下载
中文
注册
我要评分
文档获取效率
文档正确性
内容完整性
文档易理解
在线提单
论坛求助
昇腾小AI

aclnnRemainderTensorScalar&aclnnInplaceRemainderTensorScalar

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

  • aclnnRemainderTensorScalar和aclnnInplaceRemainderTensorScalar实现相同的功能,使用区别如下,请根据自身实际场景选择合适的算子。

    • aclnnRemainderTensorScalar:需新建一个输出张量对象存储计算结果。
    • aclnnInplaceRemainderTensorScalar:无需新建输出张量对象,直接在输入张量的内存中存储计算结果。
  • 每个算子分为两段式接口,必须先调用“aclnnRemainderTensorScalarGetWorkspaceSize”或者”aclnnInplaceRemainderTensorScalarGetWorkspaceSize“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“aclnnRemainderTensorScalar”或者”aclnnInplaceRemainderTensorScalar“接口执行计算。

    • aclnnStatus aclnnRemainderTensorScalarGetWorkspaceSize(const aclTensor *self, const aclScalar *other, aclTensor *out, uint64_t *workspaceSize, aclOpExecutor **executor)
    • aclnnStatus aclnnRemainderTensorScalar(void *workspace, uint64_t workspaceSize, aclOpExecutor *executor, aclrtStream stream)
    • aclnnStatus aclnnInplaceRemainderTensorScalarGetWorkspaceSize(aclTensor *selfRef, const aclScalar *other, uint64_t *workspaceSize, aclOpExecutor **executor)
    • aclnnStatus aclnnInplaceRemainderTensorScalar(void *workspace, uint64_t workspaceSize, aclOpExecutor *executor, aclrtStream stream)

功能描述

  • 算子功能: 将tensor self中的每个元素都转换为除以scalar other以后得到的余数。该结果与除数other同符号,并且该结果的绝对值是小于other的绝对值。

  • 实际计算remainder(self, other) 等效于以下公式:

    outi=selfifloor(selfi/other)otherout_i = self_i - floor(self_i / other) * other
  • 示例:

    self = tensor([[-1, -2],
                   [-3, -4]]).type(int32)
    other = 3.5   # float
    result = remainder(self, other)
    
    # result的值
    # tensor([[2.5000, 1.5000],
    #         [0.5000, 3.0000]])   float
    
    # 对于元素self中的-1来说,计算结果为 -1 - floor(-1 / 3.5) * 3.5 = 2.5
    # 可以看到,最终结果2.5的绝对值小于other 3.5。

aclnnRemainderTensorScalarGetWorkspaceSize

  • 参数说明:

    • self(aclTensor*, 计算输入):公式中的输入self,self的数据类型与other的数据类型需满足数据类型推导规则(参见互推导关系),并且推导出的数据类型必须属于INT32、INT64、FLOAT16、FLOAT、DOUBLE、BFLOAT16(仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持)类型中的一种。支持非连续的Tensor数据格式支持ND。

    • other(aclScalar*, 计算输入):公式中的输入other, self的数据类型与other的数据类型需满足数据类型推导规则(参见互推导关系),并且推导出的数据类型必须属于INT32、INT64、FLOAT16、FLOAT、DOUBLE、BFLOAT16(仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持)类型中的一种。

    • out(aclTensor*, 计算输出):公式中的输出out,数据类型支持UINT8、INT8、INT16、INT32、INT64、FLOAT16、FLOAT、DOUBLE、COMPLEX64、COMPLEX128、BFLOAT16(仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持)类型,且数据类型需要是self与other推导之后可转换的数据类型。shape需要与self一致。支持非连续的Tensor数据格式支持ND。

    • 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self、other、out是空指针。
    161002 (ACLNN_ERR_PARAM_INVALID): 1. self、out的shape不一样。
                                      2. self和other无法做数据类型推导。
                                      3. self和other推导出的数据类型不属于支持的数据类型。
                                      4. self和other推导出的数据类型无法转换为指定输出out的类型。
                                      5. self、out的维度数大于8维。

aclnnRemainderTensorScalar

  • 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aclnnRemainderTensorScalarGetWorkspaceSize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

