The greedy rewriter is used in many different flows and it has a lot of
convenience (work list management, debugging actions, tracing, etc). But
it combines two kinds of greedy behavior 1) how ops are matched, 2)
folding wherever it can.
These are independent forms of greedy and leads to inefficiency. E.g.,
cases where one need to create different phases in lowering and is
required to applying patterns in specific order split across different
passes. Using the driver one ends up needlessly retrying folding/having
multiple rounds of folding attempts, where one final run would have
sufficed.
Of course folks can locally avoid this behavior by just building their
own, but this is also a common requested feature that folks keep on
working around locally in suboptimal ways.
For downstream users, there should be no behavioral change. Updating
from the deprecated should just be a find and replace (e.g., `find ./
-type f -exec sed -i
's|applyPatternsAndFoldGreedily|applyPatternsGreedily|g' {} \;` variety)
as the API arguments hasn't changed between the two.
- Extend the SPIRV::LowerABIAttributesPass to detect when the target env
is using SPIR-V ver >= 1.4, and in this case add all the functions'
interface storage variables to the spirv.EntryPoint calls, as required
by the spec of OpEntryPoint:
"_Before version 1.4, the interface’s storage classes are limited to the
Input and Output storage classes. Starting with version 1.4, the
interface’s storage classes are all storage classes used in declaring
all global variables referenced by the entry point’s call tree_."
- Fix: generate the replacement ops (spirv.AddressOf and .AccessChain)
in the order in which the associated variable appears in the function
signature
Signed-off-by: Fabrizio Indirli <Fabrizio.Indirli@arm.com>
This commit simplifies the result type of materialization functions.
Previously: `std::optional<Value>`
Now: `Value`
The previous implementation allowed 3 possible return values:
- Non-null value: The materialization function produced a valid
materialization.
- `std::nullopt`: The materialization function failed, but another
materialization can be attempted.
- `Value()`: The materialization failed and so should the dialect
conversion. (Previously: Dialect conversion can roll back.)
This commit removes the last variant. It is not particularly useful
because the dialect conversion will fail anyway if all other
materialization functions produced `std::nullopt`.
Furthermore, in contrast to type conversions, at least one
materialization callback is expected to succeed. In case of a failing
type conversion, the current dialect conversion can roll back and try a
different pattern. This also used to be the case for materializations,
but that functionality was removed with #107109: failed materializations
can no longer trigger a rollback. (They can just make the entire dialect
conversion fail without rollback.) With this in mind, it is even less
useful to have an additional error state for materialization functions.
This commit is in preparation of merging the 1:1 and 1:N type
converters. Target materializations will have to return multiple values
instead of a single one. With this commit, we can keep the API simple:
`SmallVector<Value>` instead of `std::optional<SmallVector<Value>>`.
Note for LLVM integration: All 1:1 materializations should return
`Value` instead of `std::optional<Value>`. Instead of `std::nullopt`
return `Value()`.
This commit marks the type converter in `populate...` functions as
`const`. This is useful for debugging.
Patterns already take a `const` type converter. However, some
`populate...` functions do not only add new patterns, but also add
additional type conversion rules. That makes it difficult to find the
place where a type conversion was added in the code base. With this
change, all `populate...` functions that only populate pattern now have
a `const` type converter. Programmers can then conclude from the
function signature that these functions do not register any new type
conversion rules.
Also some minor cleanups around the 1:N dialect conversion
infrastructure, which did not always pass the type converter as a
`const` object internally.
### Description
This PR builds on #99872. It implements a minimal version of function
body vector unrolling to convert vector types into 1D and with a size
supported by SPIR-V (2, 3 or 4 depending on the original dimension). The
ops that are currently supported include those with elementwise traits
(e.g. `arith.addi`), `vector.reduction` and `vector.transpose`. This PR
also includes new LIT tests that only check for vector unrolling.
### Future Plans
- Support more ops
---------
Co-authored-by: Jakub Kuderski <kubakuderski@gmail.com>
### Description
This PR implements a minimal version of function signature conversion to
unroll vectors into 1D and with a size supported by SPIR-V (2, 3 or 4
depending on the original dimension). This PR also includes new unit
tests that only check for function signature conversion.
### Future Plans
- Check for capabilities that support vectors of size 8 or 16.
- Set up `OneToNTypeConversion` and `DialectConversion` to replace the
current implementation that uses `GreedyPatternRewriteDriver`.
- Introduce other vector unrolling patterns to cancel out the
`vector.insert_strided_slice` and `vector.extract_strided_slice` ops and
fully legalize the vector types in the function body.
- Handle `func::CallOp` and declarations.
- Restructure the code in `SPIRVConversion.cpp`.
- Create test passes for testing sets of patterns in isolation.
- Optimize the way original shape is splitted into target shapes, e.g.
`vector<5xi32>` can be splitted into `vector<4xi32>` and
`vector<1xi32>`.
---------
Co-authored-by: Jakub Kuderski <kubakuderski@gmail.com>
There are execution modes need target width as their extra operands.
SignedZeroInfNanPreserve is one of them. This patch adds `target width`
as one of SPIR-V ABI attributes.
