* Copyright (c) 2025 Huawei Technologies Co., Ltd.
* This program is free software, you can redistribute it and/or modify it under the terms and conditions of
* CANN Open Software License Agreement Version 2.0 (the "License").
* Please refer to the License for details. You may not use this file except in compliance with the License.
* THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
* INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
* See LICENSE in the root of the software repository for the full text of the License.
*/
#include "invert_permutation_aicpu.h"
#include "cpu_kernel_utils.h"
#include "cpu_types.h"
#include "log.h"
#include "status.h"
#include "utils/kernel_util.h"
namespace {
const uint32_t kOutputNum = 1;
const uint32_t kInputNum = 1;
const char *const kInvertPermutation = "InvertPermutation";
const int64_t kParallelDataNums = 64 * 1024;
}
namespace aicpu {
uint32_t InvertPermutation::Compute(CpuKernelContext &ctx) {
KERNEL_HANDLE_ERROR(
NormalCheck(ctx, kInputNum, kOutputNum),
"InvertPermutation check input and output number failed.");
Tensor *x = ctx.Input(0);
int32_t dims = x->GetTensorShape()->GetDims();
KERNEL_CHECK_FALSE((dims == 1), KERNEL_STATUS_PARAM_INVALID,
"InvertPermutation input dim need 1D, but get %d.", dims);
int64_t num = x->NumElements();
KERNEL_CHECK_FALSE((num >= 0), KERNEL_STATUS_INNER_ERROR,
"InvertPermutation get the num of input elements faild.");
auto data_type = x->GetDataType();
switch (data_type) {
case (DT_INT32):
return InvertPermutationCompute<int32_t>(x, num, ctx);
case (DT_INT64):
return InvertPermutationCompute<int64_t>(x, num, ctx);
default:
KERNEL_LOG_ERROR("InvertPermutation kernel data type [%s] not support.",
DTypeStr(data_type).c_str());
return KERNEL_STATUS_PARAM_INVALID;
}
}
template <typename T>
uint32_t InvertPermutation::InvertPermutationCompute(Tensor *x, int64_t num,
CpuKernelContext &ctx) {
T *x_addrs = reinterpret_cast<T *>(x->GetData());
T *y_addrs = reinterpret_cast<T *>(ctx.Output(0)->GetData());
size_t total = static_cast<uint64_t>(num);
for (size_t i = 0; i < total; i++) {
y_addrs[i] = -1;
}
if (total < kParallelDataNums) {
for (size_t i = 0; i < total; i++) {
const T data_ith = x_addrs[i];
KERNEL_CHECK_FALSE((data_ith >= 0 && data_ith < static_cast<T>(total)), KERNEL_STATUS_PARAM_INVALID, "%ld is not between 0 and %zu.", static_cast<int64_t>(data_ith), total);
KERNEL_CHECK_FALSE((y_addrs[data_ith] == -1), KERNEL_STATUS_PARAM_INVALID, "%ld is duplicated in the input.", static_cast<int64_t>(data_ith));
y_addrs[data_ith] = i;
}
} else {
int64_t max_core_num = 1;
int64_t cpu_num = aicpu::CpuKernelUtils::GetCPUNum(ctx) - kResvCpuNum;
max_core_num = std::max(max_core_num, cpu_num);
auto shard_invert_permutation = [&x_addrs, &y_addrs, &num](size_t start, size_t end) {
for (size_t i = start; i < end; i++) {
const T data_ith = x_addrs[i];
KERNEL_CHECK_FALSE((data_ith >= 0 && data_ith < static_cast<T>(num)), KERNEL_STATUS_PARAM_INVALID, "%ld is not between 0 and %ld.", static_cast<int64_t>(data_ith), num);
KERNEL_CHECK_FALSE((y_addrs[data_ith] == -1), KERNEL_STATUS_PARAM_INVALID, "%ld is duplicated in the input.", static_cast<int64_t>(data_ith));
y_addrs[data_ith] = i;
}
return KERNEL_STATUS_OK;
};
KERNEL_HANDLE_ERROR(
CpuKernelUtils::ParallelFor(ctx, num, num / max_core_num,
shard_invert_permutation),
"InvertPermutation Compute failed.");
}
return KERNEL_STATUS_OK;
}
REGISTER_CPU_KERNEL(kInvertPermutation, InvertPermutation);
}