OwningRewritePatternList has been deprecated for ~10 months now, we can remove
the leftover using directives at this point.
Differential Revision: https://reviews.llvm.org/D118287
These transformations already operate on memref operations (as part of
splitting up the standard dialect). Now that the operations have moved,
it's time for these transformations to move as well.
Differential Revision: https://reviews.llvm.org/D118285
We already convert to BitVector internally, and other APIs (namely Operation::eraseOperands)
already use BitVector as well. Switching over provides a common format between
API and also reduces the amount of format conversions necessary.
Fixes#53325
Differential Revision: https://reviews.llvm.org/D118083
The option parser currently does not properly handle nested options, meaning that
in some cases we can print pass pipelines that we can't actually parse back in. For example,
from #52885 we currently can't parse in inliner pipelines that have nested pipeline strings.
This commit adds handling for string values (e.g. "...") and nested options
(e.g. `foo{baz{bar=10 pizza=11}}`).
Fixes#52885
Differential Revision: https://reviews.llvm.org/D118078
Rationale:
Demonstrates the maximum tile size allowed for the f32 <= bf16 x bf16 op
Reviewed By: dcaballe
Differential Revision: https://reviews.llvm.org/D118277
This diff modifies the tablegen specification and code generation for
BitEnumAttr attributes in MLIR Operation Definition Specification (ODS) files.
Specifically:
- there is a new tablegen class for "none" values (i.e. no bits set)
- single-bit enum cases are specified via bit index (i.e. [0, 31]) instead of
the resulting enum integer value
- there is a new tablegen class to represent a "grouped" bitwise OR of other
enum values
This diff is intended as an initial step towards improving "fastmath"
optimization support in MLIR, to allow more precise control of whether certain
floating point optimizations are applied in MLIR passes. "Fast" math options
for floating point MLIR operations would (following subsequent RFC and
discussion) be specified by using the improved enum bit support in this diff.
For example, a "fast" enum value would act as an alias for a group of other
cases (e.g. finite-math-only, no-signed-zeros, etc.), in a way that is similar
to support in C/C++ compilers (clang, gcc).
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D117029
This is part of splitting up the standard dialect. The move makes sense anyways,
given that the memref dialect already holds memref.atomic_rmw which is the non-region
sibling operation of std.generic_atomic_rmw (the relationship is even more clear given
they have nearly the same description % how they represent the inner computation).
Differential Revision: https://reviews.llvm.org/D118209
The GPU dialect currently contains an explicit reference to LLVMFuncOp
during verification to handle the situation where the kernel has already been
converted. This commit changes that reference to instead use FunctionOpInterface,
which has two main benefits:
* It allows for removing an otherwise unnecessary dependency on the LLVM dialect
* It removes hardcoded assumptions about the lowering path and use of the GPU dialect
Differential Revision: https://reviews.llvm.org/D118172
This reverts commit ef82063207.
- It conflicts with the existing llvm::size in STLExtras, which will now
never be called.
- Calling it without llvm:: breaks C++17 compat
This is for compatibility with existing bufferization passes. Also clean up memref type generation a bit.
Differential Revision: https://reviews.llvm.org/D118243
This revision adds enough support to allow InlineAsmOp to work properly with indirect memory constraints "*m".
These require an explicit "elementtype" TypeAttr on the operands to pass LLVM verification and need to be provided.
Reviewed By: bkramer
Differential Revision: https://reviews.llvm.org/D118006
The existing implementation called DenseMap::insert, which is a no-op if the map already contains an entry with the same key.
Differential Revision: https://reviews.llvm.org/D118165
If we are extracting it is more useful to push the index_cast past the
extraction. This increases the chance the tensor.extract can evaluated at
compile time.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D118204
There is not much of a benefit to reshape a from element vs reloading it.
Updated to progagate shape manipulations into the output type of
tensor.from_elements.
Reviewed By: NatashaKnk
Differential Revision: https://reviews.llvm.org/D118201
Add methods to
- Get block argument that is tied with an opOperand
- Get the yield value that is tied with a output opOperand.
Differential Revision: https://reviews.llvm.org/D118085
Fusion of reshape ops by linearization incorrectly inverted the
indexing map before linearizing dimensions. This leads to incorrect
indexing maps used in the fused operation.
Differential Revision: https://reviews.llvm.org/D117908
This pattern is not written to handle operations with `linalg.index`
operations in its body, i.e. operations that have index semantics.
Differential Revision: https://reviews.llvm.org/D117856
This is mostly a copy of the existing tensor.from_elements bufferization. Once TensorInterfaceImpl.cpp is moved to the tensor dialect, the existing rewrite pattern can be deleted.
Differential Revision: https://reviews.llvm.org/D117775
This is mostly a copy of the existing tensor.generate bufferization. Once TensorInterfaceImpl.cpp is moved to the tensor dialect, the existing rewrite pattern can be deleted.
Differential Revision: https://reviews.llvm.org/D117770
This patch supports the atomic construct (capture) following section 2.17.7 of OpenMP 5.0 standard. Also added tests for the same.
Reviewed By: peixin, kiranchandramohan
Differential Revision: https://reviews.llvm.org/D115851
This is a pretty important debugging option to stay hidden. Also,
improve its cmd-line description; the current description gives no hint
that this is the one to use to have locations printed inline.
Out-of-line locations are also unproductive to work with in many cases
where the locations are actually compact, which is also why this option
should be more visible. This revision doesn't change the default on it
though.
Reviewed By: rriddle, jpienaar
Differential Revision: https://reviews.llvm.org/D117186
A lot of dialects have dependencies that are unnecessary, either because of copy/paste
of files when creating things or some other means. This commit cleans up a bunch of
the simple ones:
* Copy/Paste or missed during refactoring
Most of the dependencies cleaned up here look like copy/paste errors when creating
new dialects/transformations, or because the dependency wasn't removed during a
refactoring (e.g. when splitting the standard dialect).
