an instruction may define a value a lot higher up in the dominance tree, but
the actual write generally needs to happen right before branching to the
PHI node. Otherwise, the writes of different branches into PHI nodes may get
intermixed if they lay higher up in the dominance tree.
llvm-svn: 246441
There is no reason the loops in a region need to touch either entry or exit
block. Hence, we need to look through all loops that may touch the region as
well as their children to understand if our region has at least two loops.
llvm-svn: 246433
While ignoring read-only scalar dependences it was not necessary to consider
store instructins, but as store instructions can be the target of a scalar
read-only dependency we need to consider them for the construction of scalar
read-only dependences.
llvm-svn: 246429
Even though these are not strictly necessary for sequential code generation,
we still model both for sequential and parallel code generation to reduce
the set of configurations that needs to be tested. If this turns out, against
what we currently see, to be significant overhead, we can decide to limit this
feature again to parallel code-generation use cases only.
llvm-svn: 246420
In order to compute domain conditions for conditionals we will now
traverse the region in the ScopInfo once and build the domains for
each block in the region. The SCoP statements can then use these
constraints when they build their domain.
The reason behind this change is twofold:
1) This removes a big chunk of preprocessing logic from the
TempScopInfo, namely the Conditionals we used to build there.
Additionally to moving this logic it is also simplified. Instead
of walking the dominance tree up for each basic block in the
region (as we did before), we now traverse the region only
once in order to collect the domain conditions.
2) This is the first step towards the isl based domain creation.
The second step will traverse the region similar to this step,
however it will propagate back edge conditions. Once both are in
place this conditional handling will allow multiple exit loops
additional logic.
Reviewers: grosser
Differential Revision: http://reviews.llvm.org/D12428
llvm-svn: 246398
Our code generation currently does not support scalar references to metadata
values. Hence, it would crash if we try to model scalar dependences to metadata
values. Fortunately, for one of the common uses, debug information, we can
for now just ignore the relevant intrinsics and consequently the issue of how
to model scalar dependences to metadata.
llvm-svn: 246388
If a region does not have more than one loop, we do not identify it as
a Scop in ScopDetection. The main optimizations Polly is currently performing
(tiling, preparation for outer-loop vectorization and loop fusion) are unlikely
to have a positive impact on individual loops. In some cases, Polly's run-time
alias checks or conditional hoisting may still have a positive impact, but those
are mostly enabling transformations which LLVM already performs for individual
loops. As we do not focus on individual loops, we leave them untouched to not
introduce compile time regressions and execution time noise. This results in
good compile time reduction (oourafft: -73.99%, smg2000: -56.25%).
Contributed-by: Pratik Bhatu <cs12b1010@iith.ac.in>
Reviewers: grosser
Differential Revision: http://reviews.llvm.org/D12268
llvm-svn: 246161
Use ISL to compute the loop trip count when scalar evolution is unable to do
so.
Contributed-by: Matthew Simpson <mssimpso@codeaurora.org>
Differential Revision: http://reviews.llvm.org/D9444
llvm-svn: 246142
Originally, we intersected the iteration space with the AssumedContext before
computing the minimal/maximal memory offset in our run-time alias checks. With
this patch we drop this intersection as the AssumedContext can - for larger or
more complex scops - become very complicated (contain many disjuncts). When
intersecting an object with many disjuncts with other objects, the number of
disjuncts in these other objects also increases quickly. As a result, the
compile time is unnecessarily increased. This patch now drops the intersection
with the assumed context to ensure we do not pay unnecessary compile time
costs.
With this patch we see -3.17% reduction in compile time for 3mm with default
flags and -17.87% when compiling 3mm with -DPOLYBENCH_USE_C99_PROTO flag. We
did not observe any regressions in LNT.
Contributed-by: Pratik Bhatu <cs12b1010@iith.ac.in>
Reviewers: grosser
Differential Revision: http://reviews.llvm.org/D12198
llvm-svn: 245617
To avoid multiple exits and the resulting complicated conditions when
creating a SCoP we now use the single hasFeasibleRuntimeContext()
check to decide if a SCoP should be dismissed right after
construction. If building runtime checks failed the assumed context is
made infeasible, hence the optimized version will never be executed
and the SCoP can be dismissed.
llvm-svn: 245593
If nothing is executed we can bail out early. Otherwise we can use the
constraints that ensure at least one statement is executed for
simplification.
llvm-svn: 245585
We will record if a SAI is the base of another SAI or derived from it.
