Non-32-bit scalar types requires special hardware support that may
not exist on all Vulkan-capable GPUs. This is reflected as non-32-bit
scalar types require special capabilities or extensions to be used.
This commit makes SPIRVTypeConverter target environment aware so
that it can properly convert standard types to what is accepted on
the target environment.
Right now if a scalar type bitwidth is not supported in the target
environment, we use 32-bit unconditionally. This requires Vulkan
runtime to also feed in data with a matched bitwidth and layout,
especially for interface types. The Vulkan runtime can do that by
inspecting the SPIR-V module. Longer term, we might want to introduce
a way to control how such case are handled and explicitly fail
if wanted.
Differential Revision: https://reviews.llvm.org/D76244
Types should be checked with the type hierarchy. This should result in
better responsibility division and API surface.
Differential Revision: https://reviews.llvm.org/D76243
This commit unifies target environment queries into a new wrapper
class spirv::TargetEnv and shares across various places needing
the functionality. We still create multiple instances of TargetEnv
though given the parent components (type converters, passes,
conversion targets) have different lifetimes.
In the meantime, LowerABIAttributesPass is updated to take into
consideration the target environment, which requires updates to
tests to provide that.
Differential Revision: https://reviews.llvm.org/D76242
Previously we only consider the version/extension/capability requirement
on the op itself. This commit updates SPIRVConversionTarget to also
take into consideration the values' types when deciding op legality.
Differential Revision: https://reviews.llvm.org/D75876
Previously in SPIRVTypeConverter, we always convert memref types
to StorageBuffer regardless of their memory spaces. This commit
fixes that to let the conversion to look into memory space
properly. For this purpose, a mapping between SPIR-V storage class
and memref memory space is introduced. The mapping is arbitary
decided at the moment and the hope is that we can leverage
string memory space later to be more clear.
Now spv.interface_var_abi cannot contain storage class unless it's
attached to a scalar value, where we need the storage class as side
channel information. Verifications and tests are properly adjusted.
Differential Revision: https://reviews.llvm.org/D76241
Summary:
Although bool and int1 are the same sometimes, using bool constant matches the
semantic better. In any cases, we don't have to care the type of conditions if
we remove the intial value. The type is determined automatically by the returned
type of logical operations.
Differential Revision: https://reviews.llvm.org/D76338
Summary: The usage story in for NDEBUG isn't fleshed out yet, so this revision ensures that none of the diagnostic code exists in the binary.
Differential Revision: https://reviews.llvm.org/D76372
Summary: This is somewhat complex(annoying) as it involves directly tracking the uses within each of the callgraph nodes, and updating them as needed during inlining. The benefit of this is that we can have a more exact cost model, enable inlining some otherwise non-inlinable cases, and also ensure that newly dead callables are properly disposed of.
Differential Revision: https://reviews.llvm.org/D75476
Summary:
This adds support in RewriterGen for calling into the new `PatternRewriter::notifyMatchFailure` hook. This lets derived pattern rewriters display this information to users, an example from DialectConversion is shown below:
```
Legalizing operation : 'std.and'(0x60e0000066a0) {
* Fold {
} -> FAILURE : unable to fold
* Pattern : 'std.and -> (spv.BitwiseAnd)' {
** Failure : operand 0 of op 'std.and' failed to satisfy constraint: '8/16/32/64-bit integer or vector of 8/16/32/64-bit integer values of length 2/3/4'
} -> FAILURE : pattern failed to match
* Pattern : 'std.and -> (spv.LogicalAnd)' {
** Failure : operand 0 of op 'std.and' failed to satisfy constraint: 'bool or vector of bool values of length 2/3/4'
} -> FAILURE : pattern failed to match
} -> FAILURE : no matched legalization pattern
```
Differential Revision: https://reviews.llvm.org/D76335
Summary:
This revision restructures the calling of vector transforms to make it more flexible to ask for lowering through LLVM matrix intrinsics.
This also makes sure we bail out in degenerate cases (i.e. 1) in which LLVM complains about not being able to scalarize.
Differential Revision: https://reviews.llvm.org/D76266
LLVM has a documented mechanism for passing configuration information
to an out of tree project using cmake. See
https://llvm.org/docs/CMake.html#embedding-llvm-in-your-project. This
patch adds similar support for MLIR.
Using this requires something like:
cmake_minimum_required(VERSION 3.4.3)
project(SimpleProject)
find_package(MLIR REQUIRED CONFIG)
include_directories(${LLVM_INCLUDE_DIRS})
include_directories(${MLIR_INCLUDE_DIRS})
link_directories(${LLVM_BUILD_LIBRARY_DIR})
add_definitions(${LLVM_DEFINITIONS})
set(CMAKE_MODULE_PATH
${LLVM_CMAKE_DIR}
${MLIR_CMAKE_DIR}
)
include(AddLLVM)
include(TableGen)
include(AddMLIR)
add_executable(test-opt test-opt.cpp)
llvm_update_compile_flags(test-opt)
get_property(dialect_libs GLOBAL PROPERTY MLIR_DIALECT_LIBS)
get_property(conversion_libs GLOBAL PROPERTY MLIR_CONVERSION_LIBS)
message(dialects=${dialect_libs})
set(LIBS
${dialect_libs}
${conversion_libs}
MLIRLoopOpsTransforms
MLIRLoopAnalysis
MLIRAnalysis
MLIRDialect
MLIREDSC
MLIROptLib
MLIRParser
MLIRPass
MLIRQuantizerFxpMathConfig
MLIRQuantizerSupport
MLIRQuantizerTransforms
MLIRSPIRV
MLIRSPIRVTestPasses
MLIRSPIRVTransforms
MLIRTransforms
MLIRTransformUtils
MLIRTestDialect
MLIRTestIR
MLIRTestPass
MLIRTestTransforms
MLIRSupport
MLIRIR
MLIROptLib
LLVMSupport
LLVMCore
LLVMAsmParser
)
target_link_libraries(test-opt ${LIBS})
Differential Revision: https://reviews.llvm.org/D76047
Summary:
Renamed QuantOps to Quant to avoid the Ops suffix. All dialects will contain
ops, so the Ops suffix is redundant.
