Added lowering support for IS_DEVICE_PTR and HAS_DEVICE_ADDR clauses for
OMP TARGET directive and added related tests for these changes.
IS_DEVICE_PTR and HAS_DEVICE_ADDR clauses apply to OMP TARGET directive
OpenMP spec states
`The **is_device_ptr** clause indicates that its list items are device
pointers.`
`The **has_device_addr** clause indicates that its list items already
have device addresses and therefore they may be directly accessed from a
target device.`
Whereas USE_DEVICE_PTR and USE_DEVICE_ADDR clauses apply to OMP TARGET
DATA directive and OpenMP spec for them states
`Each list item in the **use_device_ptr** clause results in a new list
item that is a device pointer that refers to a device address`
`Each list item in a **use_device_addr** clause that is present in the
device data environment is treated as if it is implicitly mapped by a
map clause on the construct with a map-type of alloc`
Currently, by-ref reductions will allocate the per-thread reduction
variable in the initialization region. Adding a cleanup region allows
that allocation to be undone. This will allow flang to support reduction
of arrays stored on the heap.
This conflation of allocation and initialization in the initialization
should be fixed in the future to better match the OpenMP standard, but
that is beyond the scope of this patch.
Extends the `omp.parallel` op by adding a `private` clause to model
[first]private variables. This uses the `omp.private` op to map
privatized variables to their corresponding privatizers.
Example `omp.private` op with `private` variable:
```
omp.parallel private(@x.privatizer %arg0 -> %arg1 : !llvm.ptr) {
^bb0(%arg1: !llvm.ptr):
// ... use %arg1 ...
omp.terminator
}
```
Whether the variable is private or firstprivate is determined by the
attributes of the corresponding `omp.private` op.
This adds a new custom CopyPrivateVarList to the single operation.
Each list item is formed by a reference to the variable to be
updated, its type and the function to be used to perform the copy.
It will be translated to LLVM IR using OpenMP builder, that will
use the information in the copyprivate list to call
__kmpc_copyprivate.
This is patch 2 of 4, to add support for COPYPRIVATE in Flang.
Original PR: https://github.com/llvm/llvm-project/pull/73128
This patch reworks the way that wsloop reduction operations function to
better match the expected semantics from the OpenMP specification,
following the rework of parallel reductions.
The new semantics create a private reduction variable as a block
argument which should be used normally for all operations on that
variable in the region; this private variable is then combined with the
others into the shared variable. This way no special omp.reduction
operations are needed inside the region. These block arguments follow
the loop control block arguments.
---------
Co-authored-by: Kiran Chandramohan <kiran.chandramohan@arm.com>
This patch adds support for the depend clause in a number of OpenMP
directives/constructs related to offloading. Specifically, it adds the
handling of the depend clause when it is used with the following
constructs
- target
- target enter data
- target update data
- target exit data
This PR adds a new op to the OpenMP dialect: `PrivateClauseOp`. This op
will be later used to model `[first]private` clauses for differnt OpenMP
directives.
This is part of productizing the "delayed privatization" PoC which can
be found in #79862.
This patch reworks the way that parallel reduction operations function
to better match the expected semantics from the OpenMP specification.
Previously specific omp.reduction operations were used inside the
region, meaning that the reduction only applied when the correct
operation was used, whereas the specification states that any change to
the variable inside the region should be taken into account for the
reduction.
The new semantics create a private reduction variable as a block
argument which should be used normally for all operations on that
variable in the region; this private variable is then combined with the
others into the shared variable. This way no special omp.reduction
operations are needed inside the region.
This patch only makes the change for the `parallel` operation, the
change for the `wsloop` operation will be in a separate patch.
---------
Co-authored-by: Kiran Chandramohan <kiran.chandramohan@arm.com>
This patch seeks to add an initial lowering for pointers and allocatable variables
captured by implicit and explicit map in Flang OpenMP for Target operations that
take map clauses e.g. Target, Target Update. Target Exit/Enter etc.
