- Provide the operator overloads for constructing (semi-)affine expressions in
Python by combining existing expressions with constants.
- Make AffineExpr, AffineMap and IntegerSet hashable in Python.
- Expose the AffineExpr composition functionality.
Reviewed By: gysit, aoyal
Differential Revision: https://reviews.llvm.org/D113010
This better decouples transfer read/write from vector-only rewrite of conv.
This form is close to ready to plop into a new vector.conv op and the vector.transfer operations to be generalized as part of generic vectorization once the properties ConvolutionOpInterface are inferred from the indexing maps.
This also results in a nice perf boost in the dw == 1 cases.
Differential revision: https://reviews.llvm.org/D112822
This refactoring prepares conv1d vectorization for a future integration into
the generic codegen path.
Once transfer_read / transfer_write vectorization also supports sliding windows,
the special pattern for conv can disappear.
This will also likely need a vector.conv operation.
Differential Revision: https://reviews.llvm.org/D112797
The current setup of LinalgTransformationFilter allows a
transformation to trigger when either
1) The StringAttr is not set and no filter identifier is specified.
2) The StringAttr is set and its value matches (one of) the provided
identifier.
This misses the case where the transformation should trigger either
when the attribute is not set or its value matches (one of) the
provided identifier. Since `Identifier` does not allow empty strings,
add a boolean option to match when the attribute is not set. This
option is by default off.
Differential Revision: https://reviews.llvm.org/D113057
The 2-D case can be rewritten to generate quite fewer instructions and a single vector.shuffle which seems to provide a nice performance boost.
Add this arrow to our quiver by exposing it with a new vector transform option.
Differential Revision: https://reviews.llvm.org/D113062
We'd like to take a progressive approach towards Fconvolution op
CodeGen, by 1) tiling it to fit compute hierarchy first, and then
2) tiling along window dimensions with size 1 to reduce the problem
to be matmul-like. After that, we can 3) downscale high-D convolution
ops to low-D by removing the size-1 window dimensions. The final
step would be 4) vectorizing the low-D convolution op directly.
We have patterns for 1), 2), and 4). This commit adds a pattern for
3) for `linalg.conv_2d_nhwc_hwcf` ops as a starter. Supporting other
high-D convolution ops should be similar and mechanical.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D112928
Symbol tables are a largely useful top-level IR construct, for example, they
make it easy to access functions in a module by name instead of traversing the
list of module's operations to find the corresponding function.
Depends On D112886
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D112821
Inserting a symbol into a SymbolTable may lead to the name of the symbol being
changed in order to ensure uniqueness of symbol names in the table. Return this
new name to spare the caller the need to extract it from the symbol operation.
Depends On D112700
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D112886
This commit moves parts of the existing bufferization code into external op interface implementations. Furthermore, Comprehensive Bufferize is adapted to use the new interface.
Future commits will decouple the interface and its op implementations from Comprehensive Bufferize and the Linalg dialect, as well as split them into multiple files with their own build targets. This commit leaves the file structure and build rules mostly unchanged.
Differential Revision: https://reviews.llvm.org/D112900
This commit adds a new op interface: BufferizableOpInterface. In the future, ops that implement this interface can be bufferized using Comprehensive Bufferize.
Note: The interface methods of this interface correspond to the "op interface" in ComprehensiveBufferize.cpp.
Differential Revision: https://reviews.llvm.org/D112974
In order to support fusion with mma matrix type we need to be able to
execute elementwise operations on them. This add an op to be able to
support some basic elementwise operations. This is a is not a full
solution as it only supports a limited scope or operations. Ideally we would
want to be able to fuse with more kind of operations.
Differential Revision: https://reviews.llvm.org/D112857
wmma intrinsics have a large number of combinations, ideally we want to be able
to target all the different variants. To avoid a combinatorial explosion in the
number of mlir op we use attributes to represent the different variation of
load/store/mma ops. We also can generate with tablegen helpers to know which
combinations are available. Using this we can avoid having too hardcode a path
for specific shapes and can support more types.
This patch also adds boiler plates for tf32 op support.
Differential Revision: https://reviews.llvm.org/D112689
Add the shufflevector conversion. It only handles the static, i.e., IntegerAttr, index.
Co-authored: Xinyi Liu <xyliuhelen@gmail.com>
Reviewed by: antiagainst
Differential revision: https://reviews.llvm.org/D112161
This makes the class usable with types that do not provide their own operator<.
Update MLIR Linalg ComprehensiveBufferize to take advantage of the new template param.
Differential Revision: https://reviews.llvm.org/D112052
Provide support for removing an operation from the block that contains it and
moving it back to detached state. This allows for the operation to be moved to
a different block, a common IR manipulation for, e.g., module merging.
Also fix a potential one-past-end iterator dereference in Operation::moveAfter
discovered in the process.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D112700
The list of operations that do neither read nor write, but create an alias when bufferizing inplace, is getting longer. This commit adds a helper function so that we do not have to spell out the entire list each time.
Differential Revision: https://reviews.llvm.org/D112515
This patch reorders mergeLocalIds usage to merge locals only after number of
dimensions and symbols are same. This does not change any functionality
because it does not matter in what order identifiers are merged, since
the reason to do it is to ensure that two FACs are aligned.
The order ensured in this patch simplifies a subsequent patch to improve
mergeLocalIds which requires dimensions and symbols to be aligned.
Reviewed By: bondhugula
Differential Revision: https://reviews.llvm.org/D112841
Added a type with different pointer/index bit width. Also
added some sanity CHECKs on the stored indices.
Reviewed By: wrengr
Differential Revision: https://reviews.llvm.org/D112778
When operand is a subview we don't infer in_bounds and some default cases (e.g case in the tests) will crash with `operand is NULL` when converting to LLVM
Reviewed By: ThomasRaoux
Differential Revision: https://reviews.llvm.org/D112772
Add a strategy pass that pads and hoists after tiling and fusion.
Depends On D112412
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D112480
Adding a padding and hoisting pattern, a test pass, and tests. The patch prepares the split of tiling/fusion and padding.
Depends On D112255
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D112412
Using [1] for representing shape of a scalar is incorrect, and will break with vectors of size 1.
- remove redundant helper functions
- fix couple of style warnings
Reviewed By: cota
Differential Revision: https://reviews.llvm.org/D112764
InsertionGuards move constructor is currently the compiler synthesized implementation which is very bug prone. A move constructed InsertionGuard will get the same builder and insertion point as the one it is constructed from, leading to insertion point being restored twice. This can even happen in non obvious situations on some compilers, such as when returning a move constructible struct from a function.
This patch fixes the issue by properly implementing the move constructor. An InsertionGuard that was used to move construct another InsertionGuard is simply inactive and will not restore the insertion point.
I chose to explicitly delete the move assign operator as its semantics are not clear cut. If one were to strictly follow the rule of 5, you'd have to restore the insertion point before then taking ownership of the others guards fields. I believe that to be rather confusing and/or surprising however. One may still get such semantics using llvm::Optional or std::optional and the emplace method if really needed.
Differential Revision: https://reviews.llvm.org/D112749
Adapt hoistPaddingOnTensors to leave replacing and erasing the old pad tensor operation to the caller. This change makes the function pattern friendly.
Depends On D112003
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D112255
Adapt the rewriteAsPaddedOp method to use the OpBuilder instead of the PatterRewriter.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D112003
Cullen Rhodes