* Unnecessary hard coding of constant operations in matchers
There are a few instances where a dialect had a dependency because it
was hardcoding checks for constant operations instead of using the better m_Constant
approach.
Differential Revision: https://reviews.llvm.org/D118062
This has been a major TODO for a very long time, and is necessary for establishing a proper
dialect-free dependency layering for the Transforms library. Code was moved to effectively
two main locations:
* Affine/
There was quite a bit of affine dialect related code in Transforms/ do to historical reasons
(of a time way into MLIR's past). The following headers were moved to:
Transforms/LoopFusionUtils.h -> Dialect/Affine/LoopFusionUtils.h
Transforms/LoopUtils.h -> Dialect/Affine/LoopUtils.h
Transforms/Utils.h -> Dialect/Affine/Utils.h
The following transforms were also moved:
AffineLoopFusion, AffinePipelineDataTransfer, LoopCoalescing
* SCF/
Only one SCF pass was in Transforms/ (likely accidentally placed here): ParallelLoopCollapsing
The SCF specific utilities in LoopUtils have been moved to SCF/Utils.h
* Misc:
mlir::moveLoopInvariantCode was also moved to LoopLikeInterface.h given
that it is a simple utility defined in terms of LoopLikeOpInterface.
Differential Revision: https://reviews.llvm.org/D117848
Transforms/ should only contain transformations that are dialect-independent and
this pass interacts with MemRef operations (making it a better fit for living in that
dialect).
Differential Revision: https://reviews.llvm.org/D117841
Transforms/ should only contain dialect-independent transformations,
and these files are a much better fit for the bufferization dialect anyways.
Differential Revision: https://reviews.llvm.org/D117839
The current lowering from GPU to NVVM does
not correctly handle the following cases when
lowering the gpu shuffle op.
1. When the active width is set to 32 (all lanes),
then the current approach computes (1 << 32) -1 which
results in poison values in the LLVM IR. We fix this by
defining the active mask as (-1) >> (32 - width).
2. In the case of shuffle up, the computation of the third
operand c has to be different from the other 3 modes due to
the op definition in the ISA reference.
(https://docs.nvidia.com/cuda/parallel-thread-execution/index.html)
Specifically, the predicate value is computed as j >= maxLane
for up and j <= maxLane for all other modes. We fix this by
computing maskAndClamp as 32 - width for this mode.
TEST: We modify the existing test and add more checks for the up mode.
Reviewed By: ThomasRaoux
Differential Revision: https://reviews.llvm.org/D118086
Adding a similar decomposition for exponential minus one to the SPIRV
backends along with the necessary tests.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D118081
Control-Flow Sink moves operations whose only uses are in conditionally-executed regions into those regions so that paths in which their results are not needed do not perform unnecessary computation.
Depends on D115087
Reviewed By: jpienaar, rriddle, bondhugula
Differential Revision: https://reviews.llvm.org/D115088
Add a transpose option to hoist padding to transpose the padded tensor before storing it into the packed tensor. The early transpose improves the memory access patterns of the actual compute kernel. The patch introduces a transpose right after the hoisted pad tensor and a second transpose inside the compute loop. The second transpose can either be fused into the compute operation or will canonicalize away when lowering to vector instructions.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D117893
No longer go through an external model. Also put BufferizableOpInterface into the same build target as the BufferizationDialect. This allows for some code reuse between BufferizationOps canonicalizers and BufferizableOpInterface implementations.
Differential Revision: https://reviews.llvm.org/D117987
Both insertion points are valid. This is to make BufferizableOpInteface-based bufferization compatible with existing partial bufferization test cases. (So less changes are necessary to unit tests.)
Differential Revision: https://reviews.llvm.org/D117986
This is the only op that is not supported via BufferizableOpInterfaceImpl bufferization. Once this op is supported we can switch `tensor-bufferize` over to the new unified bufferization.
Differential Revision: https://reviews.llvm.org/D117985
This is in preparation of unifying the existing bufferization with One-Shot bufferization.
A subsequent commit will replace `tensor-bufferize`'s implementation with the BufferizableOpInterface-based implementation and move over missing test cases.
Differential Revision: https://reviews.llvm.org/D117984
Only using that change in StringRef already decreases the number of
preoprocessed lines from 7837621 to 7776151 for LLVMSupport
Perhaps more interestingly, it shows that many files were relying on the
inclusion of StringRef.h to have the declaration from STLExtras.h. This
patch tries hard to patch relevant part of llvm-project impacted by this
hidden dependency removal.
Potential impact:
- "llvm/ADT/StringRef.h" no longer includes <memory>,
"llvm/ADT/Optional.h" nor "llvm/ADT/STLExtras.h"
Related Discourse thread:
https://llvm.discourse.group/t/include-what-you-use-include-cleanup/5831
This patch moves merging of duplicate divisions to presburger utility
functions. This is required to support division merging in structures other
than IntegerPolyhedron.
Reviewed By: arjunp
Differential Revision: https://reviews.llvm.org/D118001
Pass a ValueRange instead of an ArrayRef<Value> for better compatibility. Also provide an additional function overload that automatically deallocates the buffer if specified.
Differential Revision: https://reviews.llvm.org/D118025
This patch changes names of identifiers and their corresponding getters in
PresburgerSet to match those of IntegerPolyhedron.
Reviewed By: arjunp
Differential Revision: https://reviews.llvm.org/D117998
When the coefficients of dividend are negative, the gcd may be negative
which will change the sign of dividend and overflow denominator.