Differential Revision: https://reviews.llvm.org/D76318
Summary:
This revision adds a new hook, `notifyMatchFailure`, that allows for notifying the rewriter that a match failure is coming with the provided reason. This hook takes as a parameter a callback that fills a `Diagnostic` instance with the reason why the match failed. This allows for the rewriter to decide how this information can be displayed to the end-user, and may completely ignore it if desired(opt mode). For now, DialectConversion is updated to include this information in the debug output.
Differential Revision: https://reviews.llvm.org/D76203
MLIR supports terminators that have the same successor block with different
block operands, which cannot be expressed in the LLVM's phi-notation as the
block identifier is used to tell apart the predecessors. This limitation can be
worked around by branching to a new block instead, with this new block
unconditionally branching to the original successor and forwarding the
argument. Until now, this transformation was performed during the conversion
from the Standard to the LLVM dialect. This does not scale well to multiple
dialects targeting the LLVM dialect as all of them would have to be aware of
this limitation and perform the preparatory transformation. Instead, do it as a
separate pass and run it immediately before the translation.
Differential Revision: https://reviews.llvm.org/D75619
A memref argument is converted into a pointer-to-struct argument
of type `{T*, T*, i64, i64[N], i64[N]}*` in the wrapper function,
where T is the converted element type and N is the memref rank.
Differential Revision: https://reviews.llvm.org/D76059
Summary: This adds bitfields that map to the dialect attribute verifier hooks. This also moves over the Test dialect to have its declaration generated.
Differential Revision: https://reviews.llvm.org/D76254
Summary: PatternState was a mechanism to pass state between the match and rewrite calls of a RewritePattern. With the rise of matchAndRewrite, this class is unused and unnecessary. This revision removes PatternState and simplifies PatternMatchResult to just be a LogicalResult. A future revision will replace all usages of PatternMatchResult/matchSuccess/matchFailure with LogicalResult equivalents.
Differential Revision: https://reviews.llvm.org/D76202
- rename vars that had inst suffixes (due to ops earlier being
known as insts); other renames for better readability
- drop unnecessary matches in test cases
- iterate without block terminator
- comment/doc updates
- instBodySkew -> affineForOpBodySkew
Signed-off-by: Uday Bondhugula <uday@polymagelabs.com>
Differential Revision: https://reviews.llvm.org/D76214
Summary:
This regional op in the QuantOps dialect will be used to wrap
high-precision ops into atomic units for quantization. All the values
used by the internal ops are captured explicitly by the op inputs. The
quantization parameters of the inputs and outputs are stored in the
attributes.
Subscribers: jfb, mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, aartbik, Joonsoo, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75972
Summary: This generates the class declarations for dialects using the existing 'Dialect' tablegen classes.
Differential Revision: https://reviews.llvm.org/D76185
Setting MLIR_TABLEGEN_EXE would prevent building the native tool which is used in cross-compiling
Differential Revision: https://reviews.llvm.org/D75299
Summary:
- remove stale declarations on flat affine constraints
- avoid allocating small vectors where possible
- clean up code comments, rename some variables
Differential Revision: https://reviews.llvm.org/D76117
Summary:
To enable this, two changes are needed:
1) Add an optional attribute `padding` to linalg.conv.
2) Compute if the indices accessing is out of bound in the loops. If so, use the
padding value `0`. Otherwise, use the value derived from load.
In the patch, the padding only works for lowering without other transformations,
e.g., tiling, fusion, etc.
Differential Revision: https://reviews.llvm.org/D75722
Summary:
This revision adds lowering of vector.contract to llvm.intr.matrix_multiply.
Note that there is currently a mismatch between the MLIR vector dialect which
expects row-major layout and the LLVM matrix intrinsics which expect column
major layout.
As a consequence, we currently only match a vector.contract with indexing maps
that express column-major matrix multiplication.
Other cases would require additional transposes and it is better to wait for
LLVM intrinsics to provide a per-operation attribute that would specify which
layout is expected.
A separate integration test, not submitted to MLIR core, has independently
verified that correct execution occurs on a 2x2x2 matrix multiplication.
Differential Revision: https://reviews.llvm.org/D76014
Summary:
The direct lowering of vector.broadcast into LLVM has been replaced by
progressive lowering into elementary vector ops. This also required a
small refactoring of a llvm.mlir test that used a direct vector.broadcast
operator (just to define a matmul).
Reviewers: nicolasvasilache, andydavis1, rriddle
Reviewed By: nicolasvasilache
Subscribers: mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, liufengdb, Joonsoo, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D76143
Summary: A number of transform import StandardOps despite not being dependent on it. Cleaned it up to better understand what dialects each of these transforms depend on.
Differential Revision: https://reviews.llvm.org/D76112
Previously we only consider the version/capability/extension requirements
on ops themselves. Some types in SPIR-V also require special extensions
or capabilities to be used. For example, non-32-bit integers/floats
will require different capabilities and/or extensions depending on
where they are used because it may mean special hardware abilities.
This commit adds query methods to SPIR-V type class hierarchy to support
querying extensions and capabilities. We don't go through ODS for
auto-generating such information given that we don't have them in
SPIR-V machine readable grammar and there are just a few types.
Differential Revision: https://reviews.llvm.org/D75875
Previously extensions and capabilities requirements are returned as
SmallVector<SmallVector>. It's an anti-pattern; this commit improves
a bit by returning as SmallVector<ArrayRef>. This is possible because
the internal sequence is always known statically (from the spec)
so that we can use a static constant array for it and get an ArrayRef.
Differential Revision: https://reviews.llvm.org/D75874
This commits changes the definition of spv.module to use the #spv.vce
attribute for specifying (version, capabilities, extensions) triple
so that we can have better API and custom assembly form. Since now
we have proper modelling of the triple, (de)serialization is wired up
to use them.