Currently this is done by treating the type that lowers to a descriptor
(allocatable/pointer/assumed shape) as a map of a record type (e.g. a structure) as that's
effectively what descriptor types lower to in LLVM-IR and what they're represented as
in the Fortran runtime (written in C/C++). The descriptor effectively lowers to a structure
containing scalar and array elements that represent various aspects of the underlying
data being mapped (lower bound, upper bound, extent being the main ones of interest
in most cases) and a pointer to the allocated data. In this current iteration of the mapping
we map the structure in it's entirety and then attach the underlying data pointer and map
the data to the device, this allows most of the required data to be resident on the device
for use. Currently we do not support the addendum (another block of pointer data), but
it shouldn't be too difficult to extend this to support it.
The MapInfoOp generation for descriptor types is primarily handled in an optimization
pass, where it expands BoxType (descriptor types) map captures into two maps, one for
the structure (scalar elements) and the other for the pointer data (base address) and
links them in a Parent <-> Child relationship. The later lowering processes will then treat
them as a conjoined structure with a pointer member map.
This patch adds the omp.distribute operation to the OMP dialect. The
purpose is to be able to represent the distribute construct in OpenMP
with the associated clauses. The effect of the operation is to
distributes the loop iterations of the loop(s) contained inside the
region across multiple teams.
After the removal of the OpenMP early outlining MLIR pass in #67319, the
`EarlyOutliningInterface` stopped doing any useful work. It used to be
necessary to tie the name of the function from which a target region was
outlined to that new function, so it would be used when translating to
LLVM IR in place of the outlined function's name.
This is not necessary anymore, so this patch removes all references to
this interface and uses of the `omp.outline_parent_name` discardable
attribute in tests.
If -nogpulib option is passed by the user, then the OpenMP device
runtime is not used and we should not emit globals to configure
debugging at compile-time for the device runtime.
Link to -nogpulib flag implementation for Clang:
https://reviews.llvm.org/D125314
Add an op in the OMP dialect to model the `target update` direcive. This
change reuses the `MapInfoOp` used by other device directive to model
`map` clauses but verifies that the restrictions imposed by the `target
update` directive are respected.
This patch removes the val field from the `MapInfoOp`.
Previously when lowering `TargetOp`, the bounds information for the
`BoxValues` were also being mapped. Instead these ops are now cloned
inside the target region to prevent mapping of non reference typed
values.
This patch adds the MLIR translation changes required for add the IsolatedFromAbove and OutlineableOpenMPOpInterface traits to omp.target. It links the newly added block arguments to their corresponding llvm values.
Depends on #67164.
Remove usage of getElementType in OpenMPTranslation to pave way for
switching to opaque pointers in MLIR and Flang. The approach chosen
stores the elementType in a new field in MapInfo called varType. A
similar approach was chosen for AtomicReadOp in
81767f52f4
This makes these match the behaviour of optional attributes (which are
omitted when they are their default value of none). This allows for
concise assembly formats without a custom printer.
An extra print of " " is also removed, this does change any existing
uses of oilists, but if the parameter before the oilist is optional,
that would previously add an extra space.
This #68694 + some fixes for the MLIR Python tests, unfortunately GitHub
does not allow re-opening PRs 😕
This makes these match the behaviour of optional attributes (which are
omitted when they are their default value of `none`). This allows for
concise assembly formats without a custom printer.
An extra print of " " is also removed, this does change any existing
uses of oilists, but if the parameter before the oilist is optional,
that would previously add an extra space.
This patch adds two new operations:
The first is the DataBoundsOp, which is based on OpenACC's DataBoundsOp,
which holds stride, index, extent, lower bound and upper bounds
which will be used in future follow up patches to perform initial
array sectioning of mapped arrays, and Fortran pointer and
allocatable mapping. Similarly to OpenACC, this new OpenMP
DataBoundsOp also comes with a new OpenMP type, which
helps to restrict operations to accepting only
DataBoundsOp as an input or output where necessary
(or other related operations that implement this type as
a return).
