The UnrollVectorPattern is can be used in a programmable fashion by:
```
OwningRewritePatternList patterns;
patterns.insert<UnrollVectorPattern<AddFOp>>(ArrayRef<int64_t>{2, 2}, ctx);
patterns.insert<UnrollVectorPattern<vector::ContractionOp>>(
ArrayRef<int64_t>{2, 2, 2}, ctx);
...
applyPatternsAndFoldGreedily(getFunction(), patterns);
```
Differential revision: https://reviews.llvm.org/D83064
Default vector.contract lowering essentially yields a series of sdot/ddot
operations. However, for some layouts a series of saxpy/daxpy operations,
chained through fma are more efficient. This CL introduces a choice between
the two lowering paths. A default heuristic is to follow.
Some preliminary avx2 performance numbers for matrix-times-vector.
Here, dot performs best for 64x64 A x b and saxpy for 64x64 A^T x b.
```
------------------------------------------------------------
A x b A^T x b
------------------------------------------------------------
GFLOPS sdot (reassoc) saxpy sdot (reassoc) saxpy
------------------------------------------------------------
1x1 0.6 0.9 0.6 0.9
2x2 2.5 3.2 2.4 3.5
4x4 6.4 8.4 4.9 11.8
8x8 11.7 6.1 5.0 29.6
16x16 20.7 10.8 7.3 43.3
32x32 29.3 7.9 6.4 51.8
64x64 38.9 79.3
128x128 32.4 40.7
------------------------------------------------------------
```
Reviewed By: nicolasvasilache, ftynse
Differential Revision: https://reviews.llvm.org/D83012
Summary:
This changes the casing of MLIRGPUtoGPURuntimeTransforms to be consistent
with other transform libraries.
Differential Revision: https://reviews.llvm.org/D82841
This revision removes the TypeConverter parameter passed to the apply* methods, and instead moves the responsibility of region type conversion to patterns. The types of a region can be converted using the 'convertRegionTypes' method, which acts similarly to the existing 'applySignatureConversion'. This method ensures that all blocks within, and including those moved into, a region will have the block argument types converted using the provided converter.
This has the benefit of making more of the legalization logic controlled by patterns, instead of being handled explicitly by the driver. It also opens up the possibility to support multiple type conversions at some point in the future.
This revision also adds a new utility class `FailureOr<T>` that provides a LogicalResult friendly facility for returning a failure or a valid result value.
Differential Revision: https://reviews.llvm.org/D81681
Summary:
Fixed build of D81618
Add a pattern for expanding tanh op into exp form.
A `tanh` is expanded into:
1) 1-exp^{-2x} / 1+exp^{-2x}, if x => 0
2) exp^{2x}-1 / exp^{2x}+1 , if x < 0.
Differential Revision: https://reviews.llvm.org/D82040
This allows for passing a lambda to addDynamicallyLegalDialect without needing to explicit wrap with Optional<DynamicLegalityCallbackFn>.
Differential Revision: https://reviews.llvm.org/D81680
It is quite common for the same type to be converted many types throughout the conversion process, and there isn't any good reason why we aren't caching that result. Especially given that we currently use identity conversion to signify legality. This revision also adds a few additional helpers to TypeConverter.
Differential Revision: https://reviews.llvm.org/D81679
This reverts commit 32c757e4f8.
Broke the build bot:
******************** TEST 'MLIR :: Examples/standalone/test.toy' FAILED ********************
[...]
/tmp/ci-KIMiRFcVZt/lib/libMLIRLinalgToLLVM.a(LinalgToLLVM.cpp.o): In function `(anonymous namespace)::ConvertLinalgToLLVMPass::runOnOperation()':
LinalgToLLVM.cpp:(.text._ZN12_GLOBAL__N_123ConvertLinalgToLLVMPass14runOnOperationEv+0x100): undefined reference to `mlir::populateExpandTanhPattern(mlir::OwningRewritePatternList&, mlir::MLIRContext*)'
Summary:
Add a pattern for expanding tanh op into exp form.
A `tanh` is expanded into:
1) 1-exp^{-2x} / 1+exp^{-2x}, if x => 0
2) exp^{2x}-1 / exp^{2x}+1 , if x < 0.
Differential Revision: https://reviews.llvm.org/D81618
This parameter gives the developers the freedom to choose their desired function
signature conversion for preparing their functions for buffer placement. It is
introduced for BufferAssignmentFuncOpConverter, and also for
BufferAssignmentReturnOpConverter, and BufferAssignmentCallOpConverter to adapt
the return and call operations with the selected function signature conversion.
