This was seemingly dropped in e2310704d8,
potentially due to a misrebase. The absence of this trait makes aliasing
analysis incorrect, leading to, e.g., buffer deallocation pass inserting
deallocations too early.
The patch in question broke the build with shared libraries due to
missing dependencies, one of which would have been circular between
MLIRStandard and MLIRMemRef if added. Fix this by moving more code
around and swapping the dependency direction. MLIRMemRef now depends on
MLIRStandard, but MLIRStandard does _not_ depend on MLIRMemRef.
Arguably, this is the right direction anyway since numerous libraries
depend on MLIRStandard and don't necessarily need to depend on
MLIRMemref.
Other otable changes include:
- some EDSC code is moved inline to MemRef/EDSC/Intrinsics.h because it
creates MemRef dialect operations;
- a utility function related to shape moved to BuiltinTypes.h/cpp
because it only realtes to shaped types and not any particular dialect
(standard dialect is erroneously believed to contain MemRefType);
- a Python test for the standard dialect is disabled completely because
the ops it tests moved to the new MemRef dialect, but it is not
exposed to Python bindings, and the change for that is non-trivial.
This reverts commit b5d9a3c923.
The commit introduced a memory error in canonicalization/operation
walking that is exposed when compiled with ASAN. It leads to crashes in
some "release" configurations.
We know that all ConstantLike operations have one result and no operands,
so check this first before doing the trait check. This change speeds up
Canonicalize on a CIRCT testcase by ~5%.
Differential Revision: https://reviews.llvm.org/D98615
Two changes:
1) Change the canonicalizer to walk the function in top-down order instead of
bottom-up order. This composes well with the "top down" nature of constant
folding and simplification, reducing iterations and re-evaluation of ops in
simple cases.
2) Explicitly enter existing constants into the OperationFolder table before
canonicalizing. Previously we would "constant fold" them and rematerialize
them, wastefully recreating a bunch fo constants, which lead to pointless
memory traffic.
Both changes together provide a 33% speedup for canonicalize on some mid-size
CIRCT examples.
One artifact of this change is that the constants generated in normal pattern
application get inserted at the top of the function as the patterns are applied.
Because of this, we get "inverted" constants more often, which is an aethetic
change to the IR but does permute some testcases.
Differential Revision: https://reviews.llvm.org/D98609
This removes some (but not all) uses of type-less CreateGEP()
and CreateInBoundsGEP() APIs, which are incompatible with opaque
pointers.
There are a still a number of tricky uses left, as well as many
more variation APIs for CreateGEP.
NestedPattern uses a BumpPtrAllocator to store child (nested) pattern
objects to decrease the overhead of dynamic allocation. This assumes all
allocations happen inside the allocator that will be freed as a whole.
However, NestedPattern contains `std::function` as a member, which
allocates internally using `new`, unaware of the BumpPtrAllocator. Since
NestedPattern only holds pointers to the nested patterns allocated in
the BumpPtrAllocator, it never calls their destructors, so the
destructor of the `std::function`s they contain are never called either,
leaking the allocated memory.
Make NestedPattern explicitly call destructors of nested patterns. This
additionally requires to actually copy the nested patterns in
copy-construction and copy-assignment instead of just sharing the
pointer to the arena-allocated list of children to avoid double-free. An
alternative solution would be to add reference counting to the list of
arena-allocated list of children.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D98485
This patch introduces progressive lowering patterns for rewriting
vector.transfer_read/write to vector.load/store and vector.broadcast
in certain supported cases.
Reviewed By: dcaballe, nicolasvasilache
Differential Revision: https://reviews.llvm.org/D97822
This patch replaces the root-terminal vectorization approach implemented in the
Affine vectorizer with a topological order approach that vectorizes all the
operations within the target loop nest. These are the most important changes
introduced by the new algorithm:
* Removed tracking of root and terminal ops. Existing vectorization
functionality is preserved and extended so that loop nests without
root-terminal chains can be vectorized.
* Vectorizing a loop nest now only requires a single topological traversal.
* A new vector loop nest is incrementally built along the vectorization
process. The original scalar loop is kept intact. No cloning guard is needed
to recover the scalar loop if vectorization fails. This approach also
simplifies the challenging task of replacing a loop operation amid the
vectorization process without invalidating the analysis information that
depends on the original loop.
* Vectorization of specific operations has been implemented as independent,
preparing them to be moved to a potential vectorization interface.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D97442
This allows for storage instances to store data that isn't uniqued in the context, or contain otherwise non-trivial logic, in the rare situations that they occur. Storage instances with trivial destructors will still have their destructor skipped. A consequence of this is that the storage instance definition must be visible from the place that registers the type.
Differential Revision: https://reviews.llvm.org/D98311
This patch fixes a heap-use-after-free introduced by the recent changes
in the vectorizer: https://reviews.llvm.org/rG95db7b4aeaad590f37720898e339a6d54313422f
The problem is due to the way candidate loops are visited. All candidate loops
are pattern-matched beforehand using the 'NestedMatch' utility. These matches may
intersect with each other so it may happen that we try to vectorize a loop that
was previously vectorized. The new vectorization algorithm replaces the original
loops that are vectorized with new loops and, therefore, any reference to the
original loops in the pre-computed matches becomes invalid.
