The ops are very similar to the std variants, but support async GPU execution.
gpu.alloc does not currently support an alignment attribute, and the new ops do not have
canonicalizers/folders like their std siblings do.
Reviewed By: herhut
Differential Revision: https://reviews.llvm.org/D91698
Introduce a conversion pass from SCF parallel loops to OpenMP dialect
constructs - parallel region and workshare loop. Loops with reductions are not
supported because the OpenMP dialect cannot model them yet.
The conversion currently targets only one level of parallelism, i.e. only
one top-level `omp.parallel` operation is produced even if there are nested
`scf.parallel` operations that could be mapped to `omp.wsloop`. Nested
parallelism support is left for future work.
Reviewed By: kiranchandramohan
Differential Revision: https://reviews.llvm.org/D91982
This revision will make it easier to create new ops base on the strided memref abstraction outside of the std dialect.
OffsetSizeAndStrideOpInterface is an interface for ops that allow specifying mixed dynamic and static offsets, sizes and strides variadic operands.
Ops that implement this interface need to expose the following methods:
1. `getArrayAttrRanks` to specify the length of static integer
attributes.
2. `offsets`, `sizes` and `strides` variadic operands.
3. `static_offsets`, resp. `static_sizes` and `static_strides` integer
array attributes.
The invariants of this interface are:
1. `static_offsets`, `static_sizes` and `static_strides` have length
exactly `getArrayAttrRanks()`[0] (resp. [1], [2]).
2. `offsets`, `sizes` and `strides` have each length at most
`getArrayAttrRanks()`[0] (resp. [1], [2]).
3. if an entry of `static_offsets` (resp. `static_sizes`,
`static_strides`) is equal to a special sentinel value, namely
`ShapedType::kDynamicStrideOrOffset` (resp. `ShapedType::kDynamicSize`,
`ShapedType::kDynamicStrideOrOffset`), then the corresponding entry is
a dynamic offset (resp. size, stride).
4. a variadic `offset` (resp. `sizes`, `strides`) operand must be present
for each dynamic offset (resp. size, stride).
This interface is useful to factor out common behavior and provide support
for carrying or injecting static behavior through the use of the static
attributes.
Differential Revision: https://reviews.llvm.org/D92011
It is a simple conversion that only requires to change the region argument
types, generalize it from ParallelOp.
Reviewed By: kiranchandramohan
Differential Revision: https://reviews.llvm.org/D91989
The existing implementation of the conversion from SCF Parallel operation to
SCF "for" loops in order to further convert those loops to branch-based CFG has
been cloning the loop and reduction body operations into the new loop because
ConversionPatternRewriter was missing support for moving blocks while replacing
their arguments. This functionality now available, use it to implement the
conversion and avoid cloning operations, which may lead to doubling of the IR
size during the conversion.
In addition, this fixes an issue with converting nested SCF "if" conditionals
present in "parallel" operations that would cause the conversion infrastructure
to stop because of the repeated application of the pattern converting "newly"
created "if"s (which were in fact just moved). Arguably, this should be fixed
at the infrastructure level and this fix is a workaround.
Reviewed By: herhut
Differential Revision: https://reviews.llvm.org/D91955
All these potential null pointer dereferences are reported by my static analyzer for null smart pointer dereferences, which has a different implementation from `alpha.cplusplus.SmartPtr`.
The checked pointers in this patch are initialized by Target::createXXX functions. When the creator function pointer is not correctly set, a null pointer will be returned, or the creator function may originally return a null pointer.
Some of them may not make sense as they may be checked before entering the function, but I fixed them all in this patch. I submit this fix because 1) similar checks are found in some other places in the LLVM codebase for the same return value of the function; and, 2) some of the pointers are dereferenced before they are checked, which may definitely trigger a null pointer dereference if the return value is nullptr.
Reviewed By: tejohnson, MaskRay, jpienaar
Differential Revision: https://reviews.llvm.org/D91410
Depends On D89963
**Automatic reference counting algorithm outline:**
1. `ReturnLike` operations forward the reference counted values without
modifying the reference count.
2. Use liveness analysis to find blocks in the CFG where the lifetime of
reference counted values ends, and insert `drop_ref` operations after
the last use of the value.
3. Insert `add_ref` before the `async.execute` operation capturing the
value, and pairing `drop_ref` before the async body region terminator,
to release the captured reference counted value when execution
completes.
4. If the reference counted value is passed only to some of the block
successors, insert `drop_ref` operations in the beginning of the blocks
that do not have reference coutned value uses.
Reviewed By: silvas
Differential Revision: https://reviews.llvm.org/D90716
Null types are commonly used as an error marker. Catch them in the constructor
of Operation if they are present in the result type list, as otherwise this
could lead to further surprising behavior when querying op result types.
Fix AsyncToLLVM and StandardToLLVM that were using null types when constructing
operations.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D91770
Additionally, clear a data structure to ensure a proper state if multiple conversion attempts are needed.
Differential Revision: https://reviews.llvm.org/D91791
Make the interface match the one of ConvertToLLVMPattern::getDataPtr() (to be removed in a separate change).
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D91599
std.alloc only supports memrefs with identity layout, which means we can simplify the lowering to LLVM and compute strides only from (static and dynamic) sizes.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D91549
This commit does the renaming mentioned in the title in order to bring
`spv` dialect closer to the MLIR naming conventions.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D91609
These includes have been deprecated in favor of BuiltinDialect.h, which contains the definitions of ModuleOp and FuncOp.
Differential Revision: https://reviews.llvm.org/D91572
The current code allows strided layouts, but the number of elements allocated is ambiguous. It could be either the number of elements in the shape (the current implementation), or the amount of elements required to not index out-of-bounds with the given maps (which would require evaluating the layout map).
If we require the canonical layouts, the two will be the same.
Reviewed By: nicolasvasilache, ftynse
Differential Revision: https://reviews.llvm.org/D91523
The logic of vector on boolean was missed. This patch adds the logic and test on
it.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D91403
Depends On D89958
1. Adds `async.group`/`async.awaitall` to group together multiple async tokens/values
2. Rewrite scf.parallel operation into multiple concurrent async.execute operations over non overlapping subranges of the original loop.
Example:
```
scf.for (%i, %j) = (%lbi, %lbj) to (%ubi, %ubj) step (%si, %sj) {
"do_some_compute"(%i, %j): () -> ()
}
```
Converted to:
```
%c0 = constant 0 : index
%c1 = constant 1 : index
// Compute blocks sizes for each induction variable.
%num_blocks_i = ... : index
%num_blocks_j = ... : index
%block_size_i = ... : index
%block_size_j = ... : index
// Create an async group to track async execute ops.