If the parameter is set, buffer placement won't also deallocate the returned
buffers.
Differential Revision: https://reviews.llvm.org/D81137
This revision adds a helper function to hoist vector.transfer_read /
vector.transfer_write pairs out of immediately enclosing scf::ForOp
iteratively, if the following conditions are true:
1. The 2 ops access the same memref with the same indices.
2. All operands are invariant under the enclosing scf::ForOp.
3. No uses of the memref either dominate the transfer_read or are
dominated by the transfer_write (i.e. no aliasing between the write and
the read across the loop)
To improve hoisting opportunities, call the `moveLoopInvariantCode` helper
function on the candidate loop above which to hoist. Hoisting the transfers
results in scf::ForOp yielding the value that originally transited through
memory.
This revision additionally exposes `moveLoopInvariantCode` as a helper in
LoopUtils.h and updates SliceAnalysis to support return scf::For values and
allow hoisting across multiple scf::ForOps.
Differential Revision: https://reviews.llvm.org/D81199
This revision adds a helper function to hoist alloc/dealloc pairs and
alloca op out of immediately enclosing scf::ForOp if both conditions are true:
1. all operands are defined outside the loop.
2. all uses are ViewLikeOp or DeallocOp.
This is now considered Linalg-specific and will be generalized on a per-need basis.
Differential Revision: https://reviews.llvm.org/D81152
Summary:
Progressive lowering of vector.transpose into an operation that
is closer to an intrinsic, and thus the hardware ISA. Currently
under the common vector transform testing flag, as we prepare
deploying this transformation in the LLVM lowering pipeline.
Reviewers: nicolasvasilache, reidtatge, andydavis1, ftynse
Reviewed By: nicolasvasilache, ftynse
Subscribers: mehdi_amini, rriddle, jpienaar, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, liufengdb, stephenneuendorffer, Joonsoo, grosul1, frgossen, Kayjukh, jurahul, llvm-commits
Tags: #llvm, #mlir
Differential Revision: https://reviews.llvm.org/D80772
Add BufferAssignmentCallOpConverter as a pattern rewriter for Buffer
Placement. It matches the signature of the caller operation with the callee
after rewriting the callee with FunctionAndBlockSignatureConverter.
Differential Revision: https://reviews.llvm.org/D80785
This revision adds custom rewrites for patterns that arise during linalg structured
ops vectorization. These patterns allow the composition of linalg promotion,
vectorization and removal of redundant copies.
The patterns are voluntarily limited and restrictive atm.
More robust behavior will be implemented once more powerful side effect modeling and analyses are available on view/subview.
On the transfer_read side, the following pattern is rewritten:
```
%alloc = ...
[optional] %view = std.view %alloc ...
%subView = subview %allocOrView ...
[optional] linalg.fill(%allocOrView, %cst) ...
...
linalg.copy(%in, %subView) ...
vector.transfer_read %allocOrView[...], %cst ...
```
into
```
[unchanged] %alloc = ...
[unchanged] [optional] %view = std.view %alloc ...
[unchanged] [unchanged] %subView = subview %allocOrView ...
...
vector.transfer_read %in[...], %cst ...
```
On the transfer_write side, the following pattern is rewriten:
```
%alloc = ...
[optional] %view = std.view %alloc ...
%subView = subview %allocOrView...
...
vector.transfer_write %..., %allocOrView[...]
linalg.copy(%subView, %out)
```
Differential Revision: https://reviews.llvm.org/D80728
This utility factors out the machinery required to add iterArgs and yield values to an scf.ForOp.
Differential Revision: https://reviews.llvm.org/D80656
Buffer placement can now operates on functions that return buffers. These
buffers escape from the deallocation phase of buffer placement.
Differential Revision: https://reviews.llvm.org/D80696
https://reviews.llvm.org/D79246 introduces alignment propagation for vector transfer operations. Unfortunately, the alignment calculation is incorrect and can result in crashes.
This revision fixes the calculation by using the natural alignment of the memref elemental type, instead of the resulting vector type.
If more alignment is desired, it can be done in 2 ways:
1. use a proper vector.type_cast to transform a memref<axbxcxdxf32> into a memref<axbxvector<cxdxf32>> giving a natural alignment of vector<cxdxf32>
2. add an alignment attribute to vector transfer operations and propagate it.
With this change the alignment in the relevant tests goes down from 128 to 4.
Lastly, a few minor cleanups are performed and the custom `isMinorIdentityMap` is deprecated.
