aclnnInplaceCopy

Supported Products

Product	Supported
Atlas A3 training products/Atlas A3 inference products	√
Atlas A2 training products/Atlas A2 inference products	√
Atlas 200I/500 A2 inference products	√
Atlas inference products	√
Atlas training products	√

Function

Description: Copies elements in src to the selfRef tensor and returns selfRef.
Formula:
${selfRef}_{i} = {src}_{i}$

Example:

Input selfRef:
tensor([[1, 2],
        [3, 4]])
Input src:
tensor([[5, 6],
        [7, 8]])

Output selfRef:
tensor([[5, 6],
        [7, 8]])

Prototype

Each operator has two-phase API calls. First, aclnnInplaceCopyGetWorkspaceSize is called to obtain the input parameters and compute the required workspace size based on the process. Then, aclnnInplaceCopy is called to perform computation.

aclnnStatus aclnnInplaceCopyGetWorkspaceSize(aclTensor *selfRef, const aclTensor *src, uint64_t *workspaceSize, aclOpExecutor **executor)
aclnnStatus aclnnInplaceCopy(void *workspace, uint64_t workspaceSize, aclOpExecutor *executor, const aclrtStream stream)

aclnnInplaceCopyGetWorkspaceSize

Parameters:
- selfRef (aclTensor*, computation input | computation output): selfRef in the formula. The copy between complex numbers is supported only when selfRef is contiguous. The shape must meet the broadcast relationship with src. Non-contiguous tensors are supported. The data format can be ND.
  - Atlas A2 training products/Atlas A2 inference products and Atlas A3 training products/Atlas A3 inference products: The data type can be INT8, INT16, INT32, INT64, UINT8, FLOAT16, FLOAT32, BOOL, DOUBLE, COMPLEX64, COMPLEX128, UINT16, UINT32, UINT64 or BFLOAT16.
  - Atlas training products, Atlas inference products, and Atlas 200I/500 A2 inference products: The data type can be INT8, INT16, INT32, INT64, UINT8, FLOAT16, FLOAT32, BOOL, DOUBLE, COMPLEX64, COMPLEX128, UINT16, UINT32 or UINT64.
- src (aclTensor*, computation input): src in the formula. The copy between complex numbers is supported only when selfRef is contiguous. The shape must meet the broadcast relationship with selfRef. Non-contiguous tensors are supported. The data format can be ND.
  - Atlas A2 training products/Atlas A2 inference products and Atlas A3 training products/Atlas A3 inference products: The data type can be INT8, INT16, INT32, INT64, UINT8, FLOAT16, FLOAT32, BOOL, DOUBLE, COMPLEX64, COMPLEX128, UINT16, UINT32, UINT64 or BFLOAT16.
  - Atlas training products, Atlas inference products, and Atlas 200I/500 A2 inference products: The data type can be INT8, INT16, INT32, INT64, UINT8, FLOAT16, FLOAT32, BOOL, DOUBLE, COMPLEX64, COMPLEX128, UINT16, UINT32 or UINT64.
- workspaceSize (uint64_t*, output): size of the workspace to be allocated on the device.
- executor (aclOpExecutor**, output): operator executor, containing the operator computation process.

Returns:

aclnnStatus: status code. For details, see aclnn Return Codes.

The first-phase API implements input parameter validation. The following error codes may be returned:
161001 (ACLNN_ERR_PARAM_NULLPTR): 1. selfRef or src is a null pointer.
161002 (ACLNN_ERR_PARAM_INVALID): 1. The data type of selfRef or src is