These non-finite math ops are supported by SPIR-V but not by WGSL.
Assume finite floating point values and expand these ops into `false`.
Previously, this worked by adding fast math flags during conversion from
arith to spirv, but this got removed in
https://github.com/llvm/llvm-project/pull/86578.
Also do some misc cleanups in the surrounding code.
Investigate the lowering of MemRef Load/Store ops and implement
additional folding of created ops
Aims to improve readability of generated lowered SPIR-V code.
Part of work llvm#70704
This commit renames 4 pattern rewriter API functions:
* `updateRootInPlace` -> `modifyOpInPlace`
* `startRootUpdate` -> `startOpModification`
* `finalizeRootUpdate` -> `finalizeOpModification`
* `cancelRootUpdate` -> `cancelOpModification`
The term "root" is a misnomer. The root is the op that a rewrite pattern
matches against
(https://mlir.llvm.org/docs/PatternRewriter/#root-operation-name-optional).
A rewriter must be notified of all in-place op modifications, not just
in-place modifications of the root
(https://mlir.llvm.org/docs/PatternRewriter/#pattern-rewriter). The old
function names were confusing and have contributed to various broken
rewrite patterns.
Note: The new function names use the term "modify" instead of "update"
for consistency with the `RewriterBase::Listener` terminology
(`notifyOperationModified`).
WebGPU does not currently support extended arithmetic, this is an issue
when we want to lower from SPIR-V. This commit adds a pattern to
transform and emulate spirv.IAddCarry with spirv.IAdd operations
Fixes#65154
ConversionPatterns do not (and should not) modify the type converter that they are using.
* Make `ConversionPattern::typeConverter` const.
* Make member functions of the `LLVMTypeConverter` const.
* Conversion patterns take a const type converter.
* Various helper functions (that are called from patterns) now also take a const type converter.
Differential Revision: https://reviews.llvm.org/D157601
This would mean allowing vector type conversion involving sub-byte
element types.
Reviewed By: kuhar
Differential Revision: https://reviews.llvm.org/D157791
1-element vectors are not valid in SPIR-V and fail `Bitcast` op verification.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D156207
In SPIR-V, the capabilities for storage and compute are separate.
We have good handling of the storage side in general via MemRef
type conversion and various `memref` dialect ops.
Once the value was loaded properly, if the compute capability is
supported directly, we don't need to emulate like the storage side
with int32. However, we do need to make sure casting ops are
properly inserted to chain the flow to go back to the original
bitwidth.
Right now that is done in the each individual pattern directly,
which put lots of pressure that shouldn't be on the patterns and
causes duplication and trickiness w.r.t. capability check and such.
Instead, we should handle such casting within the SPIR-V conversion
framework using `addSourceMaterialization`, where we can check with
the target environment to make sure the corresponding compute
capability is allowed and then we can materialize and SPIR-V casting
op.
Along the way, we can drop all the duplicated cast materialization
registration in various places.
Reviewed By: kuhar
Differential Revision: https://reviews.llvm.org/D155118
The SPIR-V spec does not specify the mangling for these variables, so
the conversion to SPIR-V should be flexible enough to allow adding a
custom prefix and suffix to the core name.
Differential Revision: https://reviews.llvm.org/D153951
Signed-off-by: Victor Perez <victor.perez@codeplay.com>
The MLIR classes Type/Attribute/Operation/Op/Value support
cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast
functionality in addition to defining methods with the same name.
This change begins the migration of uses of the method to the
corresponding function call as has been decided as more consistent.
Note that there still exist classes that only define methods directly,
such as AffineExpr, and this does not include work currently to support
a functional cast/isa call.
Caveats include:
- This clang-tidy script probably has more problems.
- This only touches C++ code, so nothing that is being generated.
Context:
- https://mlir.llvm.org/deprecation/ at "Use the free function variants
for dyn_cast/cast/isa/…"
- Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443
Implementation:
This first patch was created with the following steps. The intention is
to only do automated changes at first, so I waste less time if it's
reverted, and so the first mass change is more clear as an example to
other teams that will need to follow similar steps.
Steps are described per line, as comments are removed by git:
0. Retrieve the change from the following to build clang-tidy with an
additional check:
https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check
1. Build clang-tidy
2. Run clang-tidy over your entire codebase while disabling all checks
and enabling the one relevant one. Run on all header files also.
3. Delete .inc files that were also modified, so the next build rebuilds
them to a pure state.
4. Some changes have been deleted for the following reasons:
- Some files had a variable also named cast
- Some files had not included a header file that defines the cast
functions
- Some files are definitions of the classes that have the casting
methods, so the code still refers to the method instead of the
function without adding a prefix or removing the method declaration
at the same time.