Differential Revision: https://reviews.llvm.org/D80734
D80142 restructured MLIR-to-GPU-binary conversion to support multiple
targets. It also modified cmake files to link relevant LLVM components
in test/lib, which broke shared-library builds, and likely made the
conversions unusable outside mlir-opt (or other tools that link in test
library targets). Link these components to GPUCommon instead.
Differential Revision: https://reviews.llvm.org/D80739
Make ConvertKernelFuncToCubin pass to be generic:
- Rename to ConvertKernelFuncToBlob.
- Allow specifying triple, target chip, target features.
- Initializing LLVM backend is supplied by a callback function.
- Lowering process from MLIR module to LLVM module is via another callback.
- Change mlir-cuda-runner to adopt the revised pass.
- Add new tests for lowering to ROCm HSA code object (HSACO).
- Tests for CUDA and ROCm are kept in separate directories.
Differential Revision: https://reviews.llvm.org/D80142
alloc/dealloc/copies.
Add options to LinalgPromotion to use callbacks for implementating the
allocation, deallocation of buffers used for the promoted subviews,
and to copy data into and from the original subviews to the allocated
buffers.
Also some misc. cleanup of the code.
Differential Revision: https://reviews.llvm.org/D80365
Modifying the loop nest builder for generating scf.parallel loops to
not generate scf.parallel loops for non-parallel iterator types in
Linalg operations. The existing implementation incorrectly generated
scf.parallel for all tiled loops. It is rectified by refactoring logic
used while lowering to loops that accounted for this.
Differential Revision: https://reviews.llvm.org/D80188
This revision adds the additional lowering and exposes the patterns at a finer granularity for better programmatic reuse. The unit test makes use of the finer grained pattern for simpler checks.
As the ContractionOpLowering is exposed programmatically, cleanup opportunities appear and static class methods are turned into free functions with static visibility.
Differential Revision: https://reviews.llvm.org/D80375
Summary:
Previously, the only support partial lowering from vector transfers to SCF was
going through loops. This requires a dedicated allocation and extra memory
roundtrips because LLVM aggregates cannot be indexed dynamically (for more
details see the [deep-dive](https://mlir.llvm.org/docs/Dialects/Vector/#deeperdive)).
This revision allows specifying full unrolling which removes this additional roundtrip.
This should be used carefully though because full unrolling will spill, negating the
benefits of removing the interim alloc in the first place.
Proper heuristics are left for a later time.
Differential Revision: https://reviews.llvm.org/D80100
Summary:
This revision refactors the Linalg tiling pass to be written as pattern applications and retires the use of the folder in Linalg tiling.
In the early days, tiling was written as a pass that would create (partially) folded and canonicalized operations on the fly for better composability.
As this evolves towards composition of patterns, the pass-specific folder is counter-productive and is retired.
The tiling options struct evolves to take a tile size creation function which allows materializing tile sizes on the fly (in particular constant tile sizes). This plays better with folding and DCE.
With the folder going away in Tiling, the check on whether subviews are the same in linalg fusion needs to be more robust. This revision also implements such a check.
In the current form, there are still some canonicalizations missing due to AffineMin/Max ops fed by scf::ForOp. These will be improved at a later time.
Differential Revision: https://reviews.llvm.org/D80267
This patch introduces interfaces for read and write ops with affine
restrictions. I used `read`/`write` intead of `load`/`store` for the
interfaces so that they can also be implemented by dma ops.
For now, they are only implemented by affine.load, affine.store,
affine.vector_load and affine.vector_store.
For testing purposes, this patch also migrates affine loop fusion and
required analysis to use the new interfaces. No other changes are made
beyond that.
Co-authored-by: Alex Zinenko <zinenko@google.com>
Reviewed By: bondhugula, ftynse
Differential Revision: https://reviews.llvm.org/D79829
Making these two converters more generic. FunctionAndBlockSignatureConverter now
moves only memref results (after type conversion) to the function argument and
keeps other legal function results unchanged. NonVoidToVoidReturnOpConverter is
renamed to NoBufferOperandsReturnOpConverter. It removes only the buffer
operands from the operands of the converted ReturnOp and inserts CopyOps to copy
each buffer to the target function argument.
Differential Revision: https://reviews.llvm.org/D79329
For now the promoted buffer is indexed using the `full view`. The full view might be
slightly bigger than the partial view (which is accounting for boundaries).
Unfortunately this does not compose easily with other transformations when multiple buffers
with shapes related to each other are involved.
Take `linalg.matmul A B C` (with A of size MxK, B of size KxN and C of size MxN) and suppose we are:
- Tiling over M by 100
- Promoting A only
This is producing a `linalg.matmul promoted_A B subview_C` where `promoted_A` is a promoted buffer
of `A` of size (100xK) and `subview_C` is a subview of size mxK where m could be smaller than 100 due
to boundaries thus leading to a possible incorrect behavior.