With the new UpdateVCEPass, we don't need to manually specify the
required extensions and capabilities anymore when creating a spv.module.
One just need to call UpdateVCEPass before serialization to get the
needed version/extensions/capabilities.
Differential Revision: https://reviews.llvm.org/D75872
Creates an operation pass that deduces and attaches the minimal version/
capabilities/extensions requirements for spv.module ops.
For each spv.module op, this pass requires a `spv.target_env` attribute on
it or an enclosing module-like op to drive the deduction. The reason is
that an op can be enabled by multiple extensions/capabilities. So we need
to know which one to pick. `spv.target_env` gives the hard limit as for
what the target environment can support; this pass deduces what are
actually needed for a specific spv.module op.
Differential Revision: https://reviews.llvm.org/D75870
Previously we only look at the directly passed-in op for a potential
spv.target_env attribute. This commit switches to use a larger range
and recursively check enclosing symbol tables.
Differential Revision: https://reviews.llvm.org/D75869
We also need the (version, capabilities, extensions) triple on the
spv.module op. Thus far we have been using separate 'extensions'
and 'capabilities' attributes there and 'version' is missing. Creating
a separate attribute for the trip allows us to reuse the assembly
form and verification.
Differential Revision: https://reviews.llvm.org/D75868
This was a previous experiment that didn't pan out and needs to be
replaced, given no current use or tests, deleting instead and can start
new version fresh.
HasNoSideEffect can now be implemented using the MemoryEffectInterface, removing the need to check multiple things for the same information. This also removes an easy foot-gun for users as 'Operation::hasNoSideEffect' would ignore operations that dynamically, or recursively, have no side effects. This also leads to an immediate improvement in some of the existing users, such as DCE, now that they have access to more information.
Differential Revision: https://reviews.llvm.org/D76036
The current mechanism for identifying is a bit hacky and extremely adhoc, i.e. we explicit check 1-result, 0-operand, no side-effect, and always foldable and then assume that this is a constant. Adding a trait adds structure to this, and makes checking for a constant much more efficient as we can guarantee that all of these things have already been verified.
Differential Revision: https://reviews.llvm.org/D76020
These terminator operations don't really have any side effects, and this allows for more accurate side-effect analysis for region operations. For example, currently we can't detect like a loop.for or affine.for are dead because the affine.terminator is "side effecting".
Note: Marking as NoSideEffect doesn't mean that these operations can be opaquely erased.
Differential Revision: https://reviews.llvm.org/D75888
Summary:
This replaces the direct lowering of vector.outerproduct to LLVM with progressive lowering into elementary vectors ops to avoid having the similar lowering logic at several places.
NOTE1: with the new progressive rule, the lowered llvm is slightly more elaborate than with the direct lowering, but the generated assembly is just as optimized; still if we want to stay closer to the original, we should add a "broadcast on extract" to shuffle rewrite (rather than special cases all the lowering steps)
NOTE2: the original outerproduct lowering code should now be removed but some linalg test work directly on vector and contain some dead code, so this requires another CL
Reviewers: nicolasvasilache, andydavis1
Reviewed By: nicolasvasilache, andydavis1
Subscribers: mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, liufengdb, Joonsoo, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75956
Summary:
The C runner utils API was still not vanilla enough for certain use
cases on embedded ARM SDKs, this enables such cases.
Adding people more widely for historical Windows related build issues.
Differential Revision: https://reviews.llvm.org/D76031
MLIR does not have a Sphinx configuration, this is just leading to build
failures at the moment.
The website https://mlir.llvm.org/ is using the Hugo generator to
process the markdown files.
Summary:
affineDataCopyGenerate is a monolithinc function that
combines several steps for good reasons, but it makes customizing
the behaivor even harder. The major two steps by affineDataCopyGenerate are:
a) Identify interesting memrefs and collect their uses.
b) Create new buffers to forward these uses.
Step (a) actually has requires tremendous customization options. One could see
that from the recently added filterMemRef parameter.
This patch adds a function that only does (b), in the hope that (a)
can be directly implemented by the callers. In fact, (a) is quite
simple if the caller has only one buffer to consider, or even one use.
Differential Revision: https://reviews.llvm.org/D75965
Summary:
Now that, thanks to ntv, we have the ability to parse and represent an affine
map with rank-0 results, viz. (i,j) -> (), we can pay off some engineering debt
in special casing the verification of such affine maps in dot-product flavored
vector.contract operations.
Reviewers: nicolasvasilache, andydavis1, rriddle
Reviewed By: nicolasvasilache
Subscribers: mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, liufengdb, Joonsoo, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D76028
Summary: In some situations the name of the attribute is not representable as a bare-identifier, this revision adds support for those cases by formatting the name as a string instead. This has the added benefit of removing the identifier regex from the verifier.
Differential Revision: https://reviews.llvm.org/D75973
The three libs where recently added to the `mlir-cpu-runner`'s
`CMakeLists.txt` file. This prevent the runner to compile on other
platform (e.g. Power in my case). Native codegen is pulled in
by the ExecutionEngine library, so this is redundant in any case.
Differential Revision: https://reviews.llvm.org/D75916
Summary:
This patch add some builtin operation for the gpu.all_reduce ops.
- for Integer only: `and`, `or`, `xor`
- for Float and Integer: `min`, `max`
This is useful for higher level dialect like OpenACC or OpenMP that can lower to the GPU dialect.
Differential Revision: https://reviews.llvm.org/D75766
Summary:
1-bit integer is tricky in different dialects sometimes. E.g., there is no
arithmetic instructions on 1-bit integer in SPIR-V, i.e., `spv.IMul %0, %1 : i1`
is not valid. Instead, `spv.LogicalAnd %0, %1 : i1` is valid. Creating the op
directly makes lowering easier because we don't need to match a complicated
pattern like `!(!lhs && !rhs)`. Also, this matches the semantic better.