%group = async.create_group
scf.for %bi = %c0 to %num_blocks_i step %c1 {
%block_start_i = ... : index
%block_end_i = ... : index
scf.for %bj = %c0 t0 %num_blocks_j step %c1 {
%block_start_j = ... : index
%block_end_j = ... : index
// Execute the body of original parallel operation for the current
// block.
%token = async.execute {
scf.for %i = %block_start_i to %block_end_i step %si {
scf.for %j = %block_start_j to %block_end_j step %sj {
"do_some_compute"(%i, %j): () -> ()
}
}
}
// Add produced async token to the group.
async.add_to_group %token, %group
}
}
// Await completion of all async.execute operations.
async.await_all %group
```
In this example outer loop launches inner block level loops as separate async
execute operations which will be executed concurrently.
At the end it waits for the completiom of all async execute operations.
Reviewed By: ftynse, mehdi_amini
Differential Revision: https://reviews.llvm.org/D89963
This exposes a hook to configure legality of operations such that only
`scf.parallel` operations that have mapping attributes are marked as
illegal. Consequently, the transformation can now also be applied to
mixed forms.
Differential Revision: https://reviews.llvm.org/D91340
- Move isSupportedMemRefType() to ConvertToLLVMPatterns and check if the
memref element type is supported there.
Differential Revision: https://reviews.llvm.org/D91374
This patch introduces a new conversion pattern for `spv.ExecutionMode`.
`spv.ExecutionMode` may contain important information about the entry
point, which we want to preserve. For example, `LocalSize` provides
information about the work-group size that can be reused. Hence, the
pattern for entry-point ops changes to the following:
- `spv.EntryPoint` is still simply removed
- Info from `spv.ExecutionMode` is used to create a global struct variable,
which looks like:
```
struct {
int32_t executionMode;
int32_t values[]; // optional values
};
```
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D89989
VectorInsertDynamicOp in SPIRV dialect
conversion from vector.insertelement to spirv VectorInsertDynamicOp
Differential Revision: https://reviews.llvm.org/D90927
The pass combines patterns of ExpandAtomic, ExpandMemRefReshape,
StdExpandDivs passes. The pass is meant to legalize STD for conversion to LLVM.
Differential Revision: https://reviews.llvm.org/D91082
- Convert `global_memref` to LLVM::GlobalOp.
- Convert `get_global_memref` to a memref descriptor with a pointer to the first element
of the global stashed in it.
- Extend unit test and a mlir-cpu-runner test to validate the generated LLVM IR.
Differential Revision: https://reviews.llvm.org/D90803
Since SPIR-V module has an optional name, this patch
makes a change to pass it to `ModuleOp` during conversion.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D90904
VectorExtractDynamicOp in SPIRV dialect
conversion from vector.extractelement to spirv VectorExtractDynamicOp
Differential Revision: https://reviews.llvm.org/D90679
- Eliminate duplicated information about mapping from memref -> its descriptor fields
by consolidating that mapping in two functions: getMemRefDescriptorFields and
getUnrankedMemRefDescriptorFields.
- Change convertMemRefType() and convertUnrankedMemRefType() to use these
functions.
- Remove convertMemrefSignature and convertUnrankedMemrefSignature.
Differential Revision: https://reviews.llvm.org/D90707
When the "after" region of a WhileOp is merely forwarding its arguments back to
the "before" region, i.e. WhileOp is a canonical do-while loop, a simpler CFG
subgraph that omits the "after" region with its extra branch operation can be
produced. Loop rotation from general "while" to "if { do-while }" is left for a
future canonicalization pattern when it becomes necessary.
Differential Revision: https://reviews.llvm.org/D90604
The lowering is a straightforward inlining of the "before" and "after" regions
connected by (conditional) branches. This plugs the WhileOp into the
progressive lowering scheme. Future commits may choose to target WhileOp
instead of CFG when lowering ForOp.
Differential Revision: https://reviews.llvm.org/D90603
Because cstr operations allow more instruction reordering than asserts, we only
lower cstr_broadcastable to std ops with cstr_require. This ensures that the
more drastic lowering to asserts can happen specifically with the user's desire.
Differential Revision: https://reviews.llvm.org/D89325
For the synchronous case, destroy the stream after synchronization.
Sneak in a unrelated change to report why the gpu.wait conversion pattern didn't match.
Reviewed By: herhut
Differential Revision: https://reviews.llvm.org/D89933
This class represents a rewrite pattern list that has been frozen, and thus immutable. This replaces the uses of OwningRewritePatternList in pattern driver related API, such as dialect conversion. When PDL becomes more prevalent, this API will allow for optimizing a set of patterns once without the need to do this per run of a pass.
Differential Revision: https://reviews.llvm.org/D89104
There are several pieces of pattern rewriting infra in IR/ that really shouldn't be there. This revision moves those pieces to a better location such that they are easier to evolve in the future(e.g. with PDL). More concretely this revision does the following:
* Create a Transforms/GreedyPatternRewriteDriver.h and move the apply*andFold methods there.
The definitions for these methods are already in Transforms/ so it doesn't make sense for the declarations to be in IR.
* Create a new lib/Rewrite library and move PatternApplicator there.
This new library will be focused on applying rewrites, and will also include compiling rewrites with PDL.
Differential Revision: https://reviews.llvm.org/D89103
The Pattern class was originally intended to be used for solely matching operations, but that use never materialized. All of the pattern infrastructure uses RewritePattern, and the infrastructure for pure matching(Matchers.h) is implemented inline. This means that this class isn't a useful abstraction at the moment, so this revision refactors it to solely encapsulate the "metadata" of a pattern. The metadata includes the various state describing a pattern; benefit, root operation, etc. The API on PatternApplicator is updated to now operate on `Pattern`s as nothing special from `RewritePattern` is necessary.
This refactoring is also necessary for the upcoming use of PDL patterns alongside C++ rewrite patterns.
Differential Revision: https://reviews.llvm.org/D86258
The conversion between PDL and the interpreter is split into several different parts.
** The Matcher:
The matching section of all incoming pdl.pattern operations is converted into a predicate tree and merged. Each pattern is first converted into an ordered list of predicates starting from the root operation. A predicate is composed of three distinct parts:
* Position
- A position refers to a specific location on the input DAG, i.e. an
existing MLIR entity being matched. These can be attributes, operands,
operations, results, and types. Each position also defines a relation to
its parent. For example, the operand `[0] -> 1` has a parent operation
position `[0]` (the root).
* Question
- A question refers to a query on a specific positional value. For
example, an operation name question checks the name of an operation
position.
* Answer
- An answer is the expected result of a question. For example, when
matching an operation with the name "foo.op". The question would be an
operation name question, with an expected answer of "foo.op".
