* Copyright (c) 2026 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 <iostream>
#include <memory>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_glu_backward.h"
#define CHECK_RET(cond, return_expr) \
do { \
if (!(cond)) { \
return_expr; \
} \
} while (0)
#define LOG_PRINT(message, ...) \
do { \
printf(message, ##__VA_ARGS__); \
} while (0)
struct TensorDeleter {
void operator()(aclTensor* t) const {
if (t) aclDestroyTensor(t);
}
};
struct MemDeleter {
void operator()(void* p) const {
if (p) aclrtFree(p);
}
};
struct StreamDeleter {
void operator()(aclrtStream s) const {
if (s) aclrtDestroyStream(s);
}
};
using AclTensorPtr = std::unique_ptr<aclTensor, TensorDeleter>;
using AclMemPtr = std::unique_ptr<void, MemDeleter>;
using AclStreamPtr = std::unique_ptr<std::remove_pointer_t<aclrtStream>, StreamDeleter>;
int64_t GetShapeSize(const std::vector<int64_t>& shape) {
int64_t shape_size = 1;
for (auto i : shape) {
shape_size *= i;
}
return shape_size;
}
int Init(int32_t deviceId, AclStreamPtr& stream) {
auto ret = aclInit(nullptr);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclInit failed. ERROR: %d\n", ret); return ret);
ret = aclrtSetDevice(deviceId);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSetDevice failed. ERROR: %d\n", ret); return ret);
aclrtStream rawStream = nullptr;
ret = aclrtCreateStream(&rawStream);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtCreateStream failed. ERROR: %d\n", ret); return ret);
stream.reset(rawStream);
return 0;
}
template <typename T>
int CreateAclTensor(const std::vector<T>& hostData, const std::vector<int64_t>& shape, AclMemPtr& deviceAddr,
aclDataType dataType, AclTensorPtr& tensor) {
auto size = GetShapeSize(shape) * sizeof(T);
void* rawAddr = nullptr;
auto ret = aclrtMalloc(&rawAddr, size, ACL_MEM_MALLOC_HUGE_FIRST);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtMalloc failed. ERROR: %d\n", ret); return ret);
deviceAddr.reset(rawAddr);
ret = aclrtMemcpy(rawAddr, size, hostData.data(), size, ACL_MEMCPY_HOST_TO_DEVICE);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtMemcpy failed. ERROR: %d\n", ret); return ret);
std::vector<int64_t> strides(shape.size(), 1);
for (int64_t i = shape.size() - 2; i >= 0; i--) {
strides[i] = shape[i + 1] * strides[i + 1];
}
auto* rawTensor = aclCreateTensor(shape.data(), shape.size(), dataType, strides.data(), 0, aclFormat::ACL_FORMAT_ND,
shape.data(), shape.size(), rawAddr);
tensor.reset(rawTensor);
return 0;
}
int main() {
int32_t deviceId = 0;
AclStreamPtr stream;
auto ret = Init(deviceId, stream);
CHECK_RET(ret == 0, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);
std::vector<int64_t> selfShape = {2, 4, 6};
std::vector<int64_t> gradOutShape = {1, 4, 6};
std::vector<int64_t> outShape = {2, 4, 6};
AclMemPtr selfDeviceAddr;
AclMemPtr gradOutDeviceAddr;
AclMemPtr outDeviceAddr;
AclTensorPtr self;
AclTensorPtr gradOut;
AclTensorPtr out;
void* workspaceAddr = nullptr;
uint64_t workspaceSize = 0;
std::vector<float> selfHostData = {
1.4021, -2.2242, -0.2267, 1.0049, -0.9057, -1.2181,
1.4031, -0.1750, -0.2705, -1.3884, 0.2565, -0.7543,
-0.3368, 0.3036, 0.4370, 1.9198, 0.0974, 0.9725,
-1.7963, -1.9863, 2.2742, 1.0436, 1.6882, 0.7845,
0.5129, 0.7107, -0.0894, -1.7567, 1.5542, 1.5608,
-1.0318, -0.5742, 0.1330, -1.4514, -2.5802, -0.4738,
0.2548, -1.7638, 0.8152, 0.5531, 0.1251, 0.8516,
-0.0048, -0.9011, -0.4680, 0.2906, -0.0880, 0.3975
};
std::vector<float> gradOutHostData = {
1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0
};
std::vector<float> outHostData(48, 0);
int64_t dim = 0;
aclOpExecutor* executor = nullptr;
auto size = GetShapeSize(outShape);
std::vector<float> resultData(size, 0);
ret = CreateAclTensor(selfHostData, selfShape, selfDeviceAddr, aclDataType::ACL_FLOAT, self);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("CreateAclTensor self failed. ERROR: %d\n", ret); return ret);
ret = CreateAclTensor(gradOutHostData, gradOutShape, gradOutDeviceAddr, aclDataType::ACL_FLOAT, gradOut);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("CreateAclTensor gradOut failed. ERROR: %d\n", ret); return ret);
ret = CreateAclTensor(outHostData, outShape, outDeviceAddr, aclDataType::ACL_FLOAT, out);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("CreateAclTensor out failed. ERROR: %d\n", ret); return ret);
ret = aclnnGluBackwardGetWorkspaceSize(gradOut.get(), self.get(), dim, out.get(), &workspaceSize, &executor);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnGluBackwardGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);
AclMemPtr workspaceGuard;
if (workspaceSize > 0) {
void* rawWorkspace = nullptr;
ret = aclrtMalloc(&rawWorkspace, workspaceSize, ACL_MEM_MALLOC_HUGE_FIRST);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("allocate workspace failed. ERROR: %d\n", ret); return ret);
workspaceAddr = rawWorkspace;
workspaceGuard.reset(rawWorkspace);
}
ret = aclnnGluBackward(workspaceAddr, workspaceSize, executor, stream.get());
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnGluBackward failed. ERROR: %d\n", ret); return ret);
ret = aclrtSynchronizeStream(stream.get());
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSynchronizeStream failed. ERROR: %d\n", ret); return ret);
ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), outDeviceAddr.get(),
size * sizeof(float), ACL_MEMCPY_DEVICE_TO_HOST);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("copy result from device to host failed. ERROR: %d\n", ret); return ret);
for (int64_t i = 0; i < size; i++) {
LOG_PRINT("result[%ld] is: %f\n", i, resultData[i]);
}
aclrtResetDevice(deviceId);
aclFinalize();
return ret;
}