The patch also adds the MapInfoOp which rolls up some of
the old map information stored in target
operations into this new operation, and adds new
information that will be utilised in the lowering of mapped
variables, e.g. the aforementioned DataBoundsOp, but also a
new ByCapture OpenMP MLIR attribute, and isImplicit boolean
attribute. Both the ByCapture and isImplicit arguments will
affect the lowering from the OpenMP dialect to LLVM-IR in
minor but important ways, such as shifting the final maptype
or generating different load/store combinations to maintain
semantics with the OpenMP standard and alignment with the
current Clang OpenMP output as best as possible.
This MapInfoOp operation is slightly based on OpenACC's
DataEntryOp, the main difference other than some slightly
different fields (e,g, isImplicit/MapType/ByCapture) is that
OpenACC's data operations "inherit" (the MLIR ODS
equivalent) from this operation, whereas in OpenMP operations
that utilise MapInfoOp's are composed of/contain them.
A series of these MapInfoOp (one per map clause list item) is
now held by target operations that represent OpenMP
directives that utilise map clauses, e.g. TargetOp. MapInfoOp's
do not have their own specialised lowering to LLVM-IR, instead
the lowering is dependent on the particular container of the
MapInfoOp's, e.g. TargetOp has its own lowering to LLVM-IR
which utilised the information stored inside of MapInfoOp's to
affect it's lowering and the end result of the LLVM-IR generated,
which in turn can differ for host and device.
This patch contains these operations, minor changes to the
printing and parsing to support them, changes to tests (only
those relevant to this segment of the patch, other test
additions and changes are in other dependent
patches in this series) and some alterations to the OpenMPToLLVM
rewriter to support the new OpenMP type and operations.
This patch is one in a series that are dependent on each
other:
https://reviews.llvm.org/D158734https://reviews.llvm.org/D158735https://reviews.llvm.org/D158737
Reviewers: kiranchandramohan, TIFitis, razvanlupusoru
Differential Revision: https://reviews.llvm.org/D158732
The OpenACC standard specifies an `atomic` construct in section 2.12 (of
3.3 spec), used to ensure that a specific location is accessed or
updated atomically. Four different clauses are allowed: `read`, `write`,
`update`, or `capture`. If no clause appears, it is as if `update` is
used.
The OpenMP specification defines the same clauses for `omp atomic`. The
types of expression and the clauses in the OpenACC spec match the OpenMP
spec exactly. The main difference is that the OpenMP specification is a
superset - it includes clauses for `hint` and `memory order`. It also
allows conditional expression statements. But otherwise, the expression
definition matches.
Thus, for OpenACC, we refactor and reuse the OpenMP implementation as
follows:
* The atomic operations are duplicated in OpenACC dialect. This is
preferable so that each language's semantics are precisely represented
even if specs have divergence.
* However, since semantics overlap, a common interface between the
atomic operations is being added. The semantics for the interfaces are
not generic enough to be used outside of OpenACC and OpenMP, and thus
new folders were added to hold common pieces of the two dialects.
* The atomic interfaces define common accessors (such as getting `x` or
`v`) which match the OpenMP and OpenACC specs. It also adds common
verifiers intended to be called by each dialect's operation verifier.
* The OpenMP write operation was updated to use `x` and `expr` to be
consistent with its other operations (that use naming based on spec).
The frontend lowering necessary to generate the dialect can also be
reused. This will be done in a follow up change.
This renaming started with the native ODS support for properties, this is completing it.
A mass automated textual rename seems safe for most codebases.
Drop also the ods prefix to keep the accessors the same as they were before
this change:
properties.odsOperandSegmentSizes
reverts back to:
properties.operandSegementSizes
The ODS prefix was creating divergence between all the places and make it harder to
be consistent.
Reviewed By: jpienaar
Differential Revision: https://reviews.llvm.org/D157173
This patch permits map operands to be not specified for the target
data operation. Also emit an error if none of the map, use_device_addr,
or use_device_ptr operands are specified.
Reviewed By: TIFitis
Differential Revision: https://reviews.llvm.org/D156170
This patch implements an early outlining transform of omp.target operations in
flang. The pass is needed because optimizations may cross target op region
boundaries, but with the outlining the resulting functions only contain a
single omp.target op plus a func.return, so there should not be any opportunity
to optimize across region boundaries.