After the predicate lists have been created and ordered(based on occurrence of common predicates and other factors), they are formed into a tree of nodes that represent the branching flow of a pattern match. This structure allows for efficient construction and merging of the input patterns. There are currently only 4 simple nodes in the tree:
* ExitNode: Represents the termination of a match
* SuccessNode: Represents a successful match of a specific pattern
* BoolNode/SwitchNode: Branch to a specific child node based on the expected answer to a predicate question.
Once the matcher tree has been generated, this tree is walked to generate the corresponding interpreter operations.
** The Rewriter:
The rewriter portion of a pattern is generated in a very straightforward manor, similarly to lowerings in other dialects. Each PDL operation that may exist within a rewrite has a mapping into the interpreter dialect. The code for the rewriter is generated within a FuncOp, that is invoked by the interpreter on a successful pattern match. Referenced values defined in the matcher become inputs the generated rewriter function.
An example lowering is shown below:
```mlir
// The following high level PDL pattern:
pdl.pattern : benefit(1) {
%resultType = pdl.type
%inputOperand = pdl.input
%root, %results = pdl.operation "foo.op"(%inputOperand) -> %resultType
pdl.rewrite %root {
pdl.replace %root with (%inputOperand)
}
}
// is lowered to the following:
module {
// The matcher function takes the root operation as an input.
func @matcher(%arg0: !pdl.operation) {
pdl_interp.check_operation_name of %arg0 is "foo.op" -> ^bb2, ^bb1
^bb1:
pdl_interp.return
^bb2:
pdl_interp.check_operand_count of %arg0 is 1 -> ^bb3, ^bb1
^bb3:
pdl_interp.check_result_count of %arg0 is 1 -> ^bb4, ^bb1
^bb4:
%0 = pdl_interp.get_operand 0 of %arg0
pdl_interp.is_not_null %0 : !pdl.value -> ^bb5, ^bb1
^bb5:
%1 = pdl_interp.get_result 0 of %arg0
pdl_interp.is_not_null %1 : !pdl.value -> ^bb6, ^bb1
^bb6:
// This operation corresponds to a successful pattern match.
pdl_interp.record_match @rewriters::@rewriter(%0, %arg0 : !pdl.value, !pdl.operation) : benefit(1), loc([%arg0]), root("foo.op") -> ^bb1
}
module @rewriters {
// The inputs to the rewriter from the matcher are passed as arguments.
func @rewriter(%arg0: !pdl.value, %arg1: !pdl.operation) {
pdl_interp.replace %arg1 with(%arg0)
pdl_interp.return
}
}
}
```
Differential Revision: https://reviews.llvm.org/D84580
This patch introduces a pass for running
`mlir-spirv-cpu-runner` - LowerHostCodeToLLVMPass.
This pass emulates `gpu.launch_func` call in LLVM dialect and lowers
the host module code to LLVM. It removes the `gpu.module`, creates a
sequence of global variables that are later linked to the varables
in the kernel module, as well as a series of copies to/from
them to emulate the memory transfer to/from the host or to/from the
device sides. It also converts the remaining Standard dialect into
LLVM dialect, emitting C wrappers.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D86112
This reverts commit 4986d5eaff with
proper patches to CMakeLists.txt:
- Add MLIRAsync as a dependency to MLIRAsyncToLLVM
- Add Coroutines as a dependency to MLIRExecutionEngine
Lower from Async dialect to LLVM by converting async regions attached to `async.execute` operations into LLVM coroutines (https://llvm.org/docs/Coroutines.html):
1. Outline all async regions to functions
2. Add LLVM coro intrinsics to mark coroutine begin/end
3. Use MLIR conversion framework to convert all remaining async types and ops to LLVM + Async runtime function calls
All `async.await` operations inside async regions converted to coroutine suspension points. Await operation outside of a coroutine converted to the blocking wait operations.
Implement simple runtime to support concurrent execution of coroutines.
Reviewed By: herhut
Differential Revision: https://reviews.llvm.org/D89292
Forward missing attributes when creating the new transfer op otherwise the
builder would use default values.
Differential Revision: https://reviews.llvm.org/D89907
This still satisfies the constraints required by the affine dialect and
gives more flexibility in what iteration bounds can be used when
loewring to the GPU dialect.
Differential Revision: https://reviews.llvm.org/D89782
Now, convert-shape-to-std doesn't internally create memrefs, which was
previously a bit of a layering violation. The conversion to memrefs
should logically happen as part of bufferization.
Differential Revision: https://reviews.llvm.org/D89669
This revision adds a programmable codegen strategy from linalg based on staged rewrite patterns. Testing is exercised on a simple linalg.matmul op.
Differential Revision: https://reviews.llvm.org/D89374
For some reason the variable `cumulativeSizeInBytes` in
`getCumulativeSizeInBytes` was actually storing number of elements. I decided
to fix it and refactor the function a bit.
Differential Revision: https://reviews.llvm.org/D89336
This revision introduces support for buffer allocation for any named linalg op.
To avoid template instantiating many ops, a new ConversionPattern is created to capture the LinalgOp interface.
Some APIs are updated to remain consistent with MLIR style:
`OwningRewritePatternList * -> OwningRewritePatternList &`
`BufferAssignmentTypeConverter * -> BufferAssignmentTypeConverter &`
Differential revision: https://reviews.llvm.org/D89226
This is required or broadcasting with operands of different ranks will lead to
failures as the select op requires both possible outputs and its output type to
be the same.
Differential Revision: https://reviews.llvm.org/D89134
This revision belongs to a series of patches that reduce reliance of Linalg transformations on templated rewrite and conversion patterns.
Instead, this uses a MatchAnyTag pattern for the vast majority of cases and dispatches internally.
Differential Revision: https://reviews.llvm.org/D89133
This revision also inserts an end-to-end test that lowers tensors to buffers all the way to executable code on CPU.
Differential revision: https://reviews.llvm.org/D88998
The patch fixes the types used to access the elements of the kernel parameter structure from a pointer to the structure to a pointer to the actual parameter type.
Reviewed By: csigg
Differential Revision: https://reviews.llvm.org/D88959
Add conversion pass for Vector dialect to SPIR-V dialect and add some simple
conversion pattern for vector.broadcast, vector.insert, vector.extract.
Differential Revision: https://reviews.llvm.org/D88761
A pattern to convert `spv.CompositeInsert` and `spv.CompositeExtract`.
In LLVM, there are 2 ops that correspond to each instruction depending
on the container type. If the container type is a vector type, then
the result of conversion is `llvm.insertelement` or `llvm.extractelement`.
If the container type is an aggregate type (i.e. struct, array), the
result of conversion is `llvm.insertvalue` or `llvm.extractvalue`.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D88205
The previous code did the lowering to alloca, malloc, and aligned_malloc
in a single class with different code paths that are somewhat difficult to
follow.
