wehub-resource-sync
62 lines
2.4 KiB
C++
62 lines
2.4 KiB
C++
/* ******************************************************************************
|
||
*
|
||
*
|
||
* This program and the accompanying materials are made available under the
|
||
* terms of the Apache License, Version 2.0 which is available at
|
||
* https://www.apache.org/licenses/LICENSE-2.0.
|
||
*
|
||
* See the NOTICE file distributed with this work for additional
|
||
* information regarding copyright ownership.
|
||
* 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.
|
||
*
|
||
* SPDX-License-Identifier: Apache-2.0
|
||
******************************************************************************/
|
||
|
||
//
|
||
// @author raver119@gmail.com
|
||
//
|
||
|
||
#ifndef LIBND4J_REDUCTIONLOOPS_LONG_CPP
|
||
#define LIBND4J_REDUCTIONLOOPS_LONG_CPP
|
||
|
||
#include "ReductionLoops.hpp"
|
||
|
||
using namespace simdOps;
|
||
|
||
#include <types/types.h>
|
||
|
||
namespace sd {
|
||
|
||
template <typename X, typename Z>
|
||
template <typename OpType>
|
||
void ReductionLongLoops<X, Z>::innerloopReduce(sd::memory::Workspace* workspace, const X* x,
|
||
const sd::LongType* xShapeInfo, Z* z, const sd::LongType* zShapeInfo,
|
||
const LongType* dims, X* extraParams) {
|
||
#ifndef SD_LOOPS_INLINED
|
||
ReductionLoops<X, Z, X>::template loopReduce<OpType>(workspace, x, xShapeInfo, z, zShapeInfo, dims, extraParams);
|
||
#endif
|
||
}
|
||
|
||
template <typename X, typename Z>
|
||
void ReductionLongLoops<X, Z>::wrapper(int opNum, sd::memory::Workspace* workspace, const X* x,
|
||
const sd::LongType* xShapeInfo, Z* z, const sd::LongType* zShapeInfo,
|
||
const LongType* dims, X* extraParams) {
|
||
#ifndef SD_LOOPS_INLINED
|
||
DISPATCH_BY_OPNUM_TT(innerloopReduce, PARAMS(workspace, x, xShapeInfo, z, zShapeInfo, dims, extraParams),
|
||
REDUCE_LONG_OPS);
|
||
#endif
|
||
}
|
||
|
||
// Template instantiations for ReductionLongLoops are generated by TemplateProcessing.cmake
|
||
// This includes both class instantiations and member function templates (innerloopReduce<OpType>)
|
||
// for all type combinations in SD_COMMON_TYPES × SD_LONG_TYPES.
|
||
// No manual BUILD_DOUBLE_TEMPLATE needed here to avoid duplicate instantiation errors.
|
||
|
||
} // namespace sd
|
||
|
||
#endif // LIBND4J_REDUCTIONLOOPS_LONG_CPP
|