Files
wehub-resource-sync 8a852e4b4e
cffconvert / validate (push) Has been skipped
License Check / license-check (push) Failing after 2s
chore: import upstream snapshot with attribution
2026-07-13 12:14:16 +08:00

78 lines
2.6 KiB
C++

/* Copyright 2022 The TensorFlow Authors. All Rights Reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
==============================================================================*/
#include "tensorflow/dtensor/mlir/sparse_expander.h"
#include <memory>
#include <string>
#include <utility>
#include "absl/container/flat_hash_map.h"
#include "absl/log/check.h"
#include "absl/log/log.h"
#include "absl/status/status.h"
#include "mlir/IR/Operation.h" // from @llvm-project
#include "tensorflow/core/framework/registration/registration.h"
#include "tensorflow/core/platform/status.h"
#include "tensorflow/dtensor/mlir/op_utils.h"
#include "tensorflow/dtensor/mlir/sparse_expander_common.h"
namespace tensorflow {
namespace dtensor {
// static
SparseExpanderRegistry* SparseExpanderRegistry::Global() {
static SparseExpanderRegistry* registry = new SparseExpanderRegistry();
return registry;
}
SparseExpanderBase* SparseExpanderRegistry::GetSparseExpansionFnForOp(
mlir::Operation* op) {
auto key = OpName(op);
auto fn = op_to_sparse_expansion_fn_map_.find(key);
if (fn == op_to_sparse_expansion_fn_map_.end()) return nullptr;
return fn->second.get();
}
InitOnStartupMarker SparseExpanderRegistry::RegisterSparseExpansionFn(
std::string opName, std::unique_ptr<SparseExpanderBase> prop) {
CHECK(op_to_sparse_expansion_fn_map_ // Crash ok
.insert_or_assign(opName, std::move(prop))
.second);
return {};
}
absl::Status RunSparseExpansion(mlir::Operation* op, mlir::Operation** output) {
// Only expand if there are any SparseTensor inputs.
if (HasAnySparseInput(op)) {
SparseExpanderBase* expander =
SparseExpanderRegistry::Global()->GetSparseExpansionFnForOp(op);
if (expander != nullptr) {
auto expanded_op = expander->ExpandOp(op);
if (expanded_op.ok()) *output = expanded_op.value();
return expanded_op.status();
} else {
VLOG(1) << "No sparse expansion found for " << OpName(op) << "\n";
*output = op;
}
} else { // If there is no SparseTensor inputs then just return the op.
*output = op;
}
return absl::OkStatus();
}
} // namespace dtensor
} // namespace tensorflow