This change moves the common code to a base class and has a separte
derived class per lowering target that contains the specifics.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D88696
While affine maps are part of the builtin memref type, there is very
limited support for manipulating them in the standard dialect. Add
transpose to the set of ops to complement the existing view/subview ops.
This is a metadata transformation that encodes the transpose into the
strides of a memref.
I'm planning to use this when lowering operations on strided memrefs,
using the transpose to remove the stride without adding a dependency on
linalg dialect.
Differential Revision: https://reviews.llvm.org/D88651
We hit an llvm_unreachable related to unranked memrefs for call ops
with scalar types. Removing the llvm_unreachable since the conversion
should gracefully bail out in the presence of unranked memrefs. Adding
tests to verify that.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D88709
This patch adds support for the 'return' and 'call' ops to the bare-ptr
calling convention. These changes also align the bare-ptr calling
convention code with the latest changes in the default calling convention
and reduce the amount of customization code needed.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D87724
- use select-ops to make the lowering simpler
- change style of FileCheck variables names to be consistent
- change some variable names in the code to be more explicit
Differential Revision: https://reviews.llvm.org/D88258
Recently, restrictions on vector reductions were made more relaxed by
accepting any width signless integer and floating-point. This CL relaxes
the restriction even more by including unsigned and signed integers.
Reviewed By: bkramer
Differential Revision: https://reviews.llvm.org/D88442
(1) simplify integer printing logic by always using 64-bit print
(2) add index support (since vector<16xindex> is planned to be added)
(3) adjust naming convention print_x -> printX
Reviewed By: bkramer
Differential Revision: https://reviews.llvm.org/D88436
This generalizes printing beyond just i1,i32,i64 and also accounts
for signed and unsigned interpretation in the output.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D88290
Allow propagating optional user defined attributes during SCF to GPU conversion. Gives opportunity to use user defined attributes in the further lowering. For example setting subgroup size, or other options for GPU dispatch. This does not break backward compatibility and does not require new attributes, just allow passing optional ones.
Differential Revision: https://reviews.llvm.org/D88203
This pass converts shape.cstr_* ops to eager (side-effecting)
error-handling code. After that conversion is done, the witnesses are
trivially satisfied and are replaced with `shape.const_witness true`.
Differential Revision: https://reviews.llvm.org/D87941
- Use TypeRange instead of ArrayRef<Type> where possible.
- Change some of the custom builders to also use TypeRange
Differential Revision: https://reviews.llvm.org/D87944
When packing function results into a structure during the standard-to-llvm
dialect conversion, do not assume the conversion was successful and propagate
nullptr as error state.
Fixes PR45184.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D87605
Type converter may fail and return nullptr on unconvertible types. The function
conversion did not include a check and was attempting to use a nullptr type to
construct an LLVM function, leading to a crash. Add a check and return early.
The rest of the call stack propagates errors properly.
Fixes PR47403.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D87075
This introduces a builder for the more general case that supports zero
elements (where the element type can't be inferred from the ValueRange,
since it might be empty).
Also, fix up some cases in ShapeToStandard lowering that hit this. It
happens very easily when dealing with shapes of 0-D tensors.
The SameOperandsAndResultElementType is redundant with the new
TypesMatchWith and prevented having zero elements.
Differential Revision: https://reviews.llvm.org/D87492
This patch adds a new named structured op to accompany linalg.matmul and
linalg.matvec. We needed it for our codegen, so I figured it would be useful
to add it to Linalg.
Reviewed By: nicolasvasilache, mravishankar
Differential Revision: https://reviews.llvm.org/D87292
Take advantage of the new `dynamic_tensor_from_elements` operation in `std`.
Instead of stack-allocated memory, we can now lower directly to a single `std`
operation.
Differential Revision: https://reviews.llvm.org/D86935
VectorToSCF.cpp:515:47: error: specialization of 'template<class TransferOpTy> mlir::LogicalResult mlir::VectorTransferRewriter<TransferOpTy>::matchAndRewrite(mlir::Operation*, mlir::PatternRewriter&) const' in different namespace [-fpermissive]
Put back anonymous namespace to work around GCC5 bug.
VectorToSCF.cpp:241:61: error: specialization of 'template<class ConcreteOp> mlir::LogicalResult {anonymous}::NDTransferOpHelper<ConcreteOp>::doReplace()' in different namespace [-fpermissive]
While there
- De-templatify code that can use function_ref
- Make BoundCaptures usable when they're const
- Address post-submit review comment (static function into global namespace)
This replaces the select chain for edge-padding with an scf.if that
performs the memory operation when the index is in bounds and uses the
pad value when it's not. For transfer_write the same mechanism is used,
skipping the store when the index is out of bounds.
The integration test has a bunch of cases of how I believe this should
work.
Differential Revision: https://reviews.llvm.org/D87241
With `dynamic_tensor_from_elements` tensor values of dynamic size can be
created. The body of the operation essentially maps the index space to tensor
elements.
Declare SCF operations in the `scf` namespace to avoid name clash with the new
`std.yield` operation. Resolve ambiguities between `linalg/shape/std/scf.yield`
operations.
Differential Revision: https://reviews.llvm.org/D86276
Vector to SCF conversion still had issues due to the interaction with the natural alignment derived by the LLVM data layout. One traditional workaround is to allocate aligned. However, this does not always work for vector sizes that are non-powers of 2.
This revision implements a more portable mechanism where the intermediate allocation is always a memref of elemental vector type. AllocOp is extended to use the natural LLVM DataLayout alignment for non-scalar types, when the alignment is not specified in the first place.
An integration test is added that exercises the transfer to scf.for + scalar lowering with a 5x5 transposition.
Differential Revision: https://reviews.llvm.org/D87150
When allowed, use 32-bit indices rather than 64-bit indices in the
SIMD computation of masks. This runs up to 2x and 4x faster on
a number of AVX2 and AVX512 microbenchmarks.
Reviewed By: bkramer
Differential Revision: https://reviews.llvm.org/D87116
Added 128 byte alignment to alloc ops created in VectorToSCF pass.
128b alignment was already introduced to this pass but not to all alloc
ops. This commit changes that by adding 128b alignment to the remaining ops.
The point of specifying alignment is to prevent possible memory alignment errors
on weakly tested architectures.
Differential Revision: https://reviews.llvm.org/D86454
Unsigned and Signless attributes use uintN_t and signed attributes use intN_t, where N is the fixed width. The 1-bit variants use bool.
Differential Revision: https://reviews.llvm.org/D86739
Adding a conversion pattern for the parallel Operation. This will
help the conversion of parallel operation with standard dialect to
parallel operation with llvm dialect. The type conversion of the block
arguments in a parallel region are controlled by the pattern for the
parallel Operation. Without this pattern, a parallel Operation with
block arguments cannot be converted from standard to LLVM dialect.
