Tested with gcc-10. Other compilers may generate additional warnings. This does not fix all warnings. There are a few extra ones in LLVMCore and MLIR.
* `OpEmitter::getAttrNameIndex`: -Wunused-function (function is private and not used anywhere)
* `PrintOpPass` copy constructor: -Wextra ("Base class should be explicitly initialized in the copy constructor")
* `LegalizeForLLVMExport.cpp`: -Woverflow (overflow is expected, silence warning by making the cast explicit)
Differential Revision: https://reviews.llvm.org/D107525
The existing vector transforms reduce the dimension of transfer_read
ops. However, beyond a certain point, the vector op actually has
to be reduced to a scalar load, since we can't load a zero-dimension
vector. This handles this case.
Note that in the longer term, it may be preferaby to support
zero-dimension vectors. see
https://llvm.discourse.group/t/should-we-have-0-d-vectors/3097.
Differential Revision: https://reviews.llvm.org/D103432
We can propagate the shape from tosa.cond_if operands into the true/false
regions then through the connected blocks. Then, using the tosa.yield ops
we can determine what all possible return types are.
Reviewed By: jpienaar
Differential Revision: https://reviews.llvm.org/D105940
Handles shape inference for identity, cast, and rescale. These were missed
during the initialy elementwise work. This includes resize shape propagation
which includes both attribute and input type based propagation.
Reviewed By: jpienaar
Differential Revision: https://reviews.llvm.org/D105845
This patch adds the critical construct to the OpenMP dialect. The
implementation models the definition in 2.17.1 of the OpenMP 5 standard.
A name and hint can be specified. The name is a global entity or has
external linkage, it is modelled as a FlatSymbolRefAttr. Hint is
modelled as an integer enum attribute.
Also lowering to LLVM IR using the OpenMP IRBuilder.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D107135
Silence clang-tidy warning in AffineOps.cpp due to the inability to see
through the typeswitch. NFC.
Differential Revision: https://reviews.llvm.org/D106125
Add ForLoopBoundSpecialization pass, which specializes scf.for loops into a "main loop" where `step` divides the iteration space evenly and into an scf.if that handles the last iteration.
This transformation is useful for vectorization and loop tiling. E.g., when vectorizing loads/stores, programs will spend most of their time in the main loop, in which only unmasked loads/stores are used. Only the in the last iteration (scf.if), slower masked loads/stores are used.
Subsequent commits will apply this transformation in the SparseDialect and in Linalg's loop tiling.
Differential Revision: https://reviews.llvm.org/D105804
Introduces a conversion from one (sparse) tensor type to another
(sparse) tensor type. See the operation doc for details. Actual
codegen for all cases is still TBD.
Reviewed By: ThomasRaoux
Differential Revision: https://reviews.llvm.org/D107205
When we vectorize a scalar constant, the vector constant is inserted before its
first user if the scalar constant is defined outside the loops to be vectorized.
It is possible that the vector constant does not dominate all its users. To fix
the problem, we find the innermost vectorized loop that encloses that first user
and insert the vector constant at the top of the loop body.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D106609
Make broadcastable needs the output shape to determine whether the operation
includes additional broadcasting. Include some canonicalizations for TOSA
to remove unneeded reshape.
Reviewed By: NatashaKnk
Differential Revision: https://reviews.llvm.org/D106846
`PadTensorOp` has verification logic to make sure
result dim must be static if all the padding values are static.
Cast folding might add more static information for the src operand
of `PadTensorOp` which might change a valid operation to be invalid.
Change the canonicalizing pattern to fix this.
Currently TFRT does not support top-level coroutines, so this functionality will allow to have a single blocking await at the top level until TFRT implements the necessary functionality.
Reviewed By: ezhulenev
Differential Revision: https://reviews.llvm.org/D106730
Change the formatting of the debug print outs to elide unnecessary information.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D106661
Historically the builtin dialect has had an empty namespace. This has unfortunately created a very awkward situation, where many utilities either have to special case the empty namespace, or just don't work at all right now. This revision adds a namespace to the builtin dialect, and starts to cleanup some of the utilities to no longer handle empty namespaces. For now, the assembly form of builtin operations does not require the `builtin.` prefix. (This should likely be re-evaluated though)
Differential Revision: https://reviews.llvm.org/D105149
Interop parallelism requires needs awaiting on results. Blocking awaits are bad for performance. TFRT supports lightweight resumption on threads, and coroutines are an abstraction than can be used to lower the kernels onto TFRT threads.
Reviewed By: ezhulenev
Differential Revision: https://reviews.llvm.org/D106508
The verifier of the llvm.call operation was not checking for mismatches between
the number of operation results and the number of results in the signature of
the callee. Furthermore, it was possible to construct an llvm.call operation
producing an SSA value of !llvm.void type, which should not exist. Add the
verification and treat !llvm.void result type as absence of call results.