Also add assertions on inputs.
Differential Revision: https://reviews.llvm.org/D75764
Summary:
This patch add some builtin operation for the gpu.all_reduce ops.
- for Integer only: `and`, `or`, `xor`
- for Float and Integer: `min`, `max`
This is useful for higher level dialect like OpenACC or OpenMP that can lower to the GPU dialect.
Differential Revision: https://reviews.llvm.org/D75766
* Adds GpuLaunchFuncToVulkanLaunchFunc conversion pass.
* Moves a serialization of the `spirv::Module` from LaunchFuncToVulkanCalls pass to newly created pass.
* Updates LaunchFuncToVulkanCalls instrumentation pass, adds `initVulkan` and `deinitVulkan` runtime calls.
* Adds `bindResource` call to bind specifc resource by the given descriptor set and descriptor binding.
* Eliminates static construction and desctruction of `VulkanRuntimeManager`.
Differential Revision: https://reviews.llvm.org/D75192
Summary:
Interfaces/ is the designated directory for these types of interfaces, and also removes the need for including them directly in IR/.
Differential Revision: https://reviews.llvm.org/D75886
The interfaces themselves aren't really analyses, they may be used by analyses though. Having them in Analysis can also create cyclic dependencies if an analysis depends on a specific dialect, that also provides one of the interfaces.
Differential Revision: https://reviews.llvm.org/D75867
This revision takes advantage of the empty AffineMap to specify the
0-D edge case. This allows removing a bunch of annoying corner cases
that ended up impacting users of Linalg.
Differential Revision: https://reviews.llvm.org/D75831
Summary:
The old interface was a temporary stopgap to allow for implementing simple LICM that took side effects of region operations into account. Now that MLIR has proper support for specifying memory effects, this interface can be deleted.
Differential Revision: https://reviews.llvm.org/D74441
Summary: This op mirrors the llvm.intr counterpart and allows lowering + type conversions in a progressive fashion.
Differential Revision: https://reviews.llvm.org/D75775
Summary:
This revision adds intrinsics for transpose, columnwise.load and columnwise.store
achieving full coverage of the llvm.matrix intrinsics.
Differential Revision: https://reviews.llvm.org/D75852
Summary:
This is the counterpart of VectorOfLength for ranks.
This will be used in lowering vector.contract operations to llvm.matrix
Differential Revision: https://reviews.llvm.org/D75771
Summary:
This way, clients can opt-out of the GDB notification listener. Also, this
changes the semantics of enabling the object cache, which seemed the wrong
way around.
Reviewers: rriddle, nicolasvasilache, ftynse, andydavis1
Reviewed By: nicolasvasilache
Subscribers: mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, liufengdb, Joonsoo, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75787
add convenience method for affine data copy generation for a loop body
Signed-off-by: Uday Bondhugula <uday@polymagelabs.com>
Differential Revision: https://reviews.llvm.org/D75822
Summary:
On Windows, building `mlir_c_runner_utils` doesn't properly export
symbols, thus resulting in an implib not being created, which causes
an error when consuming LLVM from external projects.
Differential Revision: https://reviews.llvm.org/D75769
MSVC qualifies the Effect reference contextually depending on where the
template is instantiated, leading to compiler failures if there is a
different Effect class defined.
Summary:
New classes are added to ODS to enable specifying additional information on the arguments and results of an operation. These classes, `Arg` and `Res` allow for adding a description and a set of 'decorators' along with the constraint. This enables specifying the side effects of an operation directly on the arguments and results themselves.
Example:
```
def LoadOp : Std_Op<"load"> {
let arguments = (ins Arg<AnyMemRef, "the MemRef to load from",
[MemRead]>:$memref,
Variadic<Index>:$indices);
}
```
Differential Revision: https://reviews.llvm.org/D74440
This revision introduces the infrastructure for defining side-effects and attaching them to operations. This infrastructure allows for defining different types of side effects, that don't interact with each other, but use the same internal mechanisms. At the base of this is an interface that allows operations to specify the different effect instances that are exhibited by a specific operation instance. An effect instance is comprised of the following:
* Effect: The specific effect being applied.
For memory related effects this may be reading from memory, storing to memory, etc.
* Value: A specific value, either operand/result/region argument, the effect pertains to.
* Resource: This is a global entity that represents the domain within which the effect is being applied.
MLIR serves many different abstractions, which cover many different domains. Simple effects are may have very different context, for example writing to an in-memory buffer vs a database. This revision defines uses this infrastructure to define a set of initial MemoryEffects. The are effects that generally correspond to memory of some kind; Allocate, Free, Read, Write.
This set of memory effects will be used in follow revisions to generalize various parts of the compiler, and make others more powerful(e.g. DCE).
This infrastructure was originally proposed here:
https://groups.google.com/a/tensorflow.org/g/mlir/c/v2mNl4vFCUM
Differential Revision: https://reviews.llvm.org/D74439
Putting this up mainly for discussion on
how this should be done. I am interested in MLIR from
the Julia side and we currently have a strong preference
to dynamically linking against the LLVM shared library,
and would like to have a MLIR shared library.
This patch adds a new cmake function add_mlir_library()
which accumulates a list of targets to be compiled into
libMLIR.so. Note that not all libraries make sense to
be compiled into libMLIR.so. In particular, we want
to avoid libraries which primarily exist to support
certain tools (such as mlir-opt and mlir-cpu-runner).
Note that the resulting libMLIR.so depends on LLVM, but
does not contain any LLVM components. As a result, it
is necessary to link with libLLVM.so to avoid linkage
errors. So, libMLIR.so requires LLVM_BUILD_LLVM_DYLIB=on
FYI, Currently it appears that LLVM_LINK_LLVM_DYLIB is broken
because mlir-tblgen is linked against libLLVM.so and
and independent LLVM components.
