wehub-resource-sync
57 lines
1.9 KiB
Plaintext
57 lines
1.9 KiB
Plaintext
// Copyright (c) 2023 PaddlePaddle 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 "paddle/phi/kernels/reduce_max_kernel.h"
|
|
#include "paddle/phi/core/kernel_registry.h"
|
|
#include "paddle/phi/kernels/gpu/reduce.h"
|
|
|
|
namespace phi {
|
|
|
|
template <typename T, typename Context>
|
|
PADDLE_API void MaxRawKernel(const Context& dev_ctx,
|
|
const DenseTensor& x,
|
|
const IntArray& dims,
|
|
bool keep_dim,
|
|
bool reduce_all,
|
|
DenseTensor* out) {
|
|
reduce_all = recompute_reduce_all(x, dims, reduce_all);
|
|
auto out_dtype = x.dtype();
|
|
#ifdef PADDLE_WITH_XPU_KP
|
|
Reduce<T, kps::MaxFunctor, kps::IdentityFunctor>(
|
|
dev_ctx, x, reduce_all, dims.GetData(), keep_dim, out_dtype, out);
|
|
#else
|
|
Reduce<T, kps::MaxOps>(
|
|
dev_ctx, x, reduce_all, dims.GetData(), out_dtype, out);
|
|
#endif
|
|
}
|
|
|
|
} // namespace phi
|
|
|
|
#ifdef PADDLE_WITH_XPU_KP
|
|
PD_REGISTER_KERNEL(max_raw, KPS, ALL_LAYOUT, phi::MaxRawKernel, float) {}
|
|
#else
|
|
PD_REGISTER_KERNEL(max_raw,
|
|
KPS,
|
|
ALL_LAYOUT,
|
|
phi::MaxRawKernel,
|
|
float,
|
|
double,
|
|
int,
|
|
int64_t,
|
|
phi::float16,
|
|
phi::bfloat16,
|
|
phi::float8_e4m3fn,
|
|
phi::float8_e5m2) {}
|
|
#endif
|