Other OpenMP operations without regions are marked as legal. When
translation of OpenMP operations with regions are added then patterns
for these operations can also be added.
Also uses all the standard to llvm patterns. Patterns of other dialects
can be added later if needed.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D86273
This patch allows to pass the gpu module name to SPIR-V
module during conversion. This has many benefits as we can lookup
converted to SPIR-V kernel in the symbol table.
In order to avoid symbol conflicts, `"__spv__"` is added to the
gpu module name to form the new one.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D86384
This patch introduces a hook to encode descriptor set
and binding number into `spv.globalVariable`'s symbolic name. This
allows to preserve this information, and at the same time legalize
the global variable for the conversion to LLVM dialect.
This is required for `mlir-spirv-cpu-runner` to convert kernel
arguments into LLVM.
Also, a couple of some nits added:
- removed unused comment
- changed to a capital letter in the comment
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D86515
Instead of using the TypeConverter infer the value of the alloca created based
on the init value. This will allow some ambiguous types like multidimensional
vectors to be converted correctly.
Differential Revision: https://reviews.llvm.org/D86582
Binding MemRefs of f16 needs special handling as the type is not supported on
CPU. There was a bug in the type used.
Differential Revision: https://reviews.llvm.org/D86328
Removed the Standard to LLVM conversion patterns that were previously
pulled in for testing purposes. This helps to separate the conversion
to LLVM dialect of the MLIR module with both SPIR-V and Standard
dialects in it (particularly helpful for SPIR-V cpu runner). Also,
tests were changed accordingly.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D86285
Add the unsigned complements to the existing FPToSI and SIToFP operations in the
standard dialect, with one-to-one lowerings to the corresponding LLVM operations.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D85557
If Memref has rank > 1 this pass emits N-1 loops around
TransferRead op and transforms the op itself to 1D read. Since vectors
must have static shape while memrefs don't the pass emits if condition
to prevent out of bounds accesses in case some memref dimension is smaller
than the corresponding dimension of targeted vector. This logic is fine
but authors forgot to apply `permutation_map` on loops upper bounds and
thus if condition compares induction variable to incorrect loop upper bound
(dimension of the memref) in case `permutation_map` is not identity map.
This commit aims to fix that.
This changes the behavior of constructing MLIRContext to no longer load globally
registered dialects on construction. Instead Dialects are only loaded explicitly
on demand:
- the Parser is lazily loading Dialects in the context as it encounters them
during parsing. This is the only purpose for registering dialects and not load
them in the context.
- Passes are expected to declare the dialects they will create entity from
(Operations, Attributes, or Types), and the PassManager is loading Dialects into
the Context when starting a pipeline.
This changes simplifies the configuration of the registration: a compiler only
need to load the dialect for the IR it will emit, and the optimizer is
self-contained and load the required Dialects. For example in the Toy tutorial,
the compiler only needs to load the Toy dialect in the Context, all the others
(linalg, affine, std, LLVM, ...) are automatically loaded depending on the
optimization pipeline enabled.
To adjust to this change, stop using the existing dialect registration: the
global registry will be removed soon.
1) For passes, you need to override the method:
virtual void getDependentDialects(DialectRegistry ®istry) const {}
and registery on the provided registry any dialect that this pass can produce.
Passes defined in TableGen can provide this list in the dependentDialects list
field.
2) For dialects, on construction you can register dependent dialects using the
provided MLIRContext: `context.getOrLoadDialect<DialectName>()`
This is useful if a dialect may canonicalize or have interfaces involving
another dialect.
3) For loading IR, dialect that can be in the input file must be explicitly
registered with the context. `MlirOptMain()` is taking an explicit registry for
this purpose. See how the standalone-opt.cpp example is setup:
mlir::DialectRegistry registry;
registry.insert<mlir::standalone::StandaloneDialect>();
registry.insert<mlir::StandardOpsDialect>();
Only operations from these two dialects can be in the input file. To include all
of the dialects in MLIR Core, you can populate the registry this way:
mlir::registerAllDialects(registry);
4) For `mlir-translate` callback, as well as frontend, Dialects can be loaded in
the context before emitting the IR: context.getOrLoadDialect<ToyDialect>()
Differential Revision: https://reviews.llvm.org/D85622
This changes the behavior of constructing MLIRContext to no longer load globally
registered dialects on construction. Instead Dialects are only loaded explicitly
on demand:
- the Parser is lazily loading Dialects in the context as it encounters them
during parsing. This is the only purpose for registering dialects and not load
them in the context.
- Passes are expected to declare the dialects they will create entity from
(Operations, Attributes, or Types), and the PassManager is loading Dialects into
the Context when starting a pipeline.
This changes simplifies the configuration of the registration: a compiler only
need to load the dialect for the IR it will emit, and the optimizer is
self-contained and load the required Dialects. For example in the Toy tutorial,
the compiler only needs to load the Toy dialect in the Context, all the others
(linalg, affine, std, LLVM, ...) are automatically loaded depending on the
optimization pipeline enabled.
To adjust to this change, stop using the existing dialect registration: the
global registry will be removed soon.
1) For passes, you need to override the method:
virtual void getDependentDialects(DialectRegistry ®istry) const {}
and registery on the provided registry any dialect that this pass can produce.
Passes defined in TableGen can provide this list in the dependentDialects list
field.
2) For dialects, on construction you can register dependent dialects using the
provided MLIRContext: `context.getOrLoadDialect<DialectName>()`
This is useful if a dialect may canonicalize or have interfaces involving
another dialect.
3) For loading IR, dialect that can be in the input file must be explicitly
registered with the context. `MlirOptMain()` is taking an explicit registry for
this purpose. See how the standalone-opt.cpp example is setup:
mlir::DialectRegistry registry;
registry.insert<mlir::standalone::StandaloneDialect>();
registry.insert<mlir::StandardOpsDialect>();
Only operations from these two dialects can be in the input file. To include all
of the dialects in MLIR Core, you can populate the registry this way:
mlir::registerAllDialects(registry);
4) For `mlir-translate` callback, as well as frontend, Dialects can be loaded in
the context before emitting the IR: context.getOrLoadDialect<ToyDialect>()
Differential Revision: https://reviews.llvm.org/D85622
This changes the behavior of constructing MLIRContext to no longer load globally
registered dialects on construction. Instead Dialects are only loaded explicitly
on demand:
- the Parser is lazily loading Dialects in the context as it encounters them
during parsing. This is the only purpose for registering dialects and not load
them in the context.
- Passes are expected to declare the dialects they will create entity from
(Operations, Attributes, or Types), and the PassManager is loading Dialects into
the Context when starting a pipeline.
This changes simplifies the configuration of the registration: a compiler only
need to load the dialect for the IR it will emit, and the optimizer is
self-contained and load the required Dialects. For example in the Toy tutorial,
the compiler only needs to load the Toy dialect in the Context, all the others
(linalg, affine, std, LLVM, ...) are automatically loaded depending on the
optimization pipeline enabled.