Previous version of this patch broke depencies on TableGen
targets. This appears to be because it compiled all
libraries to OBJECT libraries (probably because cmake
is generating different target names). Avoiding object
libraries results in correct dependencies.
(updated by Stephen Neuendorffer)
Differential Revision: https://reviews.llvm.org/D73130
add_llvm_library and add_llvm_executable may need to create new targets with
appropriate dependencies. As a result, it is not sufficient in some
configurations (namely LLVM_BUILD_LLVM_DYLIB=on) to only call
add_dependencies(). Instead, the explicit TableGen dependencies must
be passed to add_llvm_library() or add_llvm_executable() using the DEPENDS
keyword.
Differential Revision: https://reviews.llvm.org/D74930
In cmake, it is redundant to have a target list under target_link_libraries()
and add_dependency(). This patch removes the redundant dependency from
add_dependency().
Differential Revision: https://reviews.llvm.org/D74929
CMake allows calling target_link_libraries() without a keyword,
but this usage is not preferred when also called with a keyword,
and has surprising behavior. This patch explicitly specifies a
keyword when using target_link_libraries().
Differential Revision: https://reviews.llvm.org/D75725
This revision adds the first intrinsic for llvm.matrix.multiply.
This uses the more general `LLVM_OneResultOp` for now since the goal is
to use the
specific Matrix builders that @fhahn has created recently.
When piped through:
```
opt -O3 -enable-matrix | llc -O3 -march=x86-64 -mcpu=skylake-avx512
```
this has been verified to generate ymm instructions.
Additional function attribute support will be needed to generate proper
zmm instructions but at least things run end to end.
Benchmarking will be provided separately with the experimental
metaprogramming
[ModelBuilder](https://github.com/google/iree/tree/master/experimental/ModelBuilder)
tool when ready.
Summary:
Paying off some technical debt in VectorOps, where I introduced a special
op for a fused accumulator into reduction to avoid some issues around
printing and parsing an optional accumulator. This CL merges the two
into one op again and does things the right way (still would be nice
to have "assemblyFormat" for optional operands though....).
Reviewers: nicolasvasilache, andydavis1, ftynse, rriddle
Reviewed By: nicolasvasilache
Subscribers: mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, liufengdb, Joonsoo, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75699
Summary:
This revision removes all of the functionality related to successor operands on the core Operation class. This greatly simplifies a lot of handling of operands, as well as successors. For example, DialectConversion no longer needs a special "matchAndRewrite" for branching terminator operations.(Note, the existing method was also broken for operations with variadic successors!!)
This also enables terminator operations to define their own relationships with successor arguments, instead of the hardcoded "pass-through" behavior that exists today.
Differential Revision: https://reviews.llvm.org/D75318
This greatly simplifies the requirements for builders using this mechanism for managing variadic operands.
Differential Revision: https://reviews.llvm.org/D75317
The existing API for successor operands on operations is in the process of being removed. This revision simplifies a later one that completely removes the existing API.
Differential Revision: https://reviews.llvm.org/D75316
This attribute details the segment sizes for operand groups within the operation. This revision add support for automatically populating this attribute in the declarative parser.
Differential Revision: https://reviews.llvm.org/D75315
This interface contains the necessary components to provide the same builtin behavior that terminators have. This will be used in future revisions to remove many of the hardcoded constraints placed on successors and successor operands. The interface initially contains three methods:
```c++
// Return a set of values corresponding to the operands for successor 'index', or None if the operands do not correspond to materialized values.
Optional<OperandRange> getSuccessorOperands(unsigned index);
// Return true if this terminator can have it's successor operands erased.
bool canEraseSuccessorOperand();
// Erase the operand of a successor. This is only valid to call if 'canEraseSuccessorOperand' returns true.
void eraseSuccessorOperand(unsigned succIdx, unsigned opIdx);
```
Differential Revision: https://reviews.llvm.org/D75314
This allows for simplifying OpDefGen, as well providing specializing accessors for the different successor counts. This mirrors the existing traits for operands and results.
Differential Revision: https://reviews.llvm.org/D75313
The current setup of the GPU dialect is to model both the host and
device side codegen. For cases (like IREE) the host side modeling
might not directly fit its use case, but device-side codegen is still
valuable. First step in accessing just the device-side functionality
of the GPU dialect is to allow just creating a gpu.func operation from
a gpu.launch operation. In addition this change also "inlines"
operations into the gpu.func op at time of creation instead of this
being a later step.
Differential Revision: https://reviews.llvm.org/D75287
Summary:
This patch adds support for translation of the OpenMP barrier construct to LLVM
IR. The OpenMP IRBuilder is used for this translation. In this patch the code
for translation is added to the existing LLVM dialect translation to LLVM IR.
The patch includes code changes and a testcase.
Reviewers: jdoerfert, nicolasvasilache, ftynse, rriddle, mehdi_amini
Reviewed By: ftynse, rriddle, mehdi_amini
Differential Revision: https://reviews.llvm.org/D72962
output has zero rank.
While lowering to loops, no indices should be used in the load/store
operation if the buffer is zero-rank.
Differential Revision: https://reviews.llvm.org/D75391
Summary:
Expose expandAffineMap so that it can be used by lowerings defined outside of
MLIR core.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D75589
This commit adds timestamp query commands in Vulkan runner's
compute pipeline to gain insights into how long it takes to
run the compute shader. This commit also adds timing from CPU
side for VkQueueSubmit and vkQueueWaitIdle.
Differential Revision: https://reviews.llvm.org/D75531
Summary:
Looks like a refactor that was never completed.
This change removes some unused and ambiguous definitions.
Reviewed By: bondhugula, nicolasvasilache, rriddle
Differential Revision: https://reviews.llvm.org/D75586
This commit updates SPIR-V dialect to support integer signedness
by relaxing various checks for signless to just normal integers.
The hack for spv.Bitcast can now be removed.
Differential Revision: https://reviews.llvm.org/D75611
A previous commit added support for integer signedness in C++
IntegerType. This change introduces ODS definitions for
integer types and integer (element) attributes w.r.t. signedness.