Reviewed By: Groverkss
Differential Revision: https://reviews.llvm.org/D117911
When 2 clamp ops are in a row, they can be canonicalized into a single clamp
that uses the most constrained range
Reviewed By: rsuderman
Differential Revision: https://reviews.llvm.org/D117934
Rationale:
Although file I/O is a bit alien to MLIR itself, we provide two convenient ways
for sparse tensor I/O. The input part was already there (behind the swiss army
knife sparse_tensor.new). Now we have a sparse_tensor.out to write out data. As
before, the ops are kept vague and may change in the future. For now this
allows us to compare TACO vs MLIR very easily.
Reviewed By: bixia
Differential Revision: https://reviews.llvm.org/D117850
Implement a taylor series approximation for atan and add an atan2 lowering
that uses atan's appromation. This includes tests for edge cases and tests
for each quadrant.
Reviewed By: NatashaKnk
Differential Revision: https://reviews.llvm.org/D115682
Add TACO tests to test/Integration/Dialect/SparseTensor/taco. Add the MLIR
PyTACO implementation as tools under the directory.
Reviewed By: aartbik, mehdi_amini
Differential Revision: https://reviews.llvm.org/D117260
The cleanup was manual, but assisted by "include-what-you-use". It consists in
1. Removing unused forward declaration. No impact expected.
2. Removing unused headers in .cpp files. No impact expected.
3. Removing unused headers in .h files. This removes implicit dependencies and
is generally considered a good thing, but this may break downstream builds.
I've updated llvm, clang, lld, lldb and mlir deps, and included a list of the
modification in the second part of the commit.
4. Replacing header inclusion by forward declaration. This has the same impact
as 3.
Notable changes:
- llvm/Support/TargetParser.h no longer includes llvm/Support/AArch64TargetParser.h nor llvm/Support/ARMTargetParser.h
- llvm/Support/TypeSize.h no longer includes llvm/Support/WithColor.h
- llvm/Support/YAMLTraits.h no longer includes llvm/Support/Regex.h
- llvm/ADT/SmallVector.h no longer includes llvm/Support/MemAlloc.h nor llvm/Support/ErrorHandling.h
You may need to add some of these headers in your compilation units, if needs be.
As an hint to the impact of the cleanup, running
clang++ -E -Iinclude -I../llvm/include ../llvm/lib/Support/*.cpp -std=c++14 -fno-rtti -fno-exceptions | wc -l
before: 8000919 lines
after: 7917500 lines
Reduced dependencies also helps incremental rebuilds and is more ccache
friendly, something not shown by the above metric :-)
Discourse thread on the topic: https://llvm.discourse.group/t/include-what-you-use-include-cleanup/5831
The LangRef currently lacks a top-level production, leaving the productions attribute-alias-def and type-alias-defunused. Clarify the situation by declaring what is to be parsed by an MLIR parser at the toplevel.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D117668
In some cases, fusion can produce illegal operations if after fusion
the range of some of the loops cannot be computed from shapes of its
operands. Check for this case and abort the fusion if this happens.
Differential Revision: https://reviews.llvm.org/D117602
This is superseded by the same method on OpAsmOpInterface, which is
available on the Dialect through the Fallback mechanism,
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D117750
This extends dense attribute element access to support 8b and 16b ints.
Also extends the corresponding parts of the C api.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D117731
This allows to pipe sequences of `mlir-opt -split-input-file | mlir-opt -split-input-file`.
Depends On D117750
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D117756
Right shift can occur that is a 32-bit right shift. This is undefined behavior.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D117732
- Set the DEBUG_TYPE of SerializeToBlob to serialize-to-blob
- Add debug output to print the assembly or PTX for GPU modules before
they are assembled and linked
Note that, as SerializeToBlob is a superclass of SerializeToCubin and
SerializeToHsaco, --debug-only=serialize-to-blom will dump the
intermediate compiler result for both of these passes.
In addition, if LLVM options such as --stop-after are used to control
the GPU kernel compilation process, the debug output will contain the
appropriate intermediate IR.
Reviewed By: herhut
Differential Revision: https://reviews.llvm.org/D117519
PDLDialect being a somewhat user-facing dialect and whose ops contain exclusively other PDL ops in their regions can take advantage of `OpAsmOpInterface` to provide nicer IR.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D117828
Unbound OperationOp in the matcher (i.e. one with no uses) is already disallowed by the verifier. However, an OperationOp in the rewriter is not side-effect free -- it's creating an op!
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D117825
Querying or building constraints on types, operands, results, and attributes are side-effect free in both the matcher and rewriter. The ops should be marked as such.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D117826
Its defaulting logic must go after `project(..)` to work correctly, but `project(..)` is often in a standalone condition making this
awkward, since the rest of the condition code may also need GNUInstallDirs.
The good thing is there are the various standalone booleans, which I had missed before. This makes splitting the conditional blocks less awkward.
Reviewed By: arichardson, phosek, beanz, ldionne, #libunwind, #libc, #libc_abi
Differential Revision: https://reviews.llvm.org/D117639
This commits explicitly states that negative values and values exceeding
vector dimensions are allowed in vector.create_mask (but not in
vector.constant_mask). These values are now truncated when
canonicalizing vector.create_mask to vector.constant_mask.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D116069
When a Builder methods accepts multiple InsertPoints, when both point to
the same position, inserting instructions at one position will "move" the
other after the inserted position since the InsertPoint is pegged to the
instruction following the intended InsertPoint. For instance, when
creating a parallel region at Loc and passing the same position as AllocaIP,
creating instructions at Loc will "move" the AllocIP behind the Loc
position.
To avoid this ambiguity, add an assertion checking this condition and
fix the unittests.
In case of AllocaIP, an alternative solution could be to implicitly
split BasicBlock at InsertPoint, using the first as AllocaIP, the second
for inserting the instructions themselves. However, this solution is
specific to AllocaIP since AllocaIP will always have to be first. Hence,
this is an argument to generally handling ambiguous InsertPoints as API
sage error.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D117226
When computing the new type of a collapse_shape operation, we need to at least
take into account whether the type has an identity layout, in which case we can
easily support dynamic strides. Otherwise, the canonicalizer creates invalid
IR.