This will allow to reason about indirect base pointers later on and
allows a clearer picture of indirection also in the SCoP dump.
llvm-svn: 245584
Instead of generating code for an empty assumed context we bail out
early. As the number of assumptions we generate increases this becomes
more and more important. Additionally, this change will allow us to
hide internal contexts that are only used in runtime checks e.g., a
boundary context with constraints not suited for simplifications.
llvm-svn: 245540
This patch changes Polly to compute the data-dependences on the schedule tree
instead of a flat schedule representation. Calculating dependences directly on
the schedule tree results in some good compile-time improvements (adi : -23.35%,
3mm : -9.57%), as the structure of the schedule can be exploited for increased
efficiency.
Earlier experiments with schedule tree based dependence analysis in Polly showed
some compile-time regressions. These regressions arose due to the schedule tree
based dependence analysis not taking into account the domain constraints of the
schedule tree. As a result, the computed dependences were different and this
difference caused in some cases the schedule optimizer to take a very long time.
Since isl version fe865996 the schedule tree based dependence analysis takes
domain constraints into account, which fixes the earlier compile-time issues.
Contributed-by: Pratik Bhatu <cs12b1010@iith.ac.in>
llvm-svn: 245300
The new field in the MemoryAccess allows us to track a value related
to that access:
- For real memory accesses the value is the loaded result or the
stored value.
- For straigt line scalar accesses it is the access instruction
itself.
- For PHI operand accesses it is the operand value.
We use this value to simplify code which deduced information about the value
later in the Polly pipeline and was known to be error prone.
Reviewers: grosser, Meinsersbur
Subscribers: #polly
Differential Revision: http://reviews.llvm.org/D12062
llvm-svn: 245213
This option allows the user to provide additional information about parameter
values as an isl_set. To specify that N has the value 1024, we can provide
the context -polly-context='[N] -> {: N = 1024}'.
llvm-svn: 245175
This modifies the order in which Polly passes are executed.
Assuming a function has two scops (A and B), the order before was:
FunctionPassManager
ScopDetection
IndependentBlocks
TempScopInfo for A and B
RegionPassManager
ScopInfo for A
DependenceInfo for A
IslScheduleOptimizer for A
IslAstInfo for A
CodeGeneration for A
ScopInfo for B
DependenceInfo for B
IslScheduleOptimizer for B
IslAstInfo for B
CodeGeneration for B
After this patch:
FunctionPassManager
ScopDetection
IndependentBlocks
RegionPassManager
TempScopInfo for A
ScopInfo for A
DependenceInfo for A
IslScheduleOptimizer for A
IslAstInfo for A
CodeGeneration for A
TempScopInfo for B
ScopInfo for B
DependenceInfo for B
IslScheduleOptimizer for B
IslAstInfo for B
CodeGeneration for B
TempScopInfo for B might store information and references to the IR
that CodeGeneration for A might modify. Changing the order ensures that
the IR is not modified from the analysis of a region until code
generation.
Reviewers: grosser
Differential Revision: http://reviews.llvm.org/D12014
llvm-svn: 245091
This change has three major advantages:
- The ScopInfo becomes smaller.
- It allows to use the SCEVAffinator from outside the ScopInfo.
- A member object allows state which in turn allows e.g., caching.
Differential Revision: http://reviews.llvm.org/D9099
llvm-svn: 244730
Before we only modeled PHI nodes if at least one incoming basic block was itself
part of the region, now we always model them except if all of their operands are
part of a single non-affine subregion which we model as a black-box.
This change only affects PHI nodes in the entry block, that have exactly one
incoming edge. Before this change, we did not model them and as a result code
generation would not know how to code generate them. With this change, code
generation can code generate them like any other PHI node.
This issue was exposed by r244606. Before this change simplifyRegion would have
moved these PHI nodes out of the SCoP, so we would never have tried to code
generate them. We could implement this behavior again, but changing the IR
after the scop has been modeled and transformed always adds a risk of us
invalidating earlier analysis results. It seems more save and overall also more
consistent to just model and handle this one-entry-edge PHI nodes like any
other PHI node in the scop.
Solution proposed by: Michael Kruse <llvm@meinersbur.de>
llvm-svn: 244721
Summary: The extracted function buildBBScopStmt will be needed later to be invoked individually on the region's exit block.
