aclnnNLLLoss2dBackward
支持的产品型号
- Atlas 训练系列产品。
- Atlas A2训练系列产品/Atlas 800I A2推理产品。
接口原型
每个算子分为两段式接口,必须先调用“aclnnNLLLoss2dBackwardGetWorkspaceSize”接口获取计算所需workspace大小以及包含了算子计算流程的执行器,再调用“aclnnNLLLoss2dBackward”接口执行计算。
aclnnStatus aclnnNLLLoss2dBackwardGetWorkspaceSize(const aclTensor* gradOutput, const aclTensor* self, const aclTensor* target, const aclTensor* weight, int64_t reduction, int64_t ignoreIndex, aclTensor* totalWeight, aclTensor* out, uint64_t* workspaceSize, aclOpExecutor** executor)
aclnnStatus aclnnNLLLoss2dBackward(void *workspace, uint64_t workspaceSize, aclOpExecutor *executor, aclrtStream stream)
功能描述
算子功能:负对数似然损失反向。
aclnnNLLLoss2dBackwardGetWorkspaceSize
参数说明:
gradOutput(aclTensor*, 计算输入):Device侧的aclTensor,shape为三维(第一维是N)或者一维(且元素个数为1),数据类型支持FLOAT、FLOAT16、BFLOAT16(仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持)。支持非连续的Tensor,数据格式支持ND。
self(aclTensor*, 计算输入):Device侧的aclTensor,shape为四维,第一维是N表示batch size,第二维是C表示类别,数据类型支持FLOAT、FLOAT16、BFLOAT16(仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持)。支持非连续的Tensor,数据格式支持ND。要求self的第0维、第2维、第3维的shape分别与target的第0维、第1维、第2维的shape一致,否则返回false。
target(aclTensor*, 计算输入):Device侧的aclTensor,表示真实标签,shape为三维,第一维是N,其中每个元素的取值范围是[0, C - 1],数据类型支持INT64、UINT8。支持非连续的Tensor,数据格式支持ND。
weight(aclTensor*, 计算输入):Device侧的aclTensor,表示各个类别的权重,shape为(C, ),数据类型支持FLOAT、FLOAT16、BFLOAT16(仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持)。支持非连续的Tensor,数据格式支持ND。
reduction(int64_t, 计算输入):Host侧的int64_t,指定要应用到输出的缩减。支持 0('none') | 1('mean') | 2('sum')。'none' 表示不应用减少,'mean' 表示输出的总和将除以输出中的元素数,'sum' 表示输出将被求和。当reduction为0时,要求target的shape与gradOutput的shape一致,否则返回false。
ignoreIndex(int64_t, 计算输入):Host侧的int64_t,指定一个被忽略且不影响输入梯度的目标值。
totalWeight(aclTensor*, 计算输入):Device侧的aclTensor,数据类型支持FLOAT、FLOAT16、BFLOAT16(仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持),且数据类型为与weight相同,shape为(1,),数据格式支持ND。
out(aclTensor*, 计算输出):Device侧的aclTensor,shape与self相同,数据类型支持FLOAT、FLOAT16、BFLOAT16(仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持)。支持非连续的Tensor,数据格式支持ND。数据类型跟self一致。
workspaceSize(uint64_t*, 出参):返回需要在Device侧申请的workspace大小。
executor(aclOpExecutor**, 出参):返回op执行器,包含了算子计算流程。
返回值:
aclnnStatus:返回状态码,具体参见aclnn返回码。
161001 (ACLNN_ERR_PARAM_NULLPTR): 1. 传入的gradOutput、self、target、weight、totalWeight或out是空指针。
161002 (ACLNN_ERR_PARAM_INVALID): 1. gradOutput、self、target、weight、totalWeight或out的数据类型不在支持的范围之内。
2. target非3维tensor,self非4维tensor。
3. weight的元素个数不是C。
4. self的第0,2,3维的元素个数和target的第0,1,2维元素个数不相等。
5. totalWeight的元素个数不是1。
6. reduction是none时,gradOutput的维数不是3或者gradOutput的第0,1,2维的元素个数和target的第0,1,2维元素个数不相等。
7. reduction不是none时,gradOutput的维数大于1或者元素个数不为1。
8. reduction值不在0~2范围之内。
aclnnNLLLoss2dBackward
参数说明:
workspace(void*, 入参):在Device侧申请的workspace内存地址。
workspaceSize(uint64_t, 入参):在Device侧申请的workspace大小,由第一段接口aclnnNLLLoss2dBackwardGetWorkspaceSize获取。
executor(aclOpExecutor*, 入参):op执行器,包含了算子计算流程。
stream(aclrtStream, 入参):指定执行任务的 AscendCL Stream流。
返回值:
aclnnStatus:返回状态码,具体参见aclnn返回码。
约束与限制
无。
调用示例
示例代码如下,仅供参考,具体编译和执行过程请参考编译与运行样例。
#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_nll_loss2d_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)
int64_t GetShapeSize(const std::vector<int64_t>& shape) {
int64_t shapeSize = 1;
for (auto i : shape) {
shapeSize *= i;
}
return shapeSize;
}
int Init(int32_t deviceId, aclrtStream* stream) {
// 固定写法,AscendCL初始化
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);
ret = aclrtCreateStream(stream);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtCreateStream failed. ERROR: %d\n", ret); return ret);
return 0;
}
template <typename T>
int CreateAclTensor(const std::vector<T>& hostData, const std::vector<int64_t>& shape, void** deviceAddr,
aclDataType dataType, aclTensor** tensor) {
auto size = GetShapeSize(shape) * sizeof(T);
// 调用aclrtMalloc申请device侧内存
auto ret = aclrtMalloc(deviceAddr, size, ACL_MEM_MALLOC_HUGE_FIRST);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtMalloc failed. ERROR: %d\n", ret); return ret);
// 调用aclrtMemcpy将host侧数据拷贝到device侧内存上
ret = aclrtMemcpy(*deviceAddr, size, hostData.data(), size, ACL_MEMCPY_HOST_TO_DEVICE);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtMemcpy failed. ERROR: %d\n", ret); return ret);
// 计算连续tensor的strides
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];
}
// 调用aclCreateTensor接口创建aclTensor
*tensor = aclCreateTensor(shape.data(), shape.size(), dataType, strides.data(), 0, aclFormat::ACL_FORMAT_ND,