Longer term, both the verifier and the canoncializer need to be extended to
support the general case.
Differential Revision: https://reviews.llvm.org/D117772
Earlier `computeSingleVarRepr` was returning a pair of upper bound and
lower bound indices of the inequality contraints that can be expressed
as a floordiv of an affine function. The equality expression can also be
expressed as a floordiv but contains only one index and hence the `LocalRepr`
class is introduced to facilitate this.
Reviewed By: Groverkss
Differential Revision: https://reviews.llvm.org/D117430
This commit is the first step towards unifying core bufferization and One-Shot Bufferize.
This commit does not move over the implementations of BufferizableOpInterface yet. This will be done in separate commits. This change does also not move the unit tests yet. The tests will be moved together with op interface implementations and split into separate files.
Differential Revision: https://reviews.llvm.org/D117641
BlockArguments gained the ability to have locations attached a while ago, but they
have always been optional. This goes against the core tenant of MLIR where location
information is a requirement, so this commit updates the API to require locations.
Fixes#53279
Differential Revision: https://reviews.llvm.org/D117633
Previously the optional locations of function arguments were dropped in
`parseFunctionArgumentList`. This CL adds another output argument to the
function through which they are now returned. The values are then plumbed
through as an array of optional locations in the various places.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D117604
When the printer is requested to elide large constant, we emit an opaque
attribute instead. This patch fills the dialect name with
"elided_large_const" instead of "_" to remove some user confusion when
they later try to consume it.
Differential Revision: https://reviews.llvm.org/D117711
The goal is to allow a project to directly include the `CMakeLists.txt` of mlir via something such as `add_directory`. This currently doesn't work because the exported targets then end up being imported in the same top-level project and `cmake` is not happy. LLVM works around this by guarding the exports based on the existence of the `LLVMSupport` library. This change does the same using `MLIRSupport`. In our experience, no targets need to be added to the export in the case.
If there's a preferred way of doing this or a better lib to use as the main target to check for, please let me know.
Reviewed By: stellaraccident
Differential Revision: https://reviews.llvm.org/D117599
Benchmarks show that memref::CopyOp is curently up to 200x slower than
tiled and vectorized versions of linalg::Copy.
Add a temporary flag to allow comprehensive bufferize to generate a
linalg::GenericOp that implements a copy until this performance bug is
resolved.
Differential Revision: https://reviews.llvm.org/D117696
The constructor function was being defined without indicating its "__init__"
name, which made it interpret it as a regular fuction rather than a
constructor. When overload resolution failed, Pybind would attempt to print the
arguments actually passed to the function, including "self", which is not
initialized since the constructor couldn't be called. This would result in
"__repr__" being called with "self" referencing an uninitialized MLIR C API
object, which in turn would cause undefined behavior when attempting to print
in C++. Even if the correct name is provided, the mechanism used by
PybindAdaptors.h to bind constructors directly as "__init__" functions taking
"self" is deprecated by Pybind. The new mechanism does not seem to have access
to a fully-constructed "self" object (i.e., the constructor in C++ takes a
`pybind11::detail::value_and_holder` that cannot be forwarded back to Python).
Instead, redefine "__new__" to perform the required checks (there are no
additional initialization needed for attributes and types as they are all
wrappers around a C++ pointer). "__new__" can call its equivalent on a
superclass without needing "self".
Bump pybind11 dependency to 3.8.0, which is the first version that allows one
to redefine "__new__".
Reviewed By: stellaraccident
Differential Revision: https://reviews.llvm.org/D117646
The code in `BufferizableOpInterface`'s header/source no longer contains any analysis code. This makes it easier to run the bufferization with a different analysis or without any analysis.
Differential Revision: https://reviews.llvm.org/D117478
This separates the analysis (and its helpers/data structures) more clearly from the rest of the bufferization.
Differential Revision: https://reviews.llvm.org/D117477
This change adds full python bindings for PDL, including types and operations
with additional mixins to make operation construction more similar to the PDL
syntax.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D117458
Also move `createAlloc` and related helper functions out of BufferizationState. The goal is to make BufferizationState as small as possible. (Code cleanup)
Differential Revision: https://reviews.llvm.org/D117476
If not allow-return-memref, raise an error if a new memory allocation is returned/yielded from a block. We do not check for new allocations directly, but for ops that yield/return values that are not equivalent to values that are defined outside of the current of the block.
Note: We still need to check that scf.for yield values and bbArgs are aliasing to ensure that getAliasingOpOperand/getAliasingOpResult is correct.
Differential Revision: https://reviews.llvm.org/D116687
This op is needed for unit testing in a subsequent revision. (This is the first op that has a block that yields equivalent values via the op's results.)
Note: Bufferization of scf.execute_region ops with multiple blocks is not yet supported.
Differential Revision: https://reviews.llvm.org/D117424
When the min/max are the total range of the value, it is a no-op as the values
are already restricted to that range.
Reviewed By: rsuderman
Differential Revision: https://reviews.llvm.org/D117625
Try to fix the error:
mlir/lib/Parser/Parser.cpp:553:14: error: cannot call member function 'mlir::InFlightDiagnostic mlir::detail::Parser::emitError(llvm::SMLoc, const llvm::Twine&)' without object
<< "operation location alias was never defined";
This commit refactors the FunctionLike trait into an interface (FunctionOpInterface).
FunctionLike as it is today is already a pseudo-interface, with many users checking the
presence of the trait and then manually into functionality implemented in the
function_like_impl namespace. By transitioning to an interface, these accesses are much
cleaner (ideally with no direct calls to the impl namespace outside of the implementation
of the derived function operations, e.g. for parsing/printing utilities).