aclnnInplaceRemainderTensorScalarGetWorkspaceSize

  • 参数说明

    • selfRef(aclTensor*, 计算输入|计算输出):输入输出tensor,selfRef的数据类型与other的数据类型需满足数据类型推导规则(参见互推导关系),并且推导出的数据类型必须属于INT32、INT64、FLOAT16、FLOAT、DOUBLE、BFLOAT16(仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持)类型中的一种,且需要是推导之后可转换的数据类型。支持非连续的Tensor数据格式支持ND。
    • other(aclScalar*, 计算输入):公式中的输入other, other的数据类型与selfRef的数据类型需满足数据类型推导规则(参见互推导关系),并且推导出的数据类型必须能转换为selfRef的数据类型。
    • 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selfRef、other是空指针
    161002 (ACLNN_ERR_PARAM_INVALID): 1. selfRef与other不能推导出数据类型
                                      2. selfRef与other推导出的数据类型不属于支持的数据类型
                                      3. selfRef与other推导出的数据类型不能转换为selfRef的数据类型
                                      4. selfRef的维度数大于8维

aclnnInplaceRemainderTensorScalar

  • 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aclnnInplaceRemainderTensorScalarGetWorkspaceSize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aclnnRemainderTensorScalar示例代码:

#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_remainder.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> selfShape = {3, 3};
  std::vector<int64_t> outShape = {3, 3};
  void* selfDeviceAddr = nullptr;
  void* outDeviceAddr = nullptr;
  aclTensor* self = nullptr;
  aclScalar* other = nullptr;
  aclTensor* out = nullptr;
  std::vector<int64_t> selfHostData = {0, 1, 2, 3, 4, 5, 6, 7, 8};
  std::vector<int64_t> outHostData = {0, 0, 0, 0, 0, 0, 0, 0, 0};
  int64_t Other = 3;

  // 创建self aclTensor
  ret = CreateAclTensor(selfHostData, selfShape, &selfDeviceAddr, aclDataType::ACL_INT64, &self);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建other aclScalar
  other = aclCreateScalar(&Other, aclDataType::ACL_INT64);
  CHECK_RET(other != nullptr, return ret);
  // 创建out aclTensor
  ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_INT64, &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aclnnRemainderTensorScalar第一段接口
  ret = aclnnRemainderTensorScalarGetWorkspaceSize(self, other, out, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnRemainderTensorScalarGetWorkspaceSize 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aclnnRemainderTensorScalar第二段接口
  ret = aclnnRemainderTensorScalar(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnRemainderTensorScalar 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<int64_t> 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: %ld\n", i, resultData[i]);
  }

  // 6. 释放aclTensor和aclScalar,需要根据具体API的接口定义修改
  aclDestroyTensor(self);
  aclDestroyScalar(other);
  aclDestroyTensor(out);

  // 7. 释放device资源
  aclrtFree(selfDeviceAddr);
  aclrtFree(outDeviceAddr);
  if (workspaceSize > 0) {
    aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();
  return 0;
}

aclnnInplaceRemainderTensorScalar示例代码:

#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_remainder.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> selfRefShape = {3, 3};
  void* selfRefDeviceAddr = nullptr;
  aclTensor* selfRef = nullptr;
  aclScalar* other = nullptr;
  std::vector<int64_t> selfRefHostData = {0, 1, 2, 3, 4, 5, 6, 7, 8};
  int64_t Other = 3;

  // 创建self aclTensor
  ret = CreateAclTensor(selfRefHostData, selfRefShape, &selfRefDeviceAddr, aclDataType::ACL_INT64, &selfRef);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建other aclScalar
  other = aclCreateScalar(&Other, aclDataType::ACL_INT64);
  CHECK_RET(other != nullptr, 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aclnnInplaceRemainderTensorScalar第一段接口
  ret = aclnnInplaceRemainderTensorScalarGetWorkspaceSize(selfRef, other, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnInplaceRemainderTensorScalarGetWorkspaceSize 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aclnnInplaceRemainderTensorScalar第二段接口
  ret = aclnnInplaceRemainderTensorScalar(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnInplaceRemainderTensorScalar 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(selfRefShape);
  std::vector<int64_t> resultData(size, 0);
  ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), selfRefDeviceAddr,
                    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: %ld\n", i, resultData[i]);
  }

  // 6. 释放aclTensor和aclScalar,需要根据具体API的接口定义修改
  aclDestroyTensor(selfRef);
  aclDestroyScalar(other);

  // 7. 释放device资源
  aclrtFree(selfRefDeviceAddr);
  if (workspaceSize > 0) {
    aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();

  return 0;
}
搜索结果
找到“0”个结果

当前产品无相关内容

未找到相关内容,请尝试其他搜索词