This is the second part loosely extracted from D71179 and cleaned up.
This patch provides semantic analysis support for `omp begin/end declare
variant`, mostly as defined in OpenMP technical report 8 (TR8) [0].
The sema handling makes code generation obsolete as we generate "the
right" calls that can just be handled as usual. This handling also
applies to the existing, albeit problematic, `omp declare variant
support`. As a consequence a lot of unneeded code generation and
complexity is removed.
A major purpose of this patch is to provide proper `math.h`/`cmath`
support for OpenMP target offloading. See PR42061, PR42798, PR42799. The
current code was developed with this feature in mind, see [1].
The logic is as follows:
If we have seen a `#pragma omp begin declare variant match(<SELECTOR>)`
but not the corresponding `end declare variant`, and we find a function
definition we will:
1) Create a function declaration for the definition we were about to generate.
2) Create a function definition but with a mangled name (according to
`<SELECTOR>`).
3) Annotate the declaration with the `OMPDeclareVariantAttr`, the same
one used already for `omp declare variant`, using and the mangled
function definition as specialization for the context defined by
`<SELECTOR>`.
When a call is created we inspect it. If the target has an
`OMPDeclareVariantAttr` attribute we try to specialize the call. To this
end, all variants are checked, the best applicable one is picked and a
new call to the specialization is created. The new call is used instead
of the original one to the base function. To keep the AST printing and
tooling possible we utilize the PseudoObjectExpr. The original call is
the syntactic expression, the specialized call is the semantic
expression.
[0] https://www.openmp.org/wp-content/uploads/openmp-TR8.pdf
[1] https://reviews.llvm.org/D61399#change-496lQkg0mhRN
Reviewers: kiranchandramohan, ABataev, RaviNarayanaswamy, gtbercea, grokos, sdmitriev, JonChesterfield, hfinkel, fghanim, aaron.ballman
Subscribers: bollu, guansong, openmp-commits, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D75779
If the ancestor device modifier is used and the value of the device
clause is evaluated to 1, the ancestor device shall be used for the
execution.
Since the reverse offloading is not supported yet, the target construct
execution is always initiated from the host, not from the device. So, if
the ancestor modifier is specified, just execute target region on the
host.
Added codegen for update clause in depobj. Reads the number of the
elements from the first element and updates flags for each element in
the loop.
```
omp_depend_t x;
kmp_depend_info *base = (kmp_depend_info *)x;
intptr_t num = x[-1].base_addr;
kmp_depend_info *end = x + num;
kmp_depend_info *el = base;
do {
el.flags = new_flag;
el = &el[1];
} while (el != end);
```
Added codegen for 'depend' clause in depobj directive. The depend clause
is emitted as kmp_depend_info <deps>[<number_of_items_in_clause> + 1]. The
first element in this array is reserved for storing the number of
elements in this array: <deps>[0].base_addr =
<number_of_items_in_clause>;
This extra element is required to implement 'update' and 'destroy'
clauses. It is required to know the size of array to destroy it
correctly and to update depency kind.
directive.
According to OpenMP 5.0, The atomic_default_mem_order clause specifies the default memory ordering behavior for atomic constructs that must be provided by an implementation. If the default memory ordering is specified as seq_cst, all atomic constructs on which memory-order-clause is not specified behave as if the seq_cst clause appears. If the default memory ordering is specified as relaxed, all atomic constructs on which memory-order-clause is not specified behave as if the relaxed clause appears.
If the default memory ordering is specified as acq_rel, atomic constructs on which memory-order-clause is not specified behave as if the release clause appears if the atomic write or atomic update operation is specified, as if the acquire clause appears if the atomic read operation is specified, and as if the acq_rel clause appears if the atomic captured update operation is specified.
regions.
If the lastprivate conditional is passed as shared in inner region, we
shall check if it was ever changed and use this updated value after exit
from the inner region as an update value.
regions with reductions, lastprivates or linears clauses.
If the lastprivate conditional variable is updated in inner parallel
region with reduction, lastprivate or linear clause, the value must be
considred as a candidate for lastprivate conditional. Also, tracking in
inner parallel regions is not required.
Use canonical decls instead of mangled names in the set of already
emitted decls. This allows to reduce the number of function calls for
getting declarations mangled names and speedup the compilation.