I've tried to maintain as much compatability with the current state as possible, while
also trying to clean up as much of the cruft as possible. The general migration plan for
current users of FunctionLike is as follows:
* function_like_impl -> function_interface_impl
Realistically most user calls should remove references to functions within this namespace
outside of a vary narrow set (e.g. parsing/printing utilities). Calls to the attribute name
accessors should be migrated to the `FunctionOpInterface::` equivalent, most everything
else should be updated to be driven through an instance of the interface.
* OpTrait::FunctionLike -> FunctionOpInterface
`hasTrait` checks will need to be moved to isa, along with the other various Trait vs
Interface API differences.
* populateFunctionLikeTypeConversionPattern -> populateFunctionOpInterfaceTypeConversionPattern
Fixes#52917
Differential Revision: https://reviews.llvm.org/D117272
The chunk size in schedule clause is one integer expression, which can
be either constant integer or integer variable. Fix schedule clause in
MLIR Op Def to support integer expression with different bit width.
Reviewed By: shraiysh
Differential Revision: https://reviews.llvm.org/D116073
The only benefit of FunctionPass is that it filters out function
declarations. This isn't enough to justify carrying it around, as we can
simplify filter out declarations when necessary within the pass. We can
also explore with better scheduling primitives to filter out declarations
at the pipeline level in the future.
The definition of FunctionPass is left intact for now to allow time for downstream
users to migrate.
Differential Revision: https://reviews.llvm.org/D117182
The input type of a linalg.generic can be less dynamic than its output
type. If this is the case moving a reshape across the generic op would
create invalid IR, as expand_shape cannot expand arbitrary dynamic
dimensions.
Check that the reshape is actually valid before creating the
expand_shape. This exposes the existing verification logic in reshape
utils and removes the incomplete custom implementation in fusion.
Differential Revision: https://reviews.llvm.org/D116600
Before this patch, deferred location in operation like `test.pretty_printed_region` would
break the parser, and so the IR round-trip.
Depends On D117088
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D117413
This will allow to return to the client of `parseLocationInstance` a location that is resolved later.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D117088
The current state of the top level Analysis/ directory is that it contains two libraries;
a generic Analysis library (free from dialect dependencies), and a LoopAnalysis library
that contains various analysis utilities that originated from Affine loop transformations.
This commit moves the LoopAnalysis to the more appropriate home of `Dialect/Affine/Analysis/`,
given the use and intention of the majority of the code within it. After the move, if there
are generic utilities that would fit better in the top-level Analysis/ directory, we can move
them.
Differential Revision: https://reviews.llvm.org/D117351
The leading space that is always printed at the beginning of regions is not consistent with other parts of the printing API. Moreover, this leading space can lead to undesirable assembly formats:
```
attr-dict-with-keyword $region
```
Prints as:
```
// Two spaces between `}` and `{`
attributes {foo} { ... }
```
Moreover, the leading space results in the odd generic op format:
```
"test.op"() ( {...}) : () -> ()
```
Reviewed By: rriddle, mehdi_amini
Differential Revision: https://reviews.llvm.org/D117411
The translation from the MLIR LLVM dialect to LLVM IR includes a mechanism that
ensures the successors of a block to be different blocks in case block
arguments are passed to them since the opposite cannot be expressed in LLVM IR.
This mechanism previously only worked for functions because it was written
prior to the introduction of other region-carrying operations such as the
OpenMP dialect, which also translates directly to LLVM IR. Modify this
mechanism to handle all regions in the module and not only functions.
Reviewed By: wsmoses
Differential Revision: https://reviews.llvm.org/D117548
The constructor function was being defined without indicating its "__init__"
name, which made it interpret it as a regular fuction rather than a
constructor. When overload resolution failed, Pybind would attempt to print the
arguments actually passed to the function, including "self", which is not
initialized since the constructor couldn't be called. This would result in
"__repr__" being called with "self" referencing an uninitialized MLIR C API
object, which in turn would cause undefined behavior when attempting to print
in C++.
Fix this by specifying the correct name.
This in turn uncovers the fact the the mechanism used by PybindAdaptors.h to
bind constructors directly as "__init__" functions taking "self" is deprecated
by Pybind. Instead, leverage the fact that the adaptors are intended for
attrbutes/types that cannot have additional data members and are all ultimately
instances of "PyAttribute"/"PyType" C++ class. In constructors of derived
classes, construct an instance of the base class first, then steal its internal
pointer to the C++ object to construct the instance of the derived class.
On top of that, the definition of the function was incorrectly indicated as the
method on the "None" object instead of being the method of its parent class.
This would result in a second problem when Pybind would attempt to print
warnings pointing to the parent class since the "None" does not have a
"__name__" field or its C API equivalent.
Fix this by specifying the correct parent class by looking it up by name in the
parent module.
Reviewed By: stellaraccident
Differential Revision: https://reviews.llvm.org/D117325
Also make the ODS Operator class have const iterator, and use const
references for existing API taking Operator by reference.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D117516
The majority of dialects reimplement the same boilerplate over and over,
switching the default makes it for better discoverability and make it simpler
to implement new dialects.
Differential Revision: https://reviews.llvm.org/D117524
Coroutine lowering always takes the natural alignment when spilling to
the frame (issue #53148) so using AVX2 or AVX512 in a coroutine doesn't
work. Always overalign to 64 bytes to avoid this issue until we have a
better solution.
Differential Revision: https://reviews.llvm.org/D117501
This revision fixes a bug where the iterative algorithm would walk back def-use chains to an incorrect operand.
This exposed opportunities for a larger refactoring and behavior improvement.
The new algorithm has improved folding behavior and proceeds by tracking both the
permutation of the extraction position and the internal vector permutation.
Multiple partial intersection cases with a candidate insertOp are supported.
The refactoring of the implementation should also help it generalize to strided insert/extract op.