This patch fixes the problem by classifying the candidate matches into buckets
before vectorization. Each bucket contains all the matches that intersect. The
vectorizer uses these buckets to make sure that we only vectorize *one* match from
each bucket, at most.
Differential Revision: https://reviews.llvm.org/D98382
For the use in LLVMOps.td I used the getPointerElementType()
escape hatch, as it's not obvious to me how the load type
should be properly obtained here.
Data layout information allows to answer questions about the size and alignment
properties of a type. It enables, among others, the generation of various
linear memory addressing schemes for containers of abstract types and deeper
reasoning about vectors. This introduces the subsystem for modeling data
layouts in MLIR.
The data layout subsystem is designed to scale to MLIR's open type and
operation system. At the top level, it consists of attribute interfaces that
can be implemented by concrete data layout specifications; type interfaces that
should be implemented by types subject to data layout; operation interfaces
that must be implemented by operations that can serve as data layout scopes
(e.g., modules); and dialect interfaces for data layout properties unrelated to
specific types. Built-in types are handled specially to decrease the overall
query cost.
A concrete default implementation of these interfaces is provided in the new
Target dialect. Defaults for built-in types that match the current behavior are
also provided.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D97067
verifyCompatibleShapes is not transitive. Create an n-ary version and
update SameOperandShapes and SameOperandAndResultShapes traits to use
it.
Differential Revision: https://reviews.llvm.org/D98331
If MLIR_CUDA_RUNNER_ENABLED, register a 'gpu-to-cubin' conversion pass to mlir-opt.
The next step is to switch CUDA integration tests from mlir-cuda-runner to mlir-opt + mlir-cpu-runner and remove mlir-cuda-runner.
Depends On D98279
Reviewed By: herhut, rriddle, mehdi_amini
Differential Revision: https://reviews.llvm.org/D98203
The current implementation has some inefficiencies that become noticeable when running on large modules. This revision optimizes the code, and updates some out-dated idioms with newer utilities. The main components of this optimization include:
* Add an overload of Block::eraseArguments that allows for O(N) erasure of disjoint arguments.
* Don't process entry block arguments given that we don't erase them at this point.
* Don't track individual operation results, given that we don't erase them. We can just track the parent operation.
Differential Revision: https://reviews.llvm.org/D98309
This patch replaces the root-terminal vectorization approach implemented in the
Affine vectorizer with a topological order approach that vectorizes all the
operations within the target loop nest. These are the most important changes
introduced by the new algorithm:
* Removed tracking of root and terminal ops. Existing vectorization
functionality is preserved and extended so that loop nests without
root-terminal chains can be vectorized.
* Vectorizing a loop nest now only requires a single topological traversal.
* A new vector loop nest is incrementally built along the vectorization
process. The original scalar loop is kept intact. No cloning guard is needed
to recover the scalar loop if vectorization fails. This approach also
simplifies the challenging task of replacing a loop operation amid the
vectorization process without invalidating the analysis information that
depends on the original loop.
* Vectorization of specific operations has been implemented as independent,
preparing them to be moved to a potential vectorization interface.
Reviewed By: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D97442
The dialect separation was introduced to demarkate ops operating in different
type systems. This is no longer the case after the LLVM dialect has migrated to
using built-in vector types, so the original reason for separation is no longer
valid. Squash the two dialects into one.
The code size decrease isn't quite large: the ops originally in LLVM_AVX512 are
preserved because they match LLVM IR intrinsics specialized for vector element
bitwidth. However, it is still conceptually beneficial to have only one
dialect. I originally considered to use Tablegen multiclasses to define both
the type-polymorphic op and its two intrinsic-related instantiations, but
decided against it given both the complexity of the required Tablegen input and
its dissimilarity with the rest of ODS-defined ops, both potentially resulting
in very poor maintainability.
Depends On D98327
Reviewed By: nicolasvasilache, springerm
Differential Revision: https://reviews.llvm.org/D98328
VectorOfLengthAndType accepts a cartesian product of given lengths and types
rather than types produced by co-indexed values in the corresponding lists.
Update the definitions accordingly. The type validity is already enforced by
op traits.
Reviewed By: nicolasvasilache, springerm
Differential Revision: https://reviews.llvm.org/D98327
It is to use the methods in LinalgInterfaces.cpp for additional static shape verification to match the shaped operands and loop on linalgOps. If I used the existing methods, I would face circular dependency linking issue. Now we can use them as methods of LinalgOp.
Reviewed By: hanchung
Differential Revision: https://reviews.llvm.org/D98163
Instead of configuring kernel-to-cubin/rocdl lowering through callbacks, introduce a base class that target-specific passes can derive from.
Put the base class in GPU/Transforms, according to the discussion in D98203.
The mlir-cuda-runner will go away shortly, and the mlir-rocdl-runner as well at some point. I therefore kept the existing code path working and will remove it in a separate step.
