This pattern is similar to `FoldFillWithTensorReshape`, which performs the same swapping with reshapes.
Fill the smaller extracted tensor slice instead of `x`. This allows for additional simplifications in case `x` is the result of another extract_slice.
Differential Revision: https://reviews.llvm.org/D141117
This revision adapts the pattern in LinAlg to work on DPS interface, and
adds it to canonicalization patterns of tensor dialect. The
InsertSliceOp is skipped in the pattern because it has its own logic
about folding tensor.cast ops.
Reviewed By: pifon2a
Differential Revision: https://reviews.llvm.org/D139375
Original [RFC](discourse.llvm.org/t/rfc-primitive-ops-add-broadcastop-to-linalg/66313) defined `dimensions` as a map from input to init, but a discussion in reviews.llvm.org/D138291 concluded that it's more natural for `dimensions` to represent added dims. Also this way is more consistent with `linalg.reduce`.
Differential Revision: https://reviews.llvm.org/D138408
Added dimensions can be both static and dinamic. Mapped dimension should be the same in the input and the init.
Differential Revision: https://reviews.llvm.org/D138291
The patterns to remove dead arguments and results of `linalg.generic`
operations are not necessarily canonicalizations. Instead a new entry
point `populateEraseUnusedOperandsAndResults` is added to allow using
these patterns when needed. The transformations that rely on this
pattern for cleanup now include these patterns explicitly.
Differential Revision: https://reviews.llvm.org/D138085
The pattern tries to deduce static shape from `tensor.cast` producer of linalg operation operands.
The original code unconditionally casts type of the `tensor.cast` source to `RankedTensorType`.
But the `tensor.cast` can also operate on `UnrankedTensorType`, so this cast either fail on assertion
in debug build or introduce UB in release build.
The patch replaces unconditional cast with `dyn_cast` and check for the cast result.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D137775
This is unnecessary if the generated operation type already matches
the type of the replaced value. Also use `OpFoldResult` to reduce the
number of cases the casts are needed.
Reviewed By: springerm, hanchung, antiagainst
Differential Revision: https://reviews.llvm.org/D137479
During elementwise fusion the fillOp's value was directly
referred without casting which can create mismatching dtypes.
Reviewed By: mravishankar, ThomasRaoux
Differential Revision: https://reviews.llvm.org/D137447
[RFC: EnumAttr for iterator types in Linalg](https://discourse.llvm.org/t/rfc-enumattr-for-iterator-types-in-linalg/64535)
This affect touches and probably breaks most of the code that creates `linalg.generic`. A fix would be to replace calls to `getParallelIteratorTypeName/getReductionIteratorTypeName` with `mlir::utils::IteratorType::parallel/reduction` and types from `StringRef` to `mlir::utils::IteratorType`.
Due to limitations of tablegen, shared C++ definition of IteratorType enum lives in StructuredOpsUtils.td, but each dialect should have it's own EnumAttr wrapper. To avoid conflict, all enums in a dialect are put into a separate file with a separate tablegen rule.
Test dialect td files are refactored a bit.
Printed format of `linalg.generic` temporarily remains unchanged to avoid breaking code and tests in the same change.
Differential Revision: https://reviews.llvm.org/D137658
This patch fixes:
llvm-project/mlir/lib/Dialect/Linalg/IR/LinalgOps.cpp:1855:17:
error: comparison of integers of different signs: 'int64_t' (aka
'long') and 'size_t' (aka 'unsigned long') [-Werror,-Wsign-compare]
This crash was due to incorrect usage of `hasTensorSemantics`, which has changed recently with DestinationStyleOpInterface. An op has tensor semantics if all of its inits and inputs are tensors. Previously, only inits needed to be tensors.
Differential Revision: https://reviews.llvm.org/D137243
The region should yield the first argument (input) not the last argument
(output). Also fix a few tests that were affected by this bug.
Differential Revision: https://reviews.llvm.org/D136924
The new builders get a list of additional attrs, a lambda to build the region
body and infer return types from `init`.
Differential Revision: https://reviews.llvm.org/D136838
The foldMemRefCast method is defined in memref namespace; the
foldTensorCast method is defined in tensor namespace. This revision
deletes the dup code and use the unified methods.
Reviewed By: dcaballe
Differential Revision: https://reviews.llvm.org/D136379
`getIteratorTypesArray` should be used instead. It's a better substitute for all the current usages of the interface.
The current `ArrayAttr iterator_types()` has a few problems:
* It creates an assumption operation has iterators types as an attribute, but it's not always the case. Sometime iterator types can be inferred from other attribute, or they're just static.
* ArrayAttr is an obscure contained and required extracting values in the client code.
* Makes it hard to migrate iterator types from strings to enums ([RFC](https://discourse.llvm.org/t/rfc-enumattr-for-iterator-types-in-linalg/64535/9)).
Concrete ops, like `linalg.generic` will still have iterator types as an attribute if needed.
As a side effect, this change helps a bit with migration to prefixed accessors.
Differential Revision: https://reviews.llvm.org/D135765
This will allow to model elementwise ops with this special op instead of using
GenericOp.
Also allow MapOp and ReduceOp to have no result if the output type is not a tensor.
This is needed for buffer semantics.
Differential Revision: https://reviews.llvm.org/D135754
One of the assertion is causing signed/unsigned mismatch. However this
assertion seems redundant and is no longer used.
Reviewed By: mravishankar, ThomasRaoux
Differential Revision: https://reviews.llvm.org/D135612
This patch fixes:
mlir/lib/Dialect/Linalg/IR/LinalgOps.cpp:1051:40: error: comparison
of integers of different signs: 'int64_t' (aka 'long') and
'std::size_t' (aka 'unsigned long') [-Werror,-Wsign-compare]
Some higher level operations such as torch.max generates linalg generic
that returns both the index and the value of the max operation. However
sometimes not all information is being used. This however blocks
vectorization for certain cases which causes performance degradation.
This patch aims to fix this issue.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D135388
tensor.empty/linalg.init_tensor produces an uninititalized tensor that can be used as a destination operand for destination-style ops (ops that implement `DestinationStyleOpInterface`).
This change makes it possible to implement `TilingInterface` for non-destination-style ops without depending on the Linalg dialect.
RFC: https://discourse.llvm.org/t/rfc-add-tensor-from-shape-operation/65101
Differential Revision: https://reviews.llvm.org/D135129
This patch fixes:
mlir/lib/Dialect/Linalg/IR/LinalgOps.cpp:1348:31: error: comparison
of integers of different signs: 'size_t' (aka 'unsigned long') and
'int64_t' (aka 'long') [-Werror,-Wsign-compare]
mlir/lib/ExecutionEngine/SparseTensor/File.cpp:110:3: error: default
label in switch which covers all enumeration values
[-Werror,-Wcovered-switch-default]
This op will allow to model (variadic) reductions with this special op
instead of using GenericOp.
This reverts commit 535fd753ef.
Additional fix: implement a getLibraryName method.