Update the GPU conversions to LLVM that were mistakenly assuming that it was
fine for llvm.call to produce values of !llvm.void type and ensure these calls
do not produce results.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D106937
- Fixed symbol insertion into `symNameToModuleMap`. Insertion
needs to happen whether symbols are renamed or not.
- Added check for the VCE triple and avoid dropping it.
- Disabled function deduplication. It requires more careful
rules. Right now it can remove different functions.
- Added tests for symbol rename listener.
- And some other code/comment cleanups.
Reviewed By: ergawy
Differential Revision: https://reviews.llvm.org/D106886
Specialize the DeduplicateInputs and RemoveIdentityLinalgOps patterns for GenericOp instead of implementing them for the LinalgOp interface.
This revsion is based on https://reviews.llvm.org/D105622 that moves the logic to erase identity CopyOps in a separate pattern.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D105291
Split out an EraseIdentityCopyOp from the existing RemoveIdentityLinalgOps pattern. Introduce an additional check to ensure the pattern checks the permutation maps match. This is a preparation step to specialize RemoveIdentityLinalgOps to GenericOp only.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D105622
`memref.collapse_shape` has verification logic to make sure
result dim must be static if all the collapsing src dims are static.
Cast folding might add more static information for the src operand
of `memref.collapse_shape` which might change a valid collapsing
operation to be invalid. Add `CollapseShapeOpMemRefCastFolder` pattern
to fix this.
Minor changes to `convertReassociationIndicesToExprs` to use `context`
instead of `builder` to avoid extra steps to construct temporary
builders.
Reviewed By: nicolasvasilache, mravishankar
Differential Revision: https://reviews.llvm.org/D106670
The order of testing in two sparse tensor ops was incorrect,
which could cause an invalid cast (crashing the compiler instead
of reporting the error). This revision fixes that bug.
Reviewed By: gussmith23
Differential Revision: https://reviews.llvm.org/D106841
Retaining old interface and should be constructable as previous, change would have been NFC except it this doesn't implicitly work with OpAdaptor generated in C++14.
Differential Revision: https://reviews.llvm.org/D106772
This CL adds a new RegionBranchTerminatorOpInterface to query information about operands that can be
passed to successor regions. Similar to the BranchOpInterface, it allows to freely define the
involved operands. However, in contrast to the BranchOpInterface, it expects an additional region
number to distinguish between various use cases which might require different operands passed to
different regions.
Moreover, we added new utility functions (namely getMutableRegionBranchSuccessorOperands and
getRegionBranchSuccessorOperands) to query (mutable) operand ranges for operations equiped with the
ReturnLike trait and/or implementing the newly added interface. This simplifies reasoning about
terminators in the scope of the nested regions.
We also adjusted the SCF.ConditionOp to benefit from the newly added capabilities.
Differential Revision: https://reviews.llvm.org/D105018
When the output indexing map has a permutation we need to consider in
the contraction vector type.
Differential Revision: https://reviews.llvm.org/D106469
When the source tensor of a tensor.insert_slice is not equivalent to an inplace buffer an extra copy is necessary. This revision adds the missing copy.
Reviewed By: gysit
Differential Revision: https://reviews.llvm.org/D106587
If this pass executes without shape inference its possible for unranked tensors
to appear in the IR. This pass should gracefully handle unranked tensors.
Differential Revision: https://reviews.llvm.org/D106617
- Change findDealloc() to return Optional<Operation *> and return None if > 1
dealloc is associated with the given alloc.
- Add findDeallocs() to return all deallocs associated with the given alloc.
- Fix current uses of findDealloc() to bail out if > 1 dealloc is found.
Differential Revision: https://reviews.llvm.org/D106456
Fix affine.for empty loop body folder in the presence of yield values.
The existing pattern ignored iter_args/yield values and thus crashed
when yield values had uses.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D106121
AffineForOp's folding hook is expected to fold away trivially empty
affine.for. This allows simplification to happen as part of the
canonicalizer and from wherever the folding hook is used. While more
complex analysis based zero trip count detection is available from other
passes in analysis and transforms, simple and inexpensive folding had
been missing.
Also, update/improve affine.for op documentation clarifying semantics of
the result values for zero trip count loops.
Differential Revision: https://reviews.llvm.org/D106123
Introduce a new rewrite driver (MultiOpPatternRewriteDriver) to rewrite
a supplied list of ops and other ops. Provide a knob to restrict
rewrites strictly to those ops or also to affected ops (but still not to
completely related ops).
This rewrite driver is commonly needed to run any simplification and
cleanup at the end of a transforms pass or transforms utility in a way
that only simplifies relevant IR. This makes it easy to write test cases
while not performing unrelated whole IR simplification that may
invalidate other state at the caller.