Reviewers: grosser, jdoerfert
Subscribers: jdoerfert, llvm-commits, pollydev
Projects: #polly
Differential Revision: http://reviews.llvm.org/D11878
llvm-svn: 244443
Even though read-only accesses to scalars outside of a scop do not need to be
modeled to derive valid transformations or to generate valid sequential code,
but information about them is useful when we considering memory footprint
analysis and/or kernel offloading.
llvm-svn: 243981
If set, this option instructs -view-scops and -polly-show to only print
functions that contain the specified string in their name. This allows to
look at the scops of a specific function in a large .ll file, without flooding
the screen with .dot graphs.
llvm-svn: 243882
We use the branch instruction as the location at which a PHI-node write takes
place, instead of the PHI-node itself. This allows us to identify the
basic-block in a region statement which is on the incoming edge of the PHI-node
and for which the write access was originally introduced. As a result we can,
during code generation, avoid generating PHI-node write accesses for basic
blocks that do not preceed the PHI node without having to look at the IR
again.
This change fixes a bug which was introduced in r243420, when we started to
explicitly model PHI-node reads and writes, but dropped some additional checks
that where still necessary during code generation to not emit PHI-node writes
for basic-blocks that are not on incoming edges of the original PHI node.
Compared to the code before r243420 the new code does not need to inspect the IR
any more and we also do not generate multiple redundant writes.
llvm-svn: 243852
The schedule map we derive from a schedule tree map may map statements into
schedule spaces of different dimensionality. This change adds zero padding
to ensure just a single schedule space is used and the translation from
a union_map to an isl_multi_union_pw_aff does not fail.
llvm-svn: 243849
It is common practice to keep constructors lightweight. The reasons
include:
- The vtable during the constructor's execution is set to the static
type of the object, not to the vtable of the derived class. That is,
method calls behave differently in constructors and ordinary methods.
This way it is possible to call unimplemented methods of abstract
classes, which usually results in a segmentation fault.
- If an exception is thrown in the constructor, the destructor is not
called, potentially leaking memory.
- Code in constructors cannot be called in a regular way, e.g. from
non-constructor methods of derived classes.
- Because it is common practice, people may not expect the constructor
to do more than initializing data and skip them when looking for bugs.
Not all of these are applicable to LLVM (e.g. exceptions are disabled).
This patch refactors out the computational work in the constructors of
Scop and IslAst into regular init functions and introduces static
create-functions as replacement.
Differential revision: http://reviews.llvm.org/D11491
Reviewers: grosser, jdoerfert
llvm-svn: 243677
Such codes are not interesting to optimize and most likely never appear in the
normal compilation flow. However, they show up during test case reduction with
bugpoint and trigger -- without this change -- an assert in
polly::MemoryAccess::foldAccess(). It is better to detect them in
ScopDetection itself and just bail out.
Contributed-by: Utpal Bora <cs14mtech11017@iith.ac.in>
Reviewers: grosser
Subscribers: pollydev, llvm-commits
Differential Revision: http://reviews.llvm.org/D11425
llvm-svn: 243515
Summary:
When translating PHI nodes into memory dependences during code generation we
require two kinds of memory. 'Normal memory' as for all scalar dependences and
'PHI node memory' to store the incoming values of the PHI node. With this
patch we now mark and track these two kinds of memories, which we previously
incorrectly marked as a single memory object.
Being aware of PHI node storage makes code generation easier, as we do not need
to guess what kind of storage a scalar reference requires. This simplifies the
code nicely.
Reviewers: jdoerfert
Subscribers: pollydev, llvm-commits
Differential Revision: http://reviews.llvm.org/D11554
llvm-svn: 243420
As specified in PR23888, run-time alias check generation is expensive
in terms of compile-time. This reduces the compile time by computing
minimal/maximal access only once for each base pointer
Contributed-by: Pratik Bhatu <cs12b1010@iith.ac.in>
llvm-svn: 243024
Instead of flat schedules, we now use so-called schedule trees to represent the
execution order of the statements in a SCoP. Schedule trees make it a lot easier
to analyze, understand and modify properties of a schedule, as specific nodes
in the tree can be choosen and possibly replaced.
This patch does not yet fully move our DependenceInfo pass to schedule trees,
as some additional performance analysis is needed here. (In general schedule
trees should be faster in compile-time, as the more structured representation
is generally easier to analyze and work with). We also can not yet perform the
reduction analysis on schedule trees.