This commit also updates various existing definitions' descriptions
to mention signless where suitable to make it more clear.
Positive and non-negative integer attributes are removed to avoid
the explosion of subclasses. Instead, one should use more atmoic
constraints together with Confined to model that. For example,
`Confined<..., [IntPositive]>`.
Differential Revision: https://reviews.llvm.org/D75610
Summary:
The order of the operations has fallen out of sync as operations have been renamed and new ones have been added.
Differential Revision: https://reviews.llvm.org/D75540
Recently introduced support for converting sequential reduction loops to
CFG of basic blocks in the Standard dialect makes it possible to perform
a staged conversion of parallel reduction loops into a similar CFG by
using sequential loops as an intermediate step. This is already the case
for parallel loops without reduction, so extend the pattern to support
an additional use case.
Differential Revision: https://reviews.llvm.org/D75599
Summary:
This adds an rsqrt op to the standard dialect, and lowers
it as 1 / sqrt to the LLVM dialect.
Differential Revision: https://reviews.llvm.org/D75353
Summary:
Previously, we would, for an empty file, print the somewhat confusing
Assertion `tok == curTok [...]' failed.
With this change, we now print
Parse error [...]: expected 'def' [...]
This only affects the parser from chapters 1-6, since the more advanced
chapter 7 parser is actually able to generate an empty module from an
empty file. Nonetheless, this commit also adds the additional check to
the chapter 7 parser, for consistency.
Differential Revision: https://reviews.llvm.org/D75534
Summary: This allows for attaching the attribute to CmpF as a proper argument, and thus enables the removal of a bunch of c++ code.
Differential Revision: https://reviews.llvm.org/D75539
For ODS generated operations enable querying whether there is a derived
attribute with a given name.
Rollforward of commit 5aa57c2 without using llvm::is_contained.
Summary:
Introduce support for converting loop.for operations with loop-carried values
to a CFG in the standard dialect. This is achieved by passing loop-carried
values as block arguments to the loop condition block. This block dominates
both the loop body and the block immediately following the loop, so the
arguments of this block are remain visible there.
Differential Revision: https://reviews.llvm.org/D75513
MLIR ExecutionEngine and derived tools (e.g., mlir-cpu-runner) would trigger an
assertion inside ORC JIT while ExecutionEngine is being destructed after a
failed linking due to a missing function definition. The reason for this is the
JIT lookup that may return an Error referring to strings stored internally by
the JIT. If the Error outlives the ExecutionEngine, it would want have a
dangling reference, which is currently caught by an assertion inside JIT thanks
to hand-rolled reference counting. Rewrap the error message into a string
before returning.
Differential Revision: https://reviews.llvm.org/D75508
Some attribute kinds are not supported as "value" attributes of
`llvm.mlir.constant` when translating to LLVM IR. We were correctly reporting
an error, but continuing the translation using an "undef" value instead,
leading to a surprising mix of error messages and output IR. Abort the
translation after the error is reported.
Differential Revision: https://reviews.llvm.org/D75450
Summary:
This revision fixes a -Wzero-length-array compile error that
caused e459596917 which reverted
78f9e5d098.
Also fixes a struct vs class mismatch that broke compilation with
-Werror for Windows that caused
57397eba7a.
This revision adds padding for 1-D Vector in the common case of x86
execution with a stadard data layout. This supports properly interfacing
codegen with arrays of e.g. `vector<9xf32>`.
Such vectors are already assumed padded to the next power of 2 by LLVM
codegen with the default x86 data layout:
```
define void @test_vector_add_1d_2_3(<3 x float>* nocapture readnone %0,
<3 x float>* nocapture readonly %1, i64 %2, i64 %3, i64 %4, <3 x float>*
nocapture readnone %5, <3 x float>* nocapture readonly %6, i64 %7, i64
%8, i64 %9, <3 x float>* nocapture readnone %10, <3 x float>* nocapture
%11, i64 %12, i64 %13, i64 %14) local_unnamed_addr {
%16 = getelementptr <3 x float>, <3 x float>* %6, i64 1
%17 = load <3 x float>, <3 x float>* %16, align 16
%18 = getelementptr <3 x float>, <3 x float>* %1, i64 1
%19 = load <3 x float>, <3 x float>* %18, align 16
%20 = fadd <3 x float> %17, %19
%21 = getelementptr <3 x float>, <3 x float>* %11, i64 1
```
The pointer addressing a `vector<3xf32>` is assumed aligned `@16`.
Similarly, the pointer addressing a `vector<65xf32>` is assumed aligned
`@512`.
This revision allows using objects such as `vector<3xf32>` properly with
the standard x86 data layout used in the JitRunner. Integration testing
is done out of tree, at the moment such testing fails without this
change.
Differential Revision: https://reviews.llvm.org/D75459
This revision adds a static `mlir_c_runner_utils_static` library
for the sole purpose of being linked into `mlir_runner_utils` on
Windows.
It was previously reported that:
```
`add_llvm_library(mlir_c_runner_utils SHARED CRunnerUtils.cpp)`
produces *only* a dll on windows, the linking of mlir_runner_utils fails
because target_link_libraries is looking for a .lib file as opposed to a
.dll file. I think this may be a case where either we need to use
LINK_LIBS or explicitly build a static lib as well, but I haven't tried
either yet.
```
This reverts commit 5aa57c2812.
The source code generated due to this ods change does not compile,
as it passes to few arguments to llvm::is_contained.
This revision adds a static `mlir_c_runner_utils_static` library
for the sole purpose of being linked into `mlir_runner_utils` on
Windows.
It was previously reported that:
```
`add_llvm_library(mlir_c_runner_utils SHARED CRunnerUtils.cpp)`
produces *only* a dll on windows, the linking of mlir_runner_utils fails
because target_link_libraries is looking for a .lib file as opposed to a
.dll file. I think this may be a case where either we need to use
LINK_LIBS or explicitly build a static lib as well, but I haven't tried
either yet.