The introduced utility provides more freedom to developers of transforms
and transform utilities to perform focussed and local simplification. In
several cases, it provides greater efficiency as well as more
simplification when compared to repeatedly calling
`applyOpPatternsAndFold`; in other cases, it avoids the need to
undesirably call `applyPatternsAndFoldGreedily` to do unrelated
simplification in a FuncOp.
Update a few transformations that were earlier using
applyOpPatternsAndFold (SimplifyAffineStructures,
affineDataCopyGenerate, a linalg transform).
TODO:
- OpPatternRewriteDriver can be removed as it's a special case of
MultiOpPatternRewriteDriver, i.e., both can be merged.
Differential Revision: https://reviews.llvm.org/D106232
This patch allows iterating typed enum via the ADT/Sequence utility.
It also changes the original design to better separate concerns:
- `StrongInt` only deals with safe `intmax_t` operations,
- `SafeIntIterator` presents the iterator and reverse iterator
interface but only deals with safe `StrongInt` internally.
- `iota_range` only deals with `SafeIntIterator` internally.
This design ensures that operations are always valid. In particular,
"Out of bounds" assertions fire when:
- the `value_type` is not representable as an `intmax_t`
- iterator operations make internal computation underflow/overflow
- the internal representation cannot be converted back to `value_type`
Differential Revision: https://reviews.llvm.org/D106279
This allows caller to use non-const functions, e.g., `getOperandNumber`, etc. It
is expected that OpOperand is not modified in a callback function.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D106322
Bufferization handles all unknown ops conservative. The patch ensures accessing the dimension of an output tensor does not prevent in place bufferization.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D106356
The `reifyReturnTypeShapesPerResultDim` method supports shape
inference for rsults that are ranked types. These are used lower in
the codegeneration stack than its counter part `reifyReturnTypeShapes`
which also supports unranked types, and is more suited for use higher
up the compilation stack. To have separation of concerns, this method
is split into its own interface.
See discussion : https://llvm.discourse.group/t/better-layering-for-infershapedtypeopinterface/3823
Differential Revision: https://reviews.llvm.org/D106133
This is the first step to support software pipeline for scf.for loops.
This is only the transformation to create pipelined kernel and
prologue/epilogue.
The scheduling needs to be given by user as many different algorithm
and heuristic could be applied.
This currently doesn't handle loop arguments, this will be added in a
follow up patch.
Differential Revision: https://reviews.llvm.org/D105868
This makes it more explicit what the scope of this pass is. The name
of this pass predates fusion on tensors using tile + fuse, and hence
the confusion.
Differential Revision: https://reviews.llvm.org/D106132
Added shape inference handles cases for convolution operations. This includes
conv2d, conv3d, depthwise_conv2d, and transpose_conv2d. With transpose conv
we use the specified output shape when possible however will shape propagate
if the output shape attribute has dynamic values.
Reviewed By: jpienaar
Differential Revision: https://reviews.llvm.org/D105645
Add pattern to fold a TensorCast into a PadTensorOp if the cast removes static size information.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D106278
Insert ops replacing pad_tensor in front of the associated tansfer_write / insert_slice op. Otherwise we may end up with invalid ir if one of the remaining tansfer_write / insert_slice operands is defined after the pad_tensor op.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D106162
This simplifies the vector to LLVM lowering. Previously, both vector.load/store and vector.transfer_read/write lowered directly to LLVM. With this commit, there is a single path to LLVM vector load/store instructions and vector.transfer_read/write ops must first be lowered to vector.load/store ops.
* Remove vector.transfer_read/write to LLVM lowering.
* Allow non-unit memref strides on all but the most minor dimension for vector.load/store ops.
* Add maxTransferRank option to populateVectorTransferLoweringPatterns.
* vector.transfer_reads with changing element type can no longer be lowered to LLVM. (This functionality is needed only for SPIRV.)
Differential Revision: https://reviews.llvm.org/D106118
Removed inconsistent name prefixes, added consistency checks
on debug strings, added more assertions to verify assumptions
that may be lifted in the future.
Reviewed By: gussmith23
Differential Revision: https://reviews.llvm.org/D106108
The dialect-specific cast between builtin (ex-standard) types and LLVM
dialect types was introduced long time before built-in support for
unrealized_conversion_cast. It has a similar purpose, but is restricted
to compatible builtin and LLVM dialect types, which may hamper
progressive lowering and composition with types from other dialects.
Replace llvm.mlir.cast with unrealized_conversion_cast, and drop the
operation that became unnecessary.
Also make unrealized_conversion_cast legal by default in
LLVMConversionTarget as the majority of convesions using it are partial
conversions that actually want the casts to persist in the IR. The
standard-to-llvm conversion, which is still expected to run last, cleans
up the remaining casts standard-to-llvm conversion, which is still
expected to run last, cleans up the remaining casts
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D105880
This may be necessary in partial multi-stage conversion when a container type
from dialect A containing types from dialect B goes through the conversion
where only dialect A is converted to the LLVM dialect. We will need to keep a
pointer-to-non-LLVM type in the IR until a further conversion can convert
dialect B types to LLVM types.