Depends On D98168
Reviewed By: herhut
Differential Revision: https://reviews.llvm.org/D98279
Based on the following discussion:
https://llvm.discourse.group/t/rfc-memref-memory-shape-as-attribute/2229
The goal of the change is to make memory space property to have more
expressive representation, rather then "magic" integer values.
It will allow to have more clean ASM form:
```
gpu.func @test(%arg0: memref<100xf32, "workgroup">)
// instead of
gpu.func @test(%arg0: memref<100xf32, 3>)
```
Explanation for `Attribute` choice instead of plain `string`:
* `Attribute` classes allow to use more type safe API based on RTTI.
* `Attribute` classes provides faster comparison operator based on
pointer comparison in contrast to generic string comparison.
* `Attribute` allows to store more complex things, like structs or dictionaries.
It will allows to have more complex memory space hierarchy.
This commit preserve old integer-based API and implements it on top
of the new one.
Depends on D97476
Reviewed By: rriddle, mehdi_amini
Differential Revision: https://reviews.llvm.org/D96145
This method allows for removing multiple disjoint operands at once, reducing the need to erase operands individually (which results in shifting the operand list).
Differential Revision: https://reviews.llvm.org/D98290
This class provides efficient implementations of symbol queries related to uses, such as collecting the users of a symbol, replacing all uses, etc. This provides similar benefits to use related queries, as SymbolTableCollection did for lookup queries.
Differential Revision: https://reviews.llvm.org/D98071
Provide default for gpuBinaryAnnotation so that we don't need to specify it in tests.
The annotation likely only needs to be target specific if we want to lower to e.g. both CUDA and ROCDL.
Reviewed By: herhut, bondhugula
Differential Revision: https://reviews.llvm.org/D98168
Instead of storing an array of LoopOpt attributes, which were just
wrapping std::pair<enum, int> anyway, we can have an attribute storing
a sorted ArrayRef<std::pair<enum, int>> as a single unit. This improves
here the textual format and the general API. Note that we're limiting
the options to fit into an int64_t by design, but this isn't a new
constraint.
Building the LoopOptions attribute is likely worth a specific builder
for efficient reason, that'll be the subject of a future patch.
Differential Revision: https://reviews.llvm.org/D98105
Move Target/LLVMIR.h to target/LLVMIR/Import.h to better reflect the purpose of
this file. Also move all LLVM IR target tests under the LLVMIR directory.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D98178
Return the vectorization results using a vector passed by reference instead of returning them embedded in a structure.
Differential Revision: https://reviews.llvm.org/D98182
This is using the new Attribute storage generation support in
TableGen to define the LLVM FastMathFlags.
Differential Revision: https://reviews.llvm.org/D98007
This will allow for removing the duplicated type documentation from LangRef and instead link to the builtin dialect documentation.
Differential Revision: https://reviews.llvm.org/D98093
split_at can return an error if the split index is out of bounds. If the
user knows that the index can never be out of bounds it's safe to use
extent tensors. This has a straight-forward lowering to std.subtensor.
Differential Revision: https://reviews.llvm.org/D98177
To unify the naming scheme across all ops in the SPIR-V dialect, we are
moving from spv.camelCase to spv.CamelCase everywhere. For ops that
don't have a SPIR-V spec counterpart, we use spv.mlir.snake_case.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D98014
Normally tensors will be stored in buffers before converting to SPIR-V,
given that is how a large amount of data is sent to the GPU. However,
SPIR-V supports converting from tensors directly too. This is for the
cases where the tensor just contains a small amount of elements and it
makes sense to directly inline them as a small data array in the shader.
To handle this, internally the conversion might create new local
variables. SPIR-V consumers in GPU drivers may or may not optimize that
away. So this has implications over register pressure. Therefore, a
threshold is used to control when the patterns should kick in.
Reviewed By: ThomasRaoux
Differential Revision: https://reviews.llvm.org/D98052
The two dialects are largely redundant. The former was introduced as a mirror
of the latter operating on LLVM dialect types. This is no longer necessary
since the LLVM dialect operates on built-in types. Combine the two dialects.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D98060
This patch is a follow-up on D97217. It adds a new 'Skip' result to the Operation visitor
so that a callback can stop the ongoing visit of an operation/block/region and
continue visiting the next one without fully interrupting the walk. Skipping is
needed to be able to erase an operation/block in pre-order and do not continue
visiting the internals of that operation/block.
Related to the skipping mechanism, the patch also introduces the following changes:
* Added new TestIRVisitors pass with basic testing for the IR visitors.
* Fixed missing early increment ranges in visitor implementation.
* Updated documentation of walk methods to include erasure information and walk
order information.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D97820
This patch extends the Region, Block and Operation visitors to also support pre-order walks.
We introduce a new template argument that dictates the walk order (only pre-order and
post-order are supported for now). The default order for Regions, Blocks and Operations is
post-order. Mixed orders (e.g., Region/Block pre-order + Operation post-order) could easily
be implemented, as shown in NumberOfExecutions.cpp.
Reviewed By: rriddle, frgossen, bondhugula
Differential Revision: https://reviews.llvm.org/D97217
Alex Zinenko