This also subsumes the previous `foldExtractOpFromTranspose` which is now a simple special case and can be deleted.
Differential Revision: https://reviews.llvm.org/D117322
In some cases, the result of an initTensorOp may have an attribute.
However, the Attribute was not passed to `inferResultType`, failing the
verifier. Therefore, propagate the Attribute to `inferResultType`.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D117192
The constructor function was being defined without indicating its "__init__"
name, which made it interpret it as a regular fuction rather than a
constructor. When overload resolution failed, Pybind would attempt to print the
arguments actually passed to the function, including "self", which is not
initialized since the constructor couldn't be called. This would result in
"__repr__" being called with "self" referencing an uninitialized MLIR C API
object, which in turn would cause undefined behavior when attempting to print
in C++.
Fix this by specifying the correct name.
This in turn uncovers the fact the the mechanism used by PybindAdaptors.h to
bind constructors directly as "__init__" functions taking "self" is deprecated
by Pybind. The modern method requires using "py::init", which seems to rely on
the C++ equivalent of the bound class to be available, which is not the case in
PybindAdaptors.h. A deeper inspection shows that the deprecation concerns
old-style pybind11 constructors that had to allocate the object using
placement new with "self" as memory. The PybindAdaptors.h only provides
extension classes and never allocates (the object construction is delegated to
the base class), so it does not use the deprecated functionality. Use the
implementation detail tag class to convince pybind11 that we are using the
modern constructor binding method and suppress the warning.
On top of that, the definition of the function was incorrectly indicated as the
method on the "None" object instead of being the method of its parent class.
This would result in a second problem when Pybind would attempt to print
warnings pointing to the parent class since the "None" does not have a
"__name__" field or its C API equivalent.
Fix this by specifying the correct parent class by looking it up by name in the
parent module.
Reviewed By: stellaraccident
Differential Revision: https://reviews.llvm.org/D117325
This is the original patch in my GNUInstallDirs series, now last to merge as the final piece!
It arose as a new draft of D28234. I initially did the unorthodox thing of pushing to that when I wasn't the original author, but since I ended up
- Using `GNUInstallDirs`, rather than mimicking it, as the original author was hesitant to do but others requested.
- Converting all the packages, not just LLVM, effecting many more projects than LLVM itself.
I figured it was time to make a new revision.
I have used this patch series (and many back-ports) as the basis of https://github.com/NixOS/nixpkgs/pull/111487 for my distro (NixOS), which was merged last spring (2021). It looked like people were generally on board in D28234, but I make note of this here in case extra motivation is useful.
---
As pointed out in the original issue, a central tension is that LLVM already has some partial support for these sorts of things. Variables like `COMPILER_RT_INSTALL_PATH` have already been dealt with. Variables like `LLVM_LIBDIR_SUFFIX` however, will require further work, so that we may use `CMAKE_INSTALL_LIBDIR`.
These remaining items will be addressed in further patches. What is here is now rote and so we should get it out of the way before dealing more intricately with the remainder.
Reviewed By: #libunwind, #libc, #libc_abi, compnerd
Differential Revision: https://reviews.llvm.org/D99484
The execution engine would not be functional anyway, we're already
disabling the tests, this also disable the rest of the code.
Anecdotally this reduces the number of static library built when the
builtin target is disabled goes from 236 to 218.
Here is the complete list of LLVM targets built when running
`ninja check-mlir`:
libLLVMAggressiveInstCombine.a
libLLVMAnalysis.a
libLLVMAsmParser.a
libLLVMBinaryFormat.a
libLLVMBitReader.a
libLLVMBitstreamReader.a
libLLVMBitWriter.a
libLLVMCore.a
libLLVMDebugInfoCodeView.a
libLLVMDebugInfoDWARF.a
libLLVMDemangle.a
libLLVMFileCheck.a
libLLVMFrontendOpenMP.a
libLLVMInstCombine.a
libLLVMIRReader.a
libLLVMMC.a
libLLVMMCParser.a
libLLVMObject.a
libLLVMProfileData.a
libLLVMRemarks.a
libLLVMScalarOpts.a
libLLVMSupport.a
libLLVMTableGen.a
libLLVMTableGenGlobalISel.a
libLLVMTextAPI.a
libLLVMTransformUtils.a
Differential Revision: https://reviews.llvm.org/D117287
This is the original patch in my GNUInstallDirs series, now last to merge as the final piece!
It arose as a new draft of D28234. I initially did the unorthodox thing of pushing to that when I wasn't the original author, but since I ended up
- Using `GNUInstallDirs`, rather than mimicking it, as the original author was hesitant to do but others requested.
- Converting all the packages, not just LLVM, effecting many more projects than LLVM itself.
I figured it was time to make a new revision.
I have used this patch series (and many back-ports) as the basis of https://github.com/NixOS/nixpkgs/pull/111487 for my distro (NixOS), which was merged last spring (2021). It looked like people were generally on board in D28234, but I make note of this here in case extra motivation is useful.
---
As pointed out in the original issue, a central tension is that LLVM already has some partial support for these sorts of things. Variables like `COMPILER_RT_INSTALL_PATH` have already been dealt with. Variables like `LLVM_LIBDIR_SUFFIX` however, will require further work, so that we may use `CMAKE_INSTALL_LIBDIR`.
These remaining items will be addressed in further patches. What is here is now rote and so we should get it out of the way before dealing more intricately with the remainder.
Reviewed By: #libunwind, #libc, #libc_abi, compnerd
Differential Revision: https://reviews.llvm.org/D99484
Rather than hardcoding all constants, we now use the input tensor to drive the
code setup. Of course, we still need to hardcode dim-2 of A and the final
verification in CHECK is input dependent, but overall this sets a slightly
better example of tensor setup in general.