For more information regarding schedule trees, please see Section 6 of
https://lirias.kuleuven.be/handle/123456789/497238
llvm-svn: 242130
Named isl sets can generally have any name if they remain within Polly, but only
certain strings can be parsed by isl. The new names we create ensure that we
can always copy-past isl strings from Polly to other isl tools, e.g. for
debugging.
llvm-svn: 241787
This is very preliminary support, but it seems to work for the most common case.
When observing more/different test cases, we can work on generalizing this.
llvm-svn: 240955
As Polly got a lot faster after the small-integer-optimization imath
patch, we now increase the compute out to optimize larger kernels. This
should also expose additional slow-downs for us to address.
In LNT this gives us a 3.4x speedup on 3mm, at a cost of a 2x increase in
compile time (now 0.77s). reg_detect, oorafft and adi also show some compile
time increases. This compile time cost is divided between more time in isl and
more time in LLVM's backends due to increased code size (versioning and tiling).
llvm-svn: 240840
In case we have modulo operations in the access function (supported since
r240518), the assumptions generated to ensure array accesses remain within
bounds can contain existentially quantified dimensions which results in more
complex and more difficult to handle integer sets. As a result LNT's linpack
benchmark started to fail due to excessive compile time.
We now just drop the existentially quantified dimensions. This should be
generally save, but may result in less precise assumptions which may
consequently make us fall back to the original (unoptimized) code more often. In
practice, these cases probably do not appear to often.
I had difficulties to extract a good test case, but fortunately our LNT bots
cover this one well.
llvm-svn: 240775
This removes old code that has been disabled since several weeks and was hidden
behind the flags -disable-polly-intra-scop-scalar-to-array=false and
-polly-model-phi-nodes=false. Earlier, Polly used to translate scalars and
PHI nodes to single element arrays, as this avoided the need for their special
handling in Polly. With Johannes' patches adding native support for such scalar
references to Polly, this code is not needed any more. After this commit both
-polly-prepare and -polly-independent are now mostly no-ops. Only a couple of
simple transformations still remain, but they are scheduled for removal too.
Thanks again to Johannes Doerfert for his nice work in making all this code
obsolete.
llvm-svn: 240766
Remainder operations with constant divisor can be modeled as quasi-affine
expression. This patch adds support for detecting and modeling them. We also
add a test that ensures they are correctly code generated.
This patch was extracted from a larger patch contributed by Johannes Doerfert
in http://reviews.llvm.org/D5293
llvm-svn: 240518
David Blaikie:
"find returns an iterator by value, so it's just added complexity/strangeness to
then use reference lifetime extension to give it the same semantics as if you'd
used a value type instead of a reference type."
llvm-svn: 238294
David Blaike suggested this as an alternative to the use of owningptr(s) for our
memory management, as value semantics allow to avoid the additional interface
complexity caused by owningptr while still providing similar memory consistency
guarantees. We could also have used a std::vector, but the use of std::vector
would yield possibly changing pointers which currently causes problems as for
example the memory accesses carry pointers to their parent statements. Such
pointers should not change.
Reviewer: jblaikie, jdoerfert
Differential Revision: http://reviews.llvm.org/D10041
llvm-svn: 238290
The feature itself has been committed by Johannes in r238070. As this is the
way forward, we now enable it to ensure we get test coverage.
Thank you Johannes for this nice work!
llvm-svn: 238088
To reduce compile time and to allow more and better quality SCoPs in
the long run we introduced scalar dependences and PHI-modeling. This
patch will now allow us to generate code if one or both of those
options are set. While the principle of demoting scalars as well as
PHIs to memory in order to communicate their value stays the same,
this allows to delay the demotion till the very end (the actual code
generation). Consequently:
- We __almost__ do not modify the code if we do not generate code
for an optimized SCoP in the end. Thus, the early exit as well as
the unprofitable option will now actually preven us from
introducing regressions in case we will probably not get better
code.
- Polly can be used as a "pure" analyzer tool as long as the code
generator is set to none.
- The original SCoP is almost not touched when the optimized version
is placed next to it. Runtime regressions if the runtime checks
chooses the original are not to be expected and later
optimizations do not need to revert the demotion for that part.
- We will generate direct accesses to the demoted values, thus there
are no "trivial GEPs" that select the first element of a scalar we
demoted and treated as an array.
Differential Revision: http://reviews.llvm.org/D7513
llvm-svn: 238070
Tobias Grosser