The patch also adds an interface to be able to store and retrieve the parent
function name of the original target operation. This is needed to be able to
create correct kernel function names when lowering to LLVM-IR.
Reviewed By: kiranchandramohan, domada
Differential Revision: https://reviews.llvm.org/D154879
Minor reordering of clauses in the assembly format for Target Data op to make it closer to the OpenMP standard.
Differential Revision: https://reviews.llvm.org/D152822
This patch introduces an MLIR attribute to the OpenMP dialect
representing the clauses that a 'requires' directive can define.
The `OffloadModuleInterface` is also updated to provide methods to get
and set a new dialect attribute `omp.requires`, to allow storing and using this
information during the lowering stages to LLVM IR.
Differential Revision: https://reviews.llvm.org/D147214
This reverts commit aa6b47cdaf3cddc70b7af33c1edbda502ecb3d05.
And adds a fix (adding missing libraries
to CMakeLists.txt for the OpenMPDialect)
that allows failing builds to succeed.
This attribute represents the OpenMP declare target directive, it marks a function
or global as declare target by being present but also contains information on
the device_type and capture clause (link or to). It being an attribute allows it to
mark existing constructs and be converted trivially on lowering from the OpenMP
dialect to MLIR using amendOperation.
An interface has been made for the declare target attribute, with several helper
methods for managing the attribute, this interface can be applied to MLIR
operations that are allowed to be marked as declare target (as an example, it
is by default applied to func.func, LLVMFunc, fir.GlobalOps and LLVMGlobalOps).
Reviewers: kiranchandramohan
Differential Revision: https://reviews.llvm.org/D150328
The intent of OpenMP version attribute is to be applied to a module
and then hold information on OpenMP version flag.
Information about OpenMP version can be specified in Clang
by flag -fopenmp-version and it is stored in LLVM-IR module metadata:
!llvm.module.flags = !{!0, !1}
!0 = !{i32 7, !"openmp", i32 51}
!1 = !{i32 7, !"openmp-device", i32 51}
OpenMP MLIR version attributes will allow to lower Flang frontend flag
to OpenMP MLIR code and then to LLVM IR.
There are two OpenMP MLIR version attributes. The first one "omp.version"
MLIR attribute corresponds to host OpenMP version. The second one
corresponds to "openmp-device" LLVM-IR metadata. "openmp-device" LLVM-IR
metadata is attached only for offloaded code.
Differential Revision: https://reviews.llvm.org/D150351
Reviewed By: kiranchandramohan
This patch prevents constant operations defined inside `omp.target` from being
hoisted out and into their parent functions by canonicalization passes.
Differential Revision: https://reviews.llvm.org/D148349
The intent of this attribute is for it to be applied to a module and
then hold information on runtime library (RTL) flags given to
Flang (or other OpenMP frontend) that should be lowered down to
LLVM-IR for devices as LLVM globals. The following related
flags are:
-fopenmp-target-debug
-fopenmp-assume-threads-oversubscription
-fopenmp-assume-teams-oversubscription
-fopenmp-assume-no-nested-parallelism
-fopenmp-assume-no-thread-state
These exist within Clang and are lowered into the IR when
offloading for device. This attribute allows this infromation
to be carried down from the Flang frontend to the
LLVM/OpenMP Dialect to LLVM-IR translation phase
and then be lowered to LLVM-IR.
Reviewers: kiranchandramohan
Differential Revision: https://reviews.llvm.org/D144896
This patch adds support for the OpenMP 4.0 depend clause (in, out,
inout) of the task construct to the definition of the OpenMP MLIR
dialect and translation from MLIR to LLVM IR using OMPIRBuilder.
Reviewed By: kiranchandramohan
Differential Revision: https://reviews.llvm.org/D142730
The current OpenMP implementation assumes the use of typed pointers (or rather typed pointer like types). Given the support for typed pointers in LLVM is now pending removal, the OpenMP Dialect should be able to support opaque pointers as well, given that any users of it must lower OpenMP through the LLVM Dialect.