Added codegen support for lastprivate conditional. According to the
standard, if when the conditional modifier appears on the clause, if an
assignment to a list item is encountered in the construct then the
original list item is assigned the value that is assigned to the new
list item in the sequentially last iteration or lexically last section
in which such an assignment is encountered.
We look for the assignment operations and check if the left side
references lastprivate conditional variable. Then the next code is
emitted:
if (last_iv_a <= iv) {
last_iv_a = iv;
last_a = lp_a;
}
At the end the implicit barrier is generated to wait for the end of all
threads and then in the check for the last iteration the private copy is
assigned the last value.
if (last_iter) {
lp_a = last_a; // <--- new code
a = lp_a; // <--- store of private value to the original variable.
}
This removes the OpenMPProcBindClauseKind enum in favor of
llvm::omp::ProcBindKind which lives in OpenMPConstants.h and was
introduced in D70109.
No change in behavior is expected.
Reviewed By: JonChesterfield
Differential Revision: https://reviews.llvm.org/D70289
Summary:
Basic codegen for the declarations marked as nontemporal. Also, if the
base declaration in the member expression is marked as nontemporal,
lvalue for member decl access inherits nonteporal flag from the base
lvalue.
Reviewers: rjmccall, hfinkel, jdoerfert
Subscribers: guansong, arphaman, caomhin, kkwli0, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D71708
According to OpenMP 5.0, if clause can be used in simd directive. If
condition in the if clause if false, the non-vectorized version of the
loop must be executed.
The final list of OpenMP offload targets becomes known only at the link time and since offload registration code depends on the targets list it makes sense to delay offload registration code generation to the link time instead of adding it to the host part of every fat object. This patch moves offload registration code generation from clang to the offload wrapper tool.
This is the last part of the OpenMP linker script elimination patch https://reviews.llvm.org/D64943
Differential Revision: https://reviews.llvm.org/D68746
llvm-svn: 374937
Runtime function __kmpc_push_tripcount better to call inside of the task
context for target regions. Otherwise, the libomptarget is unable to
link the provided tripcount value for nowait target regions and
completely looses this information.
llvm-svn: 372609
This patch implements the code generation for OpenMP 5.0 declare mapper
(user-defined mapper) constructs. For each declare mapper, a mapper
function is generated. These mapper functions will be called by the
runtime and/or other mapper functions to achieve user defined mapping.
The design slides can be found at
https://github.com/lingda-li/public-sharing/blob/master/mapper_runtime_design.pptx
Re-commit after revert in r367773 because r367755 changed the LLVM-IR
output such that a CHECK line failed.
Patch by Lingda Li <lildmh@gmail.com>
Differential Revision: https://reviews.llvm.org/D59474
llvm-svn: 367905
This patch implements the code generation for OpenMP 5.0 declare mapper
(user-defined mapper) constructs. For each declare mapper, a mapper
function is generated. These mapper functions will be called by the
runtime and/or other mapper functions to achieve user defined mapping.
The design slides can be found at
https://github.com/lingda-li/public-sharing/blob/master/mapper_runtime_design.pptx
Patch by Lingda Li <lildmh@gmail.com>
Differential Revision: https://reviews.llvm.org/D59474
llvm-svn: 367773
Summary:
This patch adds support for the handling of the variables under the declare target to clause.
The variables in this case are handled like link variables are. A pointer is created on the host and then mapped to the device. The runtime will then copy the address of the host variable in the device pointer.
Reviewers: ABataev, AlexEichenberger, caomhin
Reviewed By: ABataev
Subscribers: guansong, jdoerfert, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D63108
llvm-svn: 363959
Summary: This patch avoids the emission of maps for target link variables when unified memory is present.
Reviewers: ABataev, caomhin
Reviewed By: ABataev
Subscribers: guansong, jdoerfert, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D60883
llvm-svn: 363435
Summary:
This patch adds support for the registration of the requires directives with the runtime.
Each requires directive clause will enable a particular flag to be set.
The set of flags is passed to the runtime to be checked for compatibility with other such flags coming from other object files.
The registration function is called whenever OpenMP is present even if a requires directive is not present. This helps detect cases in which requires directives are used inconsistently.
Reviewers: ABataev, AlexEichenberger, caomhin
Reviewed By: ABataev, AlexEichenberger
Subscribers: jholewinski, guansong, jfb, jdoerfert, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D60568
llvm-svn: 361298
regions.