To adjust to this change, stop using the existing dialect registration: the
global registry will be removed soon.
1) For passes, you need to override the method:
virtual void getDependentDialects(DialectRegistry ®istry) const {}
and registery on the provided registry any dialect that this pass can produce.
Passes defined in TableGen can provide this list in the dependentDialects list
field.
2) For dialects, on construction you can register dependent dialects using the
provided MLIRContext: `context.getOrLoadDialect<DialectName>()`
This is useful if a dialect may canonicalize or have interfaces involving
another dialect.
3) For loading IR, dialect that can be in the input file must be explicitly
registered with the context. `MlirOptMain()` is taking an explicit registry for
this purpose. See how the standalone-opt.cpp example is setup:
mlir::DialectRegistry registry;
mlir::registerDialect<mlir::standalone::StandaloneDialect>();
mlir::registerDialect<mlir::StandardOpsDialect>();
Only operations from these two dialects can be in the input file. To include all
of the dialects in MLIR Core, you can populate the registry this way:
mlir::registerAllDialects(registry);
4) For `mlir-translate` callback, as well as frontend, Dialects can be loaded in
the context before emitting the IR: context.getOrLoadDialect<ToyDialect>()
There should be an equivalent std.floor op to std.ceil. This includes
matching lowerings for SPIRV, NVVM, ROCDL, and LLVM.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D85940
This patch adds more op/type conversion support
necessary for `spirv-runner`:
- EntryPoint/ExecutionMode: currently removed since we assume
having only one kernel function in the kernel module.
- StorageBuffer storage class is now supported. We are not
concerned with multithreading so this is fine for now.
- Type conversion enhanced, now regular offsets and strides
for structs and arrays are supported (based on
`VulkanLayoutUtils`).
- Support of `spc.AccessChain` that is modelled with GEP op
in LLVM dialect.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D86109
The function makes too strong assumption regarding parent FuncOp
which gets broken when FuncOp is first lowered to llvm function.
In this fix we generalize the assumption to allocation scope and
add assertion to produce user friendly message in case our assumption
is broken.
Differential Revision: https://reviews.llvm.org/D86086
Legacy implementation of the LLVM dialect in MLIR contained an instance of
llvm::Module as it was required to parse LLVM IR types. The access to the data
layout of this module was exposed to the users for convenience, but in practice
this layout has always been the default one obtained by parsing an empty layout
description string. Current implementation of the dialect no longer relies on
wrapping LLVM IR types, but it kept an instance of DataLayout for
compatibility. This effectively forces a single data layout to be used across
all modules in a given MLIR context, which is not desirable. Remove DataLayout
from the LLVM dialect and attach it as a module attribute instead. Since MLIR
does not yet have support for data layouts, use the LLVM DataLayout in string
form with verification inside MLIR. Introduce the layout when converting a
module to the LLVM dialect and keep the default "" description for
compatibility.
This approach should be replaced with a proper MLIR-based data layout when it
becomes available, but provides an immediate solution to compiling modules with
different layouts, e.g. for GPUs.
This removes the need for LLVMDialectImpl, which is also removed.
Depends On D85650
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D85652
This changes the behavior of constructing MLIRContext to no longer load globally registered dialects on construction. Instead Dialects are only loaded explicitly on demand:
- the Parser is lazily loading Dialects in the context as it encounters them during parsing. This is the only purpose for registering dialects and not load them in the context.
- Passes are expected to declare the dialects they will create entity from (Operations, Attributes, or Types), and the PassManager is loading Dialects into the Context when starting a pipeline.
This changes simplifies the configuration of the registration: a compiler only need to load the dialect for the IR it will emit, and the optimizer is self-contained and load the required Dialects. For example in the Toy tutorial, the compiler only needs to load the Toy dialect in the Context, all the others (linalg, affine, std, LLVM, ...) are automatically loaded depending on the optimization pipeline enabled.
Differential Revision: https://reviews.llvm.org/D85622
This changes the behavior of constructing MLIRContext to no longer load globally registered dialects on construction. Instead Dialects are only loaded explicitly on demand:
- the Parser is lazily loading Dialects in the context as it encounters them during parsing. This is the only purpose for registering dialects and not load them in the context.
- Passes are expected to declare the dialects they will create entity from (Operations, Attributes, or Types), and the PassManager is loading Dialects into the Context when starting a pipeline.
This changes simplifies the configuration of the registration: a compiler only need to load the dialect for the IR it will emit, and the optimizer is self-contained and load the required Dialects. For example in the Toy tutorial, the compiler only needs to load the Toy dialect in the Context, all the others (linalg, affine, std, LLVM, ...) are automatically loaded depending on the optimization pipeline enabled.
The convresion of memref cast operaitons from the Standard dialect to the LLVM
dialect has been emitting bitcasts from a struct type to itself. Beyond being
useless, such casts are invalid as bitcast does not operate on aggregate types.
This kept working by accident because LLVM IR bitcast construction API skips
the construction if types are equal before it verifies that the types are
acceptable in a bitcast. Do not emit such bitcasts, the memref cast that only
adds/erases size information is in fact a noop on the current descriptor as it
always contains dynamic values for all sizes.
Reviewed By: pifon2a
Differential Revision: https://reviews.llvm.org/D85899
This revision removes all of the lingering usages of Type::getKind. A consequence of this is that FloatType is now split into 4 derived types that represent each of the possible float types(BFloat16Type, Float16Type, Float32Type, and Float64Type). Other than this split, this revision is NFC.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D85566
Inital conversion of `spv._address_of` and `spv.globalVariable`.
In SPIR-V, the global returns a pointer, whereas in LLVM dialect
the global holds an actual value. This difference is handled by
`spv._address_of` and `llvm.mlir.addressof`ops that both return
a pointer. Moreover, only current invocation is in conversion's
scope.
Reviewed By: antiagainst, mravishankar
Differential Revision: https://reviews.llvm.org/D84626
This change adds initial support needed to generate OpenCL compliant SPIRV.
If Kernel capability is declared then memory model becomes OpenCL.
If Addresses capability is declared then addressing model becomes Physical64.
Additionally for Kernel capability interface variable ABI attributes are not
generated as entry point function is expected to have normal arguments.
Differential Revision: https://reviews.llvm.org/D85196
Using a shuffle for the last recursive step in progressive lowering not only
results in much more compact IR, but also more efficient code (since the
backend is no longer confused on subvector aliasing for longer vectors).
E.g. the following
%f = vector.shape_cast %v0: vector<1024xf32> to vector<32x32xf32>
yields much better x86-64 code that runs 3x faster than the original.