```
ninja -C $BUILD_DIR clang-tidy
run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\
-header-filter=mlir/ mlir/* -fix
rm -rf $BUILD_DIR/tools/mlir/**/*.inc
git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\
mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\
mlir/lib/**/IR/\
mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\
mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\
mlir/test/lib/Dialect/Test/TestTypes.cpp\
mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\
mlir/test/lib/Dialect/Test/TestAttributes.cpp\
mlir/unittests/TableGen/EnumsGenTest.cpp\
mlir/test/python/lib/PythonTestCAPI.cpp\
mlir/include/mlir/IR/
```
Differential Revision: https://reviews.llvm.org/D150123
Typically GPUs cannot access memory in sub-byte manner. So for
sub-byte integer type values, we need to either expand them to
full bytes or tightly pack them. This commit adds support for
tightly packed power-of-two sub-byte types.
Sub-byte types aren't allowed in SPIR-V spec, so there are no
compute/storage capability for them like other supported integer
types. So we don't recognize sub-byte types in `spirv::ScalarType`.
We just special case them in type converter and always convert
to use i32 under the hood.
Reviewed By: kuhar
Differential Revision: https://reviews.llvm.org/D150395
Currently conversions to interfaces may happen implicitly (e.g.
`Attribute -> TypedAttr`), failing a runtime assert if the interface
isn't actually implemented. This change marks the `Interface(ValueT)`
constructor as explicit so that a cast is required.
Where it was straightforward to I adjusted code to not require casts,
otherwise I just made them explicit.
Depends on D148491, D148492
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D148493
This commit adds conversion from complex construction and
extraction ops to SPIR-V. Other arithemtic ops can be done
via ComplexToStandard patterns.
Reviewed By: kuhar
Differential Revision: https://reviews.llvm.org/D147193
Complex types are converted to a two-element vector type to contain
the real and imaginary numbers.
Reviewed By: kuhar
Differential Revision: https://reviews.llvm.org/D147188
We do not plan to handle wide vector types with SPIR-V arrays anymore.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D143777
For the cases where we have aliases of `vector<4xf16>` and
`vector<4xf32>`, we need to do casting before composite
construction.
Reviewed By: kuhar
Differential Revision: https://reviews.llvm.org/D143042
Fix an off-by-one error in extended umul extension for WebGPU.
Revert to the long multiplication algorithm originally added to wide
integer emulation, which was deleted in D139776. It is much easier
to see why it is correct.
Add runtime tests based on the mlir-vulkan-runner. These run both with
and without umul extension.
Issue: https://github.com/llvm/llvm-project/issues/59563
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D141085
This is needed because WGSL does not yet support extended multiplication
ops.
Set up pattern/pass stuff and handle the first op: `UMulExtended`.
`SMulExtended` handling will go to a separate patch.
Issue: https://github.com/llvm/llvm-project/issues/59563
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D140995
This is part of an effort to migrate from llvm::Optional to
std::optional. 22426110c5 changed the way mlir-tblgen generates .inc
files, emitting std::optional when an Optional attribute is specified in
a .td file. It also changed several .td files hard-coding llvm::Optional
to use std::optional. However, the patch excluded a few .td files in
SPIRV and Bufferization hard-coding llvm::Optional. This patch fixes
that defect, and after this patch, references to llvm::Optional in .cpp
and .h files can be replaced mechanically.
See also: https://discourse.llvm.org/t/deprecating-llvm-optional-x-hasvalue-getvalue-getvalueor/63716
Signed-off-by: Ramkumar Ramachandra <r@artagnon.com>
Differential Revision: https://reviews.llvm.org/D140329
std::optional::value() has undesired exception checking semantics and is
unavailable in older Xcode (see _LIBCPP_AVAILABILITY_BAD_OPTIONAL_ACCESS). The
call sites block std::optional migration.
This is part of an effort to migrate from llvm::Optional to
std::optional. This patch changes the way mlir-tblgen generates .inc
files, and modifies tests and documentation appropriately. It is a "no
compromises" patch, and doesn't leave the user with an unpleasant mix of
llvm::Optional and std::optional.
A non-trivial change has been made to ControlFlowInterfaces to split one
constructor into two, relating to a build failure on Windows.
See also: https://discourse.llvm.org/t/deprecating-llvm-optional-x-hasvalue-getvalue-getvalueor/63716
Signed-off-by: Ramkumar Ramachandra <r@artagnon.com>
Differential Revision: https://reviews.llvm.org/D138934
Reland D139447, D139471 With flang actually working
- FunctionOpInterface: make get/setFunctionType interface methods
This patch removes the concept of a `function_type`-named type attribute
as a requirement for implementors of FunctionOpInterface. Instead, this
type should be provided through two interface methods, `getFunctionType`
and `setFunctionTypeAttr` (*Attr because functions may use different
concrete function types), which should be automatically implemented by
ODS for ops that define a `$function_type` attribute.
This also allows FunctionOpInterface to materialize function types if
they don't carry them in an attribute, for example.
Importantly, all the function "helper" still accept an attribute name to
use in parsing and printing functions, for example.
- FunctionOpInterface: arg and result attrs dispatch to interface
This patch removes the `arg_attrs` and `res_attrs` named attributes as a
requirement for FunctionOpInterface and replaces them with interface
methods for the getters, setters, and removers of the relevent
attributes. This allows operations to use their own storage for the
argument and result attributes.
Reviewed By: jpienaar
Differential Revision: https://reviews.llvm.org/D139736