Added more complex analysis for number of teams and number of threads in
the target regions, also merged related common code between CGOpenMPRuntime
and CGOpenMPRuntimeNVPTX classes.
llvm-svn: 358126
For the global variables the allocate directive must specify only the
predefined allocator. This allocator must be translated into the correct
form of the address space for the targets that support different address
spaces.
llvm-svn: 356702
Added initial codegen for the local variables with the #pragma omp
allocate directive. Instead of allocating the variables on the stack,
__kmpc_alloc|__kmpc_free functions are used for memory (de-)allocation.
llvm-svn: 356472
memory.
If the variable with the constant non-scalar type is firstprivatized in
the target region, the local copy is created with the data copying.
Instead, we allocate the copy in the constant memory and avoid extra
copying in the outlined target regions. This global copy is used in the
target regions without loss of the performance.
llvm-svn: 355418
Emit{Nounwind,}RuntimeCall{,OrInvoke} have been modified to take a
FunctionCallee as an argument, and CreateRuntimeFunction has been
modified to return a FunctionCallee. All callers have been updated.
Additionally, CreateBuiltinFunction is removed, as it was redundant
with CreateRuntimeFunction after some previous changes.
Differential Revision: https://reviews.llvm.org/D57668
llvm-svn: 353184
to reflect the new license.
We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.
Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.
llvm-svn: 351636
Several headers would fail to compile if other headers were not previously
included. The usual issue is that a class is forward declared, but the
full definition is needed. The requirement for the definition is use of
isa/dyn_cast or calling functions of pointer-packed data types such as
DenseMap or PointerIntPair. Add missing includes to these headers.
SVals.h required an out-of-line method definition in the .cpp file to avoid
circular inclusion of headers with BasicValueFactory.h
llvm-svn: 350913
Each we create the target regions with the teams distribute inner
region, we can better estimate number of the teams required to execute
the target region. Function __kmpc_push_target_tripcount() is used for
purpose, which accepts device_id and the number of the iterations,
performed by the associated loop.
llvm-svn: 350571
__kmpc_barrier runtime functions must be marked as convergent to prevent
some dangerous optimizations. Also, for NVPTX target all barriers must
be emitted as simple barriers.
llvm-svn: 348271
For the NVPTX target default locations should be emitted as constants +
additional info must be emitted in the reserved_2 field of the ident_t
structure. The 1st bit controls the execution mode and the 2nd bit
controls use of the lightweight runtime. The combination of the bits for
Non-SPMD mode + lightweight runtime represents special undefined mode,
used outside of the target regions for orphaned directives or functions.
Should allow and additional optimization inside of the target regions.
llvm-svn: 347425
Fixed lookup for the target regions in unused virtual functions + fixed
processing of the global variables not marked as declare target but
emitted during debug info emission.
llvm-svn: 346343
Added support for mapping of lambdas in the target regions. It scans all
the captures by reference in the lambda, implicitly maps those variables
in the target region and then later reinstate the addresses of
references in lambda to the correct addresses of the captured|privatized
variables.
llvm-svn: 345609
Summary: This patch adds a new code generation path for bound sharing directives containing distribute parallel for. The new code generation scheme applies to chunked schedules on distribute and parallel for directives. The scheme simplifies the code that is being generated by eliminating the need for an outer for loop over chunks for both distribute and parallel for directives. In the case of distribute it applies to any sized chunk while in the parallel for case it only applies when chunk size is 1.
Reviewers: ABataev, caomhin
Reviewed By: ABataev
Subscribers: jholewinski, guansong, cfe-commits
Differential Revision: https://reviews.llvm.org/D53448
llvm-svn: 345509
Fixed emission of the __kmpc_global_thread_num() so that it is not
messed up with alloca instructions anymore. Plus, fixes emission of the
__kmpc_global_thread_num() functions in the target outlined regions so
that they are not called before runtime is initialized.
llvm-svn: 343856
Summary: Set default schedule for parallel for loops to schedule(static, 1) when using SPMD mode on the NVPTX device offloading toolchain to ensure coalescing.
Reviewers: ABataev, Hahnfeld, caomhin
Reviewed By: ABataev
Subscribers: jholewinski, guansong, cfe-commits
Differential Revision: https://reviews.llvm.org/D52629
llvm-svn: 343260
Johannes Doerfert