We propose to:
- Add a new parameter to the tiling promotion allowing to enable the use of the full tile buffer.
- By default all promoted buffers will be indexed by the partial view.
Note that this could be considered as a breaking change in comparison to the way the tiling promotion
was working.
Differential Revision: https://reviews.llvm.org/D79927
The following Conversions are affected: LoopToStandard -> SCFToStandard,
LoopsToGPU -> SCFToGPU, VectorToLoops -> VectorToSCF. Full file paths are
affected. Additionally, drop the 'Convert' prefix from filenames living under
lib/Conversion where applicable.
API names and CLI options for pass testing are also renamed when applicable. In
particular, LoopsToGPU contains several passes that apply to different kinds of
loops (`for` or `parallel`), for which the original names are preserved.
Differential Revision: https://reviews.llvm.org/D79940
All ops of the SCF dialect now use the `scf.` prefix instead of `loop.`. This
is a part of dialect renaming.
Differential Revision: https://reviews.llvm.org/D79844
Conversion/ folders were originally intended to store patterns for
DialectA->DialectB conversions that depend on both dialects and do not
conceptually belong to either of the dialects. As such, DialectA->DialectA
conversion does not make sense under Conversion/ and should rather live with
the dialect it operates on.
Differential Revision: https://reviews.llvm.org/D79569
Summary:
This revision introduces a helper function to allow applying rewrite patterns, interleaved with more global transformations, in a staged fashion:
1. the first stage consists of an OwningRewritePatternList. The RewritePattern in this list are applied once and in order.
2. the second stage consists of a single OwningRewritePattern that is applied greedily until convergence.
3. the third stage consists of applying a lambda, generally used for non-local transformation effects.
This allows creating custom fused transformations where patterns can be ordered and applied at a finer granularity than a sequence of traditional compiler passes.
A test that exercises these behaviors is added.
Differential Revision: https://reviews.llvm.org/D79518
Summary: This revision introduces LinalgPromotionOptions to more easily control the application of promotion patterns. It also simplifies the different entry points into Promotion in preparation for some behavior change in subsequent revisions.
Differential Revision: https://reviews.llvm.org/D79489
This dialect contains various structured control flow operaitons, not only
loops, reflect this in the name. Drop the Ops suffix for consistency with other
dialects.
Note that this only moves the files and changes the C++ namespace from 'loop'
to 'scf'. The visible IR prefix remains the same and will be updated
separately. The conversions will also be updated separately.
Differential Revision: https://reviews.llvm.org/D79578
Summary:
Adds the loop unroll transformation for loop::ForOp.
Adds support for promoting the body of single-iteration loop::ForOps into its containing block.
Adds check tests for loop::ForOps with dynamic and static lower/upper bounds and step.
Care was taken to share code (where possible) with the AffineForOp unroll transformation to ease maintenance and potential future transition to a LoopLike construct on which loop transformations for different loop types can implemented.
Reviewers: ftynse, nicolasvasilache
Reviewed By: ftynse
Subscribers: bondhugula, mgorny, zzheng, mehdi_amini, rriddle, jpienaar, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, aartbik, liufengdb, Joonsoo, grosul1, frgossen, Kayjukh, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79184
- Exports MLIR targets to be used out-of-tree.
- mimicks `add_clang_library` and `add_flang_library`.
- Fixes libMLIR.so
After https://reviews.llvm.org/D77515 libMLIR.so was no longer containing
any object files. We originally had a cludge there that made it work with
the static initalizers and when switchting away from that to the way the
clang shlib does it, I noticed that MLIR doesn't create a `obj.{name}` target,
and doesn't export it's targets to `lib/cmake/mlir`.
This is due to MLIR using `add_llvm_library` under the hood, which adds
the target to `llvmexports`.
Differential Revision: https://reviews.llvm.org/D78773
[MLIR] Fix libMLIR.so and LLVM_LINK_LLVM_DYLIB
Primarily, this patch moves all mlir references to LLVM libraries into
either LLVM_LINK_COMPONENTS or LINK_COMPONENTS. This enables magic in
the llvm cmake files to automatically replace reference to LLVM components
with references to libLLVM.so when necessary. Among other things, this
completes fixing libMLIR.so, which has been broken for some configurations
since D77515.