This patch fixes the above and adds support for using LLVM opaque pointers with the OpenMP dialect. This is implemented by making all related code not make use of the element type of pointer arguments. The few (one) op requiring a pointer element type now use an explicit `TypeAttr` for representing the element type.
More concretely, the list of changes are:
* `omp.atomic.read` now has an extra `TypeAttr` (also in syntax) which is the element type of the values read and stored from the operands
* `omp.reduction` now has an type argument in the syntax for both the accmulator and operand since the operand type can no longer be inferred from the accumulator
* `OpenMPToLLVMIRTranslation.cpp` was rewritten to never query element types of pointers
* Adjusted the verifier to be able to handle pointers without element types
Differential Revision: https://reviews.llvm.org/D143582
This revision adapts the printers and parsers of the LLVM Dialect
AtomicRMWOp, AtomicCmpXchgOp, CallOp, and InvokeOp to support both
opaque and typed pointers by printing the pointer types explicitly.
Previously, the printers and parser of these operations silently assumed
typed pointers. This assumption is problematic if a lowering or the
LLVM IR import produce LLVM Dialect with opaque pointers and the IR is
then printed and parsed, for example, when running mlir-translate. In
LLVM IR itself all tests with typed pointers are already gone. It is
thus important to start switching to opaque pointers.
This revision can be seen as a preparation step for the switch of the
LLVM Dialect to opaque pointers. Once printing and parsing works
seamlessly, all lowerings to LLVM Dialect can be switched to produce
opaque pointers. After a transition period, LLVM Dialect itself can by
simplified to support opaque pointers only.
Reviewed By: ftynse, Dinistro
Differential Revision: https://reviews.llvm.org/D142884
This includes a basic implementation for the OpenMP 5.1 Target Data, Target Exit Data and Target Enter Data constructs
operation.
TODO:
- Depend clause support for Target Enter and Exit Data.
- Mapper and Iterator value support for Map Type Modifiers.
- Verifier for the operations.
Co-authored-by: abidmalikwaterloo <amalik@bnl.gov>
Co-authored-by: raghavendra <Raghu.Maddhipatla@amd.com>
Differential Revision: https://reviews.llvm.org/D131915
simd nontemporal construct is represented as a list of variables
which have low locality accross simd iterations
Added verifier of nontemporal clause. MLIR tests were updated to test
correctness of MLIR definition of nontemporal clause.
Differential Revision: https://reviews.llvm.org/D140553
Reviewed By: kiranchandramohan
This new option is set to `false` by default. It should be set only in Canonicalizer tests to detect faulty canonicalization patterns. I.e., patterns that prevent the canonicalizer from converging. The canonicalizer should always convergence on such small unit tests that we have in `canonicalize.mlir`.
Two faulty canonicalization patterns were detected and fixed with this change.
Differential Revision: https://reviews.llvm.org/D140873
This is generated by running
```
sed --in-place 's/[[:space:]]\+$//' mlir/**/*.td
sed --in-place 's/[[:space:]]\+$//' mlir/**/*.mlir
```
Reviewed By: rriddle, dcaballe
Differential Revision: https://reviews.llvm.org/D138866
This is required for D126305 code to propagate fastmath attributes
for Arith operations that are converted to LLVM IR intrinsics
operations.
LLVM IR intrinsic operations are using custom assembly format now
to avoid printing {fastmathFlags = #llvm.fastmath<none>}, which
is too verbose.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D136225
simd aligned construct is represented as pair of variable which needs
to be aligned and corresponding alignment value.
Added parser, printer and verifier of aligned clause. MLIR tests were
updated to test correctness of MLIR definition of aligned clause.
Differential Revision: https://reviews.llvm.org/D135865
Reviewed By: kiranchandramohan
The OpenMP standard specifies the region inside to be a structured
block. This means that branches into or out of the region is not
permitted but branches are permitted inside the OpenMP single region.
This is currently modeled by having a region of kind `AnyRegion` for
all the OpenMP ops. In future we can have a more strictly defined
region.
Fixes#58396.
Reviewed By: peixin
Differential Revision: https://reviews.llvm.org/D137023