shape.data(), shape.size(), *deviceAddr);
return 0;
}
int main() {
// 1. (固定写法)device/stream初始化,参考AscendCL对外接口列表
// 根据自己的实际device填写deviceId
int32_t deviceId = 0;
aclrtStream stream;
auto ret = Init(deviceId, &stream);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);
// 2. 构造输入与输出,需要根据API的接口自定义构造
std::vector<int64_t> gradShape = {3, 1, 1};
std::vector<int64_t> selfShape = {3, 5, 1, 1};
std::vector<int64_t> targetShape = {3, 1, 1};
std::vector<int64_t> weightShape = {5};
std::vector<int64_t> totalWeightShape = {1};
std::vector<int64_t> outShape = {3, 5, 1, 1};
void* gradDeviceAddr = nullptr;
void* selfDeviceAddr = nullptr;
void* targetDeviceAddr = nullptr;
void* weightDeviceAddr = nullptr;
void* totalWeightDeviceAddr = nullptr;
void* outDeviceAddr = nullptr;
aclTensor* grad = nullptr;
aclTensor* self = nullptr;
aclTensor* target = nullptr;
aclTensor* weight = nullptr;
aclTensor* totalWeight = nullptr;
aclTensor* out = nullptr;
std::vector<float> gradHostData = {2.7, 2.6, 2.5};
std::vector<float> selfHostData = {4.1, 4.2, 4.3, 4.4, 4.5, 4.6, 4.7, 4.8, 4.9, 5.0, 5.1, 5.2, 5.3, 5.4, 5.5};
std::vector<int64_t> targetHostData = {2, 3, 1};
std::vector<float> weightHostData = {1.0, 1.0, 1.0, 1.0, 1.0};
std::vector<float> totalWeightHostData = {1.0};
std::vector<float> outHostData = {0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0};
int64_t reduction = 0;
int64_t ignoreIndex = -100;
// 创建grad aclTensor
ret = CreateAclTensor(gradHostData, gradShape, &gradDeviceAddr, aclDataType::ACL_FLOAT, &grad);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// 创建self aclTensor
ret = CreateAclTensor(selfHostData, selfShape, &selfDeviceAddr, aclDataType::ACL_FLOAT, &self);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// 创建target aclTensor
ret = CreateAclTensor(targetHostData, targetShape, &targetDeviceAddr, aclDataType::ACL_INT64, &target);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// 创建weight aclTensor
ret = CreateAclTensor(weightHostData, weightShape, &weightDeviceAddr, aclDataType::ACL_FLOAT, &weight);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// 创建totalWeight aclTensor
ret = CreateAclTensor(totalWeightHostData, totalWeightShape, &totalWeightDeviceAddr,
aclDataType::ACL_FLOAT, &totalWeight);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// 创建out aclTensor
ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_FLOAT, &out);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// 3. 调用CANN算子库API,需要修改为具体的Api名称
uint64_t workspaceSize = 0;
aclOpExecutor* executor;
// 调用aclnnNLLLoss2dBackward第一段接口
ret = aclnnNLLLoss2dBackwardGetWorkspaceSize(grad, self, target, weight, reduction, ignoreIndex, totalWeight, out,
&workspaceSize, &executor);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnNLLLoss2dBackwardGetWorkspaceSize failed. ERROR: %d\n", ret);
return ret);
// 根据第一段接口计算出的workspaceSize申请device内存
void* workspaceAddr = nullptr;
if (workspaceSize > 0) {
ret = aclrtMalloc(&workspaceAddr, workspaceSize, ACL_MEM_MALLOC_HUGE_FIRST);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("allocate workspace failed. ERROR: %d\n", ret); return ret);
}
// 调用aclnnNLLLoss2dBackward第二段接口
ret = aclnnNLLLoss2dBackward(workspaceAddr, workspaceSize, executor, stream);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnNLLLoss2dBackward failed. ERROR: %d\n", ret); return ret);
// 4. (固定写法)同步等待任务执行结束
ret = aclrtSynchronizeStream(stream);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSynchronizeStream failed. ERROR: %d\n", ret); return ret);
// 5. 获取输出的值,将device侧内存上的结果拷贝至host侧,需要根据具体API的接口定义修改
auto size = GetShapeSize(outShape);
std::vector<float> resultData(size, 0);
ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), outDeviceAddr,
size * sizeof(resultData[0]), 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]);
}
// 6. 释放aclTensor,需要根据具体API的接口定义修改
aclDestroyTensor(grad);
aclDestroyTensor(self);
aclDestroyTensor(target);
aclDestroyTensor(weight);
aclDestroyTensor(totalWeight);
aclDestroyTensor(out);
// 7. 释放device 资源
aclrtFree(gradDeviceAddr);
aclrtFree(selfDeviceAddr);
aclrtFree(targetDeviceAddr);
aclrtFree(weightDeviceAddr);
aclrtFree(totalWeightDeviceAddr);
aclrtFree(outDeviceAddr);
if (workspaceSize > 0) {
aclrtFree(workspaceAddr);
}
aclrtDestroyStream(stream);
aclrtResetDevice(deviceId);
aclFinalize();
return 0;
}