```
Summary:
This revision fixes a -Wzero-length-array compile error that
caused e459596917 which reverted
78f9e5d098.
This revision adds padding for 1-D Vector in the common case of x86
execution with a stadard data layout. This supports properly interfacing
codegen with arrays of e.g. `vector<9xf32>`.
Such vectors are already assumed padded to the next power of 2 by LLVM
codegen with the default x86 data layout:
```
define void @test_vector_add_1d_2_3(<3 x float>* nocapture readnone %0,
<3 x float>* nocapture readonly %1, i64 %2, i64 %3, i64 %4, <3 x float>*
nocapture readnone %5, <3 x float>* nocapture readonly %6, i64 %7, i64
%8, i64 %9, <3 x float>* nocapture readnone %10, <3 x float>* nocapture
%11, i64 %12, i64 %13, i64 %14) local_unnamed_addr {
%16 = getelementptr <3 x float>, <3 x float>* %6, i64 1
%17 = load <3 x float>, <3 x float>* %16, align 16
%18 = getelementptr <3 x float>, <3 x float>* %1, i64 1
%19 = load <3 x float>, <3 x float>* %18, align 16
%20 = fadd <3 x float> %17, %19
%21 = getelementptr <3 x float>, <3 x float>* %11, i64 1
```
The pointer addressing a `vector<3xf32>` is assumed aligned `@16`.
Similarly, the pointer addressing a `vector<65xf32>` is assumed aligned
`@512`.
This revision allows using objects such as `vector<3xf32>` properly with
the standard x86 data layout used in the JitRunner. Integration testing
is done out of tree, at the moment such testing fails without this
change.
Differential Revision: https://reviews.llvm.org/D75459
as it broke the Werror build:
.../sources/llvm-project/mlir/include/mlir/ExecutionEngine/CRunnerUtils.h:85:16: error: zero size arrays are an extension [-Werror,-Wzero-length-array]
char padding[detail::nextPowerOf2<sizeof(T[Dim])>() - sizeof(T[Dim])];
^~~~~~~~~~~~~~~
This reverts commit 78f9e5d098.
Summary:
Make computeConversionSet bubble up errors from nested regions. Note
that this doesn't change top-level behavior - since the nested region
calls emitError, the error was visible before, just not surfaced as
quickly.
Differential Revision: https://reviews.llvm.org/D75369
Summary: For example, DenseElementsAttr currently does not properly round-trip unsigned integer values.
Differential Revision: https://reviews.llvm.org/D75374
Summary:
This revision adds padding for 1-D Vector in the common case of x86
execution with a stadard data layout. This supports properly interfacing
codegen with arrays of e.g. `vector<9xf32>`.
Such vectors are already assumed padded to the next power of 2 by LLVM
codegen with the default x86 data layout:
```
define void @test_vector_add_1d_2_3(<3 x float>* nocapture readnone %0,
<3 x float>* nocapture readonly %1, i64 %2, i64 %3, i64 %4, <3 x float>*
nocapture readnone %5, <3 x float>* nocapture readonly %6, i64 %7, i64
%8, i64 %9, <3 x float>* nocapture readnone %10, <3 x float>* nocapture
%11, i64 %12, i64 %13, i64 %14) local_unnamed_addr {
%16 = getelementptr <3 x float>, <3 x float>* %6, i64 1
%17 = load <3 x float>, <3 x float>* %16, align 16
%18 = getelementptr <3 x float>, <3 x float>* %1, i64 1
%19 = load <3 x float>, <3 x float>* %18, align 16
%20 = fadd <3 x float> %17, %19
%21 = getelementptr <3 x float>, <3 x float>* %11, i64 1
```
The pointer addressing a `vector<3xf32>` is assumed aligned `@16`.
Similarly, the pointer addressing a `vector<65xf32>` is assumed aligned
`@512`.
This revision allows using objects such as `vector<3xf32>` properly with
the standard x86 data layout used in the JitRunner. Integration testing
is done out of tree, at the moment such testing fails without this
change.
Reviewers: ftynse
Subscribers: mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, arpith-jacob, mgester, lucyrfox, liufengdb, Joonsoo, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75459
Summary: Added brackets to fix the loop trip count computation.
The brackets ensure the bounds are subtracted before we divide
the result by the step of the loop.
Differential Revision: https://reviews.llvm.org/D75449
Summary:
This is to ensure that the template declaration is seen before
any template specialization.
Reviewers: mravishankar, antiagainst, rriddle!
Differential Revision: https://reviews.llvm.org/D75442
A printer refactoring removed automatic newline printing in the printer
of a ModuleOp. As a consequence, mlir-opt no longer printed a newline
after the closing brace of a module, which made it hard to distinguish
when used from command line. Print the newline character explicitly in
mlir-opt.
Putting this up mainly for discussion on
how this should be done. I am interested in MLIR from
the Julia side and we currently have a strong preference
to dynamically linking against the LLVM shared library,
and would like to have a MLIR shared library.
This patch adds a new cmake function add_mlir_library()
which accumulates a list of targets to be compiled into
libMLIR.so. Note that not all libraries make sense to
be compiled into libMLIR.so. In particular, we want
to avoid libraries which primarily exist to support
certain tools (such as mlir-opt and mlir-cpu-runner).
Note that the resulting libMLIR.so depends on LLVM, but
does not contain any LLVM components. As a result, it
is necessary to link with libLLVM.so to avoid linkage
errors. So, libMLIR.so requires LLVM_BUILD_LLVM_DYLIB=on
FYI, Currently it appears that LLVM_LINK_LLVM_DYLIB is broken
because mlir-tblgen is linked against libLLVM.so and
and independent LLVM components.
Previous version of this patch broke depencies on TableGen
targets. This appears to be because it compiled all
libraries to OBJECT libraries (probably because cmake
is generating different target names). Avoiding object
libraries results in correct dependencies.