Reviewed By: bkramer, nicolasvasilache
Differential Revision: https://reviews.llvm.org/D85482
Original modeling of LLVM IR types in the MLIR LLVM dialect had been wrapping
LLVM IR types and therefore required the LLVMContext in which they were created
to outlive them, which was solved by placing the LLVMContext inside the dialect
and thus having the lifetime of MLIRContext. This has led to numerous issues
caused by the lack of thread-safety of LLVMContext and the need to re-create
LLVM IR modules, obtained by translating from MLIR, in different LLVM contexts
to enable parallel compilation. Similarly, llvm::Module had been introduced to
keep track of identified structure types that could not be modeled properly.
A recent series of commits changed the modeling of LLVM IR types in the MLIR
LLVM dialect so that it no longer wraps LLVM IR types and has no dependence on
LLVMContext and changed the ownership model of the translated LLVM IR modules.
Remove LLVMContext and LLVM modules from the implementation of MLIR LLVM
dialect and clean up the remaining uses.
The only part of LLVM IR that remains necessary for the LLVM dialect is the
data layout. It should be moved from the dialect level to the module level and
replaced with an MLIR-based representation to remove the dependency of the
LLVMDialect on LLVM IR library.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D85445
Due to the original type system implementation, LLVMDialect in MLIR contains an
LLVMContext in which the relevant objects (types, metadata) are created. When
an MLIR module using the LLVM dialect (and related intrinsic-based dialects
NVVM, ROCDL, AVX512) is converted to LLVM IR, it could only live in the
LLVMContext owned by the dialect. The type system no longer relies on the
LLVMContext, so this limitation can be removed. Instead, translation functions
now take a reference to an LLVMContext in which the LLVM IR module should be
constructed. The caller of the translation functions is responsible for
ensuring the same LLVMContext is not used concurrently as the translation no
longer uses a dialect-wide context lock.
As an additional bonus, this change removes the need to recreate the LLVM IR
module in a different LLVMContext through printing and parsing back, decreasing
the compilation overhead in JIT and GPU-kernel-to-blob passes.
Reviewed By: rriddle, mehdi_amini
Differential Revision: https://reviews.llvm.org/D85443
The RewritePattern will become one of several, and will be part of the LLVM conversion pass (instead of a separate pass following LLVM conversion).
Reviewed By: herhut
Differential Revision: https://reviews.llvm.org/D84946
Historical modeling of the LLVM dialect types had been wrapping LLVM IR types
and therefore needed access to the instance of LLVMContext stored in the
LLVMDialect. The new modeling does not rely on that and only needs the
MLIRContext that is used for uniquing, similarly to other MLIR types. Change
LLVMType::get<Kind>Ty functions to take `MLIRContext *` instead of
`LLVMDialect *` as first argument. This brings the code base closer to
completely removing the dependence on LLVMContext from the LLVMDialect,
together with additional support for thread-safety of its use.
Depends On D85371
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D85372
This prepares for the removal of llvm::Module and LLVMContext from the
mlir::LLVMDialect.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D85371
The intrinsics were already supported and vector.transfer_read/write lowered
direclty into these operations. By providing them as individual ops, however,
clients can used them directly, and it opens up progressively lowering transfer
operations at higher levels (rather than direct lowering to LLVM IR as done now).
Reviewed By: bkramer
Differential Revision: https://reviews.llvm.org/D85357
Previous type model in the LLVM dialect did not support identified structure
types properly and therefore could use stateless translations implemented as
free functions. The new model supports identified structs and must keep track
of the identified structure types present in the target context (LLVMContext or
MLIRContext) to avoid creating duplicate structs due to LLVM's type
auto-renaming. Expose the stateful type translation classes and use them during
translation, storing the state as part of ModuleTranslation.
Drop the test type translation mechanism that is no longer necessary and update
the tests to exercise type translation as part of the main translation flow.
Update the code in vector-to-LLVM dialect conversion that relied on stateless
translation to use the new class in a stateless manner.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D85297
Per Vulkan's SPIR-V environment spec: "While the OpSRem and OpSMod
instructions are supported by the Vulkan environment, they require
non-negative values and thus do not enable additional functionality
beyond what OpUMod provides."
The `getOffsetForBitwidth` function is used for lowering std.load
and std.store, whose indices are of `index` type and cannot be
negative. So we should be okay to use spv.UMod directly here to
be exact. Also made the comment explicit about the assumption.
Differential Revision: https://reviews.llvm.org/D83714
`promoteMemRefDescriptors` also converts types of every operand, not only
memref-typed ones. I think `promoteMemRefDescriptors` name does not imply that.
Differential Revision: https://reviews.llvm.org/D85325
Handle the case where the ViewOp takes in a memref that has
an memory space.
Reviewed By: ftynse, bondhugula, nicolasvasilache
Differential Revision: https://reviews.llvm.org/D85048
This patch introduces a conversion of `spv.loop` to LLVM dialect.
Similarly to `spv.selection`, op's control attributes are not mapped
to LLVM yet and therefore the conversion fails if the loop control is
not `None`. Also, all blocks within the loop should be reachable in
order for conversion to succeed.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D84245
Introduces the expand and compress operations to the Vector dialect
(important memory operations for sparse computations), together
with a first reference implementation that lowers to the LLVM IR
dialect to enable running on CPU (and other targets that support
the corresponding LLVM IR intrinsics).
Reviewed By: reidtatge
Differential Revision: https://reviews.llvm.org/D84888
A new first-party modeling for LLVM IR types in the LLVM dialect has been
developed in parallel to the existing modeling based on wrapping LLVM `Type *`
instances. It resolves the long-standing problem of modeling identified
structure types, including recursive structures, and enables future removal of
LLVMContext and related locking mechanisms from LLVMDialect.
This commit only switches the modeling by (a) renaming LLVMTypeNew to LLVMType,
(b) removing the old implementaiton of LLVMType, and (c) updating the tests. It
is intentionally minimal. Separate commits will remove the infrastructure built
for the transition and update API uses where appropriate.
Depends On D85020
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D85021
The bug was not noticed because we didn't have a lot of custom type conversions
directly to LLVM dialect.
Differential Revision: https://reviews.llvm.org/D85192
Lowering of newly defined Conv ops in TC syntax to standard
dialect is not supported and therefore this commit adds support
for it.
Differential Revision: https://reviews.llvm.org/D84840
Replaced definition of named ND ConvOps with tensor comprehension
syntax which reduces boilerplate code significantly. Furthermore,
new ops to support TF convolutions added (without strides and dilations).
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D84628
Now that we can have a memref of index type, we no longer need to materialize shapes in i64 and then index_cast.
Differential Revision: https://reviews.llvm.org/D84938
When lowering to the standard dialect, we currently support only the extent
tensor variant of the shape.rank operation. This change lets the conversion
pattern fail in a well-defined manner.