Reviewed By: wsmoses
Differential Revision: https://reviews.llvm.org/D106076
Changes include the following:
1. Single iteration reduction loops being sibling fused at innermost insertion level
are skipped from being considered as sequential loops.
Otherwise, the slice bounds of these loops is reset.
2. Promote loops that are skipped in previous step into outer loops.
3. Two utility function - buildSliceTripCountMap, getSliceIterationCount - are moved from
mlir/lib/Transforms/Utils/LoopFusionUtils.cpp to mlir/lib/Analysis/Utils.cpp
Reviewed By: bondhugula, vinayaka-polymage
Differential Revision: https://reviews.llvm.org/D104249
This format was missing from the support library. Although there are some
subtleties reading in an external format for int64 as double, there is no
good reason to omit support for this data type form the support library.
Reviewed By: gussmith23
Differential Revision: https://reviews.llvm.org/D106016
Arbitrary shifts have some complications, but shift by invariants
(viz. tensor index exp only at left hand side) can be easily
handled with the conjunctive rule.
Reviewed By: gussmith23
Differential Revision: https://reviews.llvm.org/D106002
Instead of relying on adhoc bounds calculations, use a projection-based
implementation. This simplifies the implementation and finds more static
constant sizes than previously/
Differential Revision: https://reviews.llvm.org/D106054
Tiling can be enabled with `linalg-tile-pad-tensor-ops`. Only scf::ForOp can be generated at the moment.
Differential Revision: https://reviews.llvm.org/D105460
Factor out the functionality into a new function, so that it can be used for creating PadTensorOp tiles.
Differential Revision: https://reviews.llvm.org/D105458
Adds zero-preserving unary operators from std. Also adds xor.
Performs minor refactoring to remove "zero" node, and pushed
the irregular logic for negi (not support in std) into one place.
Reviewed By: gussmith23
Differential Revision: https://reviews.llvm.org/D105928
Clean up corner cases related to elemental tensor / buffer type return values that would previously fail.
Reviewed By: ThomasRaoux
Differential Revision: https://reviews.llvm.org/D105857
Previously, linalg bufferization always had to be conservative at function boundaries and assume the most dynamic strided memref layout.
This revision introduce the mechanism to specify a linalg.buffer_layout function argument attribute that carries an affine map used to set a less pessimistic layout.
Reviewed By: ThomasRaoux
Differential Revision: https://reviews.llvm.org/D105859
Integral AND and OR follow the simple conjunction and disjuction rules
for lattice building. This revision also completes some of the Merge
refactoring by moving the remainder parts that are merger specific from
sparsification into utils files.
Reviewed By: gussmith23
Differential Revision: https://reviews.llvm.org/D105851
Pool operations perform the same shape propagation. Included the shape
propagation and tests for these avg_pool2d and max_pool2d.
Differential Revision: https://reviews.llvm.org/D105665
Right now, we only accept x/c with nonzero c, since this
conceptually can be treated as a x*(1/c) conjunction for both
FP and INT as far as lattice computations go. The codegen
keeps the division though to preserve precise semantics.
See discussion:
https://llvm.discourse.group/t/sparse-tensors-in-mlir/3389/28
Reviewed By: gussmith23
Differential Revision: https://reviews.llvm.org/D105731
Previously, comprehensive bufferization of scf.yield did not have enough information
to detect whether an enclosing scf::for bbargs would bufferize to a buffer equivalent
to that of the matching scf::yield operand.
As a consequence a separate sanity check step would be required to determine whether
bufferization occured properly.
This late check would miss the case of calling a function in an loop.
Instead, we now pass and update aliasInfo during bufferization and it is possible to
imrpove bufferization of scf::yield and drop that post-pass check.
Add an example use case that was failing previously.
This slightly modifies the error conditions, which are also updated as part of this
revision.
Differential Revision: https://reviews.llvm.org/D105803
The trait was inconsistent with the other broadcasting logic here. And
also fix printing here to use ? rather than -1 in the error.
Differential Revision: https://reviews.llvm.org/D105748
Use a modeling similar to SCF ParallelOp to support arbitrary parallel
reductions. The two main differences are: (1) reductions are named and declared
beforehand similarly to functions using a special op that provides the neutral
element, the reduction code and optionally the atomic reduction code; (2)
reductions go through memory instead because this is closer to the OpenMP
semantics.
See https://llvm.discourse.group/t/rfc-openmp-reduction-support/3367.
Reviewed By: kiranchandramohan
Differential Revision: https://reviews.llvm.org/D105358
Jacques Pienaar