(updated by Stephen Neuendorffer)
Differential Revision: https://reviews.llvm.org/D73130
add_llvm_library and add_llvm_executable may need to create new targets with
appropriate dependencies. As a result, it is not sufficient in some
configurations (namely LLVM_BUILD_LLVM_DYLIB=on) to only call
add_dependencies(). Instead, the explicit TableGen dependencies must
be passed to add_llvm_library() or add_llvm_executable() using the DEPENDS
keyword.
Differential Revision: https://reviews.llvm.org/D74930
In cmake, it is redundant to have a target list under target_link_libraries()
and add_dependency(). This patch removes the redundant dependency from
add_dependency().
Differential Revision: https://reviews.llvm.org/D74929
When compiling libLLVM.so, add_llvm_library() manipulates the link libraries
being used. This means that when using add_llvm_library(), we need to pass
the list of libraries to be linked (using the LINK_LIBS keyword) instead of
using the standard target_link_libraries call. This is preparation for
properly dealing with creating libMLIR.so as well.
Differential Revision: https://reviews.llvm.org/D74864
This is avoid the user to shoot themselves in the foot and encounter
strange crashes that are confusing until one run with TSAN.
Differential Revision: https://reviews.llvm.org/D75399
This is in preparation for the next patch D75141. The purpose is to
provide a single place where LLVM dialect registers its ops as
legal/illegal.
Reviewers: ftynse, mravishankar, herhut
Subscribers: jholewinski, bixia, sanjoy.google, mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, nicolasvasilache, csigg, arpith-jacob, mgester, lucyrfox, aartbik, liufengdb, Joonsoo, llvm-commits
Differential Revision: https://reviews.llvm.org/D75140
Putting this up mainly for discussion on
how this should be done. I am interested in MLIR from
the Julia side and we currently have a strong preference
to dynamically linking against the LLVM shared library,
and would like to have a MLIR shared library.
This patch adds a new cmake function add_mlir_library()
which accumulates a list of targets to be compiled into
libMLIR.so. Note that not all libraries make sense to
be compiled into libMLIR.so. In particular, we want
to avoid libraries which primarily exist to support
certain tools (such as mlir-opt and mlir-cpu-runner).
Note that the resulting libMLIR.so depends on LLVM, but
does not contain any LLVM components. As a result, it
is necessary to link with libLLVM.so to avoid linkage
errors. So, libMLIR.so requires LLVM_BUILD_LLVM_DYLIB=on
FYI, Currently it appears that LLVM_LINK_LLVM_DYLIB is broken
because mlir-tblgen is linked against libLLVM.so and
and independent LLVM components
(updated by Stephen Neuendorffer)
Differential Revision: https://reviews.llvm.org/D73130
add_llvm_library and add_llvm_executable may need to create new targets with
appropriate dependencies. As a result, it is not sufficient in some
configurations (namely LLVM_BUILD_LLVM_DYLIB=on) to only call
add_dependencies(). Instead, the explicit TableGen dependencies must
be passed to add_llvm_library() or add_llvm_executable() using the DEPENDS
keyword.
Differential Revision: https://reviews.llvm.org/D74930
In cmake, it is redundant to have a target list under target_link_libraries()
and add_dependency(). This patch removes the redundant dependency from
add_dependency().
Differential Revision: https://reviews.llvm.org/D74929
When compiling libLLVM.so, add_llvm_library() manipulates the link libraries
being used. This means that when using add_llvm_library(), we need to pass
the list of libraries to be linked (using the LINK_LIBS keyword) instead of
using the standard target_link_libraries call. This is preparation for
properly dealing with creating libMLIR.so as well.
Differential Revision: https://reviews.llvm.org/D74864
Previously, lib/Support/JitRunner.cpp was essentially a complete application,
performing all library initialization, along with dealing with command line
arguments and actually running passes. This differs significantly from
mlir-opt and required a dependency on InitAllDialects.h. This dependency
is significant, since it requires a dependency on all of the resulting
libraries.
This patch refactors the code so that tools are responsible for library
initialization, including registering all dialects, prior to calling
JitRunnerMain. This places the concern about what dialect to support
with the end application, enabling more extensibility at the cost of
a small amount of code duplication between tools. It also fixes
BUILD_SHARED_LIBS=on.
Differential Revision: https://reviews.llvm.org/D75272
Collect a list of conversion libraries in cmake, so we don't have to
list these explicitly in most binaries.
Differential Revision: https://reviews.llvm.org/D75222
Instead of creating extra libraries we don't really need, collect a
list of all dialects and use that instead.
Differential Revision: https://reviews.llvm.org/D75221
This matches loops with a affine.min upper bound, limiting the trip
count to a constant, and rewrites them into two loops, one with constant
upper bound and one with variable upper bound. The assumption is that
the constant upper bound loop will be unrolled and vectorized, which is
preferable if this is the hot path.
Differential Revision: https://reviews.llvm.org/D75240
Summary:
* add missing comma.
* remove "having to register them here" phrasing, since register it
is what we're doing, which made the comment a bit confusing.
* remove duplicate code.
* clarify link to chapter 3, since "folder" doesn't appear in that
chapter.
Differential Revision: https://reviews.llvm.org/D75263
Summary:
* Use bold font (not monospace) for legal/illegal.
* Say a few words about operation<->dialect precedence.
* Omit duplicate code samples.
* Indent items in bullet-point list.
Differential Revision: https://reviews.llvm.org/D75262
Summary:
* Let's use "override" when we're just doing standard baseclassing.
("Specialization" makes it sound like template specialization, which
this is not.)
* CallInterfaces.td has an include guard, so #ifdef not needed anymore.
* Omit duplicate code in code samples.
* Clarify which algorithm we're talking about.
* Mention that the ShapeInference code is code a snippet that belongs to
algorithm discussed in the paragraph above it.
* Add missing definition for createShapeInferencePass.
Differential Revision: https://reviews.llvm.org/D75260
Lei Zhang