Reviewed By: bixia
Differential Revision: https://reviews.llvm.org/D117349
`getNumRegionInvocations` was originally added for the async reference counting, but turned out to be not useful, and currently is not used anywhere (couldn't find any uses in public github repos). Removing dead code.
Reviewed By: Mogball, mehdi_amini
Differential Revision: https://reviews.llvm.org/D117347
- Generic visitors invoke operation callbacks before/in-between/after visiting the regions
attached to an operation and use a `WalkStage` to indicate which regions have been
visited.
- This can be useful for cases where we need to visit the operation in between visiting
regions attached to the operation.
Differential Revision: https://reviews.llvm.org/D116230
Clean up return value on affineDataCopyGenerate utility. Return the
actual success/failure status instead of the "number of bytes" which
isn't being used in the codebase in any way. The success/failure status
wasn't being sent out earlier.
Differential Revision: https://reviews.llvm.org/D117209
By default, copies are inserted right before the tensor OpOperand use. With this change, `bufferize` implementation can change the insertion point. This is needed for some ops where it would be illegal to insert a copy right before the use.
Differential Revision: https://reviews.llvm.org/D117291
LLVM dialect supports terminators with repeated successor blocks that take
different operands. This cannot be directly expressed in LLVM IR though since
it uses the number of the predecessor block to differentiate values in its PHI
nodes. Therefore, the translation to LLVM IR inserts dummy blocks to forward
arguments in case of repeated succesors with arguments. The insertion works
correctly. However, when connecting PHI nodes to their source values, the
assertion of the insertion having worked correctly was incorrect: it would only
trigger if repeated blocks were adjacent in the successor list (not guaranteed
by anything) and would not check if the successors have operands (no need for
dummy blocks in absence of operands since no PHIs are being created). Change
the assertion to only trigger in case of duplicate successors with operands,
and don't expect them to be adjacent.
Reviewed By: wsmoses
Differential Revision: https://reviews.llvm.org/D117214
When constructing an OperationName, the overwhelming majority of
cases are from registered operations. This revision adds a non-locked
lookup into the currently registered operations, which prevents locking
in the common case. This revision also optimizes several uses of
RegisteredOperationName that expect the operation to be registered,
e.g. such as in OpBuilder.
These changes provides a reasonable speedup (5-10%) in some
compilations, especially on platforms where locking is expensive.
Differential Revision: https://reviews.llvm.org/D117187
This guarantees the preconditions of fromCOO; whereas prior to this, one could call the constructor directly with an unsorted tensor, which would cause fromCOO to misbehave.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D117167
OpenMP runtime requires depend vec with i64 type and the alignment of
store instruction should be set as 8.
Reviewed By: kiranchandramohan, shraiysh
Differential Revision: https://reviews.llvm.org/D116300
(1) copy-and-past error in encoding alias name:
this is an annotation for a tensor (3-d) not a matrix (2-d).
(2) typo in "initialization"
Reviewed By: bixia
Differential Revision: https://reviews.llvm.org/D117255
Add TACO tests to test/Integration/Dialect/SparseTensor/taco. Add the MLIR
PyTACO implementation as tools under the directory.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D117126
The names of the generated attribute getters for ops changed some time ago. The method created from the attribute name returns the return type and an additional method of the same name with Attr as suffix is generated which returns the actual attribute as its storage type.
The code generating effects however was using the methods without the Attr suffix, which is a problem in the case of FlatSymbolRefAttr as it has a return type of llvm::StringRef. This would lead to compilation errors as the constructor of SideEffects::EffectInstance expects a SymbolRefAttr in this case.
This patch simply fixes the generated effects code to use the Attr suffixed getter to get the actual storage type of the attribute.
Differential Revision: https://reviews.llvm.org/D117194
Adding the optional decompositions have been verified to improve memory
usage on common models. Added the decomposition to the default tosa to linalg
passes.
Reviewed By: NatashaKnk
Differential Revision: https://reviews.llvm.org/D117175
Spotted this in a final grep of projects I don't usually build before
pushing https://reviews.llvm.org/D115380, which makes
`SmallVector::set_size()` private.
Update to `truncate()`, a new-ish variant of `resize()` that asserts the
new size is not bigger and that avoids pulling in the allocation and
initialization code for growing. Doesn't really look like the perf
impact of that would matter here, but since `dirLength` is known to be a
smaller size then we might as well.
Differential Revision: https://reviews.llvm.org/D117073
Enable ReassociatingReshapeOpConversion with "non-identity" layouts.
This removes an early-return in this function, which seems unnecessary and is
preventing some memref.collapse_shape from converting to LLVM (see included lit test).
It seems unnecessary because the return message says "only empty layout map is supported"
but there actually is code in this function to deal with non-empty layout maps. Maybe
it refers to an earlier state of implementation and is just out of date?
Though, there is another concern about this early return: the condition that it actually
checks, `{src,dst}MemrefType.getLayout().isIdentity()`, is not quite the same as what the
return message says, "only empty layout map is supported". Stepping through this
`getLayout().isIdentity()` code in GDB, I found that it evaluates to `.getAffineMap().isIdentity()`
which does (AffineMap.cpp:271):
```
if (getNumDims() != getNumResults())
return false;
```
This seems that it would always return false for memrefs of rank greater than 1 ?
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D114808
LLVM Dialect Constant Op translations assume that if the attribute is a
vector, it's a fixed length one, generating an invalid translation for
constant scalable vector initializations.
Differential Revision: https://reviews.llvm.org/D117125
* This is useful when you need to build mixed array from external static/dynamic arrays (e.g. from adaptor during dialect conversion)
* Also, to reduce C++ code in td and generated files
Differential Revision: https://reviews.llvm.org/D117106
This overload parses a pipeline string that contains the anchor operation type, and returns an OpPassManager
corresponding to the provided pipeline. This is useful for various situations, such as dynamic pass pipelines
which are not anchored within a parent pass pipeline.
fixes#52813
Differential Revision: https://reviews.llvm.org/D116525
Dynamic batch for rescale, gather, max_pool, avg_pool, conv2D and depthwise_conv2D. Split helper functions into a separate header file.