Unlike previously, the pattern is now that mlir libraries should almost
always use add_mlir_library. Previously, some libraries still used
add_llvm_library. However, this confuses the export of targets for use
out of tree because libraries specified with add_llvm_library are exported
by LLVM. Instead users which don't need/can't be linked into libMLIR.so
can specify EXCLUDE_FROM_LIBMLIR
A common error mode is linking with LLVM libraries outside of LINK_COMPONENTS.
This almost always results in symbol confusion or multiply defined options
in LLVM when the same object file is included as a static library and
as part of libLLVM.so. To catch these errors more directly, there's now
mlir_check_all_link_libraries.
To simplify usage of add_mlir_library, we assume that all mlir
libraries depend on LLVMSupport, so it's not necessary to separately specify
it.
tested with:
BUILD_SHARED_LIBS=on,
BUILD_SHARED_LIBS=off + LLVM_BUILD_LLVM_DYLIB,
BUILD_SHARED_LIBS=off + LLVM_BUILD_LLVM_DYLIB + LLVM_LINK_LLVM_DYLIB.
By: Stephen Neuendorffer <stephen.neuendorffer@xilinx.com>
Differential Revision: https://reviews.llvm.org/D79067
[MLIR] Move from using target_link_libraries to LINK_LIBS
This allows us to correctly generate dependencies for derived targets,
such as targets which are created for object libraries.
By: Stephen Neuendorffer <stephen.neuendorffer@xilinx.com>
Differential Revision: https://reviews.llvm.org/D79243
Three commits have been squashed to avoid intermediate build breakage.
Linalg transformations are currently exposed as DRRs.
Unfortunately RewriterGen does not play well with the line of work on named linalg ops which require variadic operands and results.
Additionally, DRR is arguably not the right abstraction to expose compositions of such patterns that don't rely on SSA use-def semantics.
This revision abandons DRRs and exposes manually written C++ patterns.
Refactorings and cleanups are performed to uniformize APIs.
This refactoring will allow replacing the currently manually specified Linalg named ops.
A collateral victim of this refactoring is the `tileAndFuse` DRR, and the one associated test, which will be revived at a later time.
Lastly, the following 2 tests do not add value and are altered:
- a dot_perm tile + interchange test does not test anything new and is removed
- a dot tile + lower to loops does not need 2-D tiling and is trimmed.
These libraries are distinct from other things in Analysis in that they
operate only on core IR concepts. This also simplifies dependencies
so that Dialect -> Analysis -> Parser -> IR. Previously, the parser depended
on portions of the the Analysis directory as well, which sometimes
caused issues with the way the cmake makefile generator discovers
dependencies on generated files during compilation.
Differential Revision: https://reviews.llvm.org/D79240
Summary:
This revision cleans up a layer of complexity in ScopedContext and uses InsertGuard instead of previously manual bookkeeping.
The method `getBuilder` is renamed to `getBuilderRef` and spurious copies of OpBuilder are tracked.
This results in some canonicalizations not happening anymore in the Linalg matmul to vector test. This test is retired because relying on DRRs for this has been shaky at best. The solution will be better support to write fused passes in C++ with more idiomatic pattern composition and application.
Differential Revision: https://reviews.llvm.org/D79208
This revision allows masked vector transfers with m-D buffers and n-D vectors to
progressively lower to m-D buffer and 1-D vector transfers.
For a vector.transfer_read, assuming a `memref<(leading_dims) x (major_dims) x (minor_dims) x type>` and a `vector<(minor_dims) x type>` are involved in the transfer, this generates pseudo-IR resembling:
```
if (any_of(%ivs_major + %offsets, <, major_dims)) {
%v = vector_transfer_read(
{%offsets_leading, %ivs_major + %offsets_major, %offsets_minor},
%ivs_minor):
memref<(leading_dims) x (major_dims) x (minor_dims) x type>,
vector<(minor_dims) x type>;
} else {
%v = splat(vector<(minor_dims) x type>, %fill)
}
```
Differential Revision: https://reviews.llvm.org/D79062
We have provided a generic buffer assignment transformation ported from
TensorFlow. This generic transformation pass automatically analyzes the values
and their aliases (also in other blocks) and returns the valid positions for
Alloc and Dealloc operations. To find these positions, the algorithm uses the
block Dominator and Post-Dominator analyses. In our proposed algorithm, we have
considered aliasing, liveness, nested regions, branches, conditional branches,
critical edges, and independency to custom block terminators. This
implementation doesn't support block loops. However, we have considered this in
our design. For this purpose, it is only required to have a loop analysis to
insert Alloc and Dealloc operations outside of these loops in some special
cases.
Differential Revision: https://reviews.llvm.org/D78484