Differential Revision: https://reviews.llvm.org/D84852
This patch introduces new intrinsics in LLVM dialect:
- `llvm.intr.floor`
- `llvm.intr.maxnum`
- `llvm.intr.minnum`
- `llvm.intr.smax`
- `llvm.intr.smin`
These intrinsics correspond to SPIR-V ops from GLSL
extended instruction set (`spv.GLSL.Floor`, `spv.GLSL.FMax`,
`spv.GLSL.FMin`, `spv.GLSL.SMax` and `spv.GLSL.SMin`
respectively). Also conversion patterns for them were added.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D84661
This is a second patch on conversion of GLSL ops to LLVM dialect.
It introduces patterns to convert `spv.InverseSqrt` and `spv.Tanh`.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D84633
This is the first patch that adds support for GLSL extended
instruction set ops. These are direct conversions, apart from `spv.Tan`
that is lowered to `sin() / cos()`.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D84627
This commit is part of a greater project which aims to add
full end-to-end support for convolutions inside mlir. The
reason behind having conv ops for each rank rather than
having one generic ConvOp is to enable better optimizations
for every N-D case which reflects memory layout of input/kernel
buffers better and simplifies code as well. We expect plain linalg.conv
to be progressively retired.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D83879
The lowering does not support all types for its source operations. This change
makes the patterns fail in a well-defined manner.
Differential Revision: https://reviews.llvm.org/D84443
The current modeling of LLVM IR types in MLIR is based on the LLVMType class
that wraps a raw `llvm::Type *` and delegates uniquing, printing and parsing to
LLVM itself. This is model makes thread-safe type manipulation hard and is
being progressively replaced with a cleaner MLIR model that replicates the type
system. In the new model, LLVMType will no longer have an underlying LLVM IR
type. Restrict access to this type in the current model in preparation for the
change.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D84389
Operating on indices and extent tensors directly, the type conversion is no
longer needed for the supported cases.
Differential Revision: https://reviews.llvm.org/D84442
This adds conversions for const_size and to_extent_tensor. Also, cast-like operations are now folded away if the source and target types are the same.
Differential Revision: https://reviews.llvm.org/D84745
Conversion of `spv.BranchConditional` now supports branch weights
that are mapped to weights vector in `llvm.cond_br`.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D84657
This patch adds support of Volatile and Nontemporal
memory accesses to `spv.Load` and `spv.Store`. These attributes are
modelled with a `volatile` and `nontemporal` flags.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D84739
Do not return error code, instead return created resource handles or void. Error reporting is done by the library function.
Reviewed By: herhut
Differential Revision: https://reviews.llvm.org/D84660
- replace DotOp, now that DRR rules have been dropped.
- Capture arguments mismatch in the parser. The number of parsed arguments must
equal the number of expected arguments.
Reviewed By: ftynse, nicolasvasilache
Differential Revision: https://reviews.llvm.org/D82952
The LowerAffine psas was a FunctionPass only for legacy
reasons. Making this Op-agnostic allows it to be used from command
line when affine expressions are within operations other than
`std.func`.
Differential Revision: https://reviews.llvm.org/D84590
This patch introduces conversion pattern for `spv.Store` and `spv.Load`.
Only op with `Function` Storage Class is supported at the moment
because `spv.GlobalVariable` has not been introduced yet. If the op
has memory access attribute, then there are the following cases.
If the access is `Aligned`, add alignment to the op builder. Otherwise
the conversion fails as other cases are not supported yet because they
require additional attributes for `llvm.store`/`llvm.load` ops: e.g.
`volatile` and `nontemporal`.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D84236
The patch introduces the conversion pattern for function-level
`spv.Variable`. It is modelled as `llvm.alloca` op. If initialized, then
additional store instruction is used. Note that there is no initialization
for arrays and structs since constants of these types are not supported in
LLVM dialect yet. Also, at the moment initialisation is only possible via
`spv.constant` (since `spv.GlobalVariable` conversion is not implemented
yet).
The input code has some scoping is not taken into account and will be
addressed in a different patch.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D84224
This concerns `from/to_extent_tensor`, `size_to_index`, `index_to_size`, and
`const_size` conversion patterns. The new lowering will work directly on indices
and extent tensors. The shape and size values will allow for error values but
are not yet supported by the dialect conversion.
Differential Revision: https://reviews.llvm.org/D84436
The operation `shape.shape_of` now returns an extent tensor `tensor<?xindex>` in
cases when no error are possible. All consuming operation will eventually accept
both, shapes and extent tensors.
Differential Revision: https://reviews.llvm.org/D84160
The default lowering of `assert` calls `abort` in case the assertion is
violated. The failure message is ignored but should be used by custom lowerings
that can assume more about their environment.
Differential Revision: https://reviews.llvm.org/D83886
This revision adds support for much deeper type conversion integration into the conversion process, and enables auto-generating cast operations when necessary. Type conversions are now largely automatically managed by the conversion infra when using a ConversionPattern with a provided TypeConverter. This removes the need for patterns to do type cast wrapping themselves and moves the burden to the infra. This makes it much easier to perform partial lowerings when type conversions are involved, as any lingering type conversions will be automatically resolved/legalized by the conversion infra.
To support this new integration, a few changes have been made to the type materialization API on TypeConverter. Materialization has been split into three separate categories:
* Argument Materialization: This type of materialization is used when converting the type of block arguments when calling `convertRegionTypes`. This is useful for contextually inserting additional conversion operations when converting a block argument type, such as when converting the types of a function signature.
* Source Materialization: This type of materialization is used to convert a legal type of the converter into a non-legal type, generally a source type. This may be called when uses of a non-legal type persist after the conversion process has finished.
* Target Materialization: This type of materialization is used to convert a non-legal, or source, type into a legal, or target, type. This type of materialization is used when applying a pattern on an operation, but the types of the operands have not yet been converted.
Differential Revision: https://reviews.llvm.org/D82831
Introduces the scatter/gather operations to the Vector dialect
(important memory operations for sparse computations), together
with a first reference implementation that lowers to the LLVM IR
dialect to enable running on CPU (and other targets that support
the corresponding LLVM IR intrinsics).
The operations can be used directly where applicable, or can be used
during progressively lowering to bring other memory operations closer to
hardware ISA support for a gather/scatter. The semantics of the operation
closely correspond to those of the corresponding llvm intrinsics.
Note that the operation allows for a dynamic index vector (which is
important for sparse computations). However, this first reference
lowering implementation "serializes" the address computation when
base + index_vector is converted to a vector of pointers. Exploring
how to use SIMD properly during these step is TBD. More general
memrefs and idiomatic versions of striding are also TBD.
Reviewed By: arpith-jacob
Differential Revision: https://reviews.llvm.org/D84039
Christian Sigg