Reviewed By: rsuderman
Differential Revision: https://reviews.llvm.org/D117031
ShapedType was created in a time before interfaces, and is one of the earliest
type base classes in the ecosystem. This commit refactors ShapedType into
an interface, which is what it would have been if interfaces had existed at that
time. The API of ShapedType and it's derived classes are essentially untouched
by this refactor, with the exception being the API surrounding kDynamicIndex
(which requires a sole home).
For now, the API of ShapedType and its name have been kept as consistent to
the current state of the world as possible (to help with potential migration churn,
among other reasons). Moving forward though, we should look into potentially
restructuring its API and possible its name as well (it should really have "Interface"
at the end like other interfaces at the very least).
One other potentially interesting note is that I've attached the ShapedType::Trait
to TensorType/BaseMemRefType to act as mixins for the ShapedType API. This
is kind of weird, but allows for sharing the same API (i.e. preventing API loss from
the transition from base class -> Interface). This inheritance doesn't affect any
of the derived classes, it is just for API mixin.
Differential Revision: https://reviews.llvm.org/D116962
This field allows for defining a code block that is placed in both the interface
and trait declarations. This is very useful when defining a set of utilities to
expose on both the Interface class and the derived attribute/operation/type.
In non-static methods, `$_attr`/`$_op`/`$_type` (depending on the type of
interface) may be used to refer to an instance of the IR entity. In the interface
declaration, this is an instance of the interface class. In the trait declaration,
this is an instance of the concrete entity class (e.g. `IntegerAttr`, `FuncOp`, etc.).
Differential Revision: https://reviews.llvm.org/D116961
Apply scale may encounter scalar, tensor, or vector operations. Expand the
lowering so that it can lower arbitrary of container types.
Reviewed By: NatashaKnk
Differential Revision: https://reviews.llvm.org/D117080
This method simply forwards to populateFunctionLikeTypeConversionPattern,
which is more general. This also helps to remove special treatment of FuncOp from
DialectConversion.
Differential Revision: https://reviews.llvm.org/D116624
This patch fixes:
mlir/lib/Dialect/Arithmetic/Transforms/ExpandOps.cpp:161:52: error:
'static_assert' with no message is a C++17 extension
[-Werror,-Wc++17-extensions]
There have been a few API pieces remaining to allow for a smooth transition for
downstream users, but these have been up for a few months now. After this only
the C API will have reference to "Identifier", but those will be reworked in a followup.
The main updates are:
* Identifier -> StringAttr
* StringAttr::get requires the context as the first parameter
- i.e. `Identifier::get("...", ctx)` -> `StringAttr::get(ctx, "...")`
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D116626
If any of the operands is NaN, return the operand instead of a new constant.
When the rhs operand is a constant, the second arith.cmpf+select ops will be folded away.
https://reviews.llvm.org/D117010 marks the two ops commutative, which will place the constant on the rhs.
Reviewed By: herhut
Differential Revision: https://reviews.llvm.org/D117011
Enable constant folding of ops within the math dialect, and introduce constant folders for ceil and log2
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D117085
This change makes it possible to use a different buffer deallocation strategy. E.g., `-buffer-deallocation` can be used, which also works for allocations that are not in destination-passing style.
Differential Revision: https://reviews.llvm.org/D117096
This op is an example for how to deal with ops who's OpResult may aliasing with one of multiple OpOperands.
Differential Revision: https://reviews.llvm.org/D116868
Add inliner interface for GPU dialect. The interface marks all GPU
dialect ops legal to inline anywhere.
Differential Revision: https://reviews.llvm.org/D116889
Enable constant folding of ops within the math dialect, and introduce constant folders for ceil and log2
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D117085
Given a while loop whose condition is given by a cmp, don't recomputed the comparison (or its inverse) in the after region, instead use a constant since the original condition must be true if we branched to the after region.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D117047
Most convolution operations need explicit padding of the input to
ensure all accesses are inbounds. In such cases, having a pad
operation can be a significant overhead. One way to reduce that
overhead is to try to fuse the pad operation with the producer of its
source.
A sequence
```
linalg.generic -> linalg.pad_tensor
```
can be replaced with
```
linalg.fill -> tensor.extract_slice -> linalg.generic ->
tensor.insert_slice.
```
if the `linalg.generic` has all parallel iterator types.
Differential Revision: https://reviews.llvm.org/D116418
This patch adds corresponding memory effects to mlir llvm-dialect load/store/addressof ops, which thus enables canonicalizations of those ops (like dead code elimination) that rely on the effect interface
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D117041
This patch fixes:
mlir/lib/Dialect/Tosa/Transforms/TosaOptionalDecompositions.cpp:28:8:
error: 'runOnFunction' overrides a member function but is not marked
'override' [-Werror,-Wsuggest-override]
Moved all TOSA decomposition patterns so that they can be optionally populated
and used by external rewrites. This avoids decomposing TOSa operations when
backends may benefit from the non-decomposed version.
Reviewed By: rsuderman, mehdi_amini
Differential Revision: https://reviews.llvm.org/D116526
Apply scale may operate on vectors, scalars, or tensors during
tiling. Relax the requirements to avoid failures.
Reviewed By: NatashaKnk
Differential Revision: https://reviews.llvm.org/D116981
Given an if of the form, simplify it by eliminating the not and swapping the regions
scf.if not(c) {
yield origTrue
} else {
yield origFalse
}
becomes
scf.if c {
yield origFalse
} else {
yield origTrue
}
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D116990
All named ops list iterators for accessing output first except
pooling ops. This commit made the pooling ops consistent with
the rest.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D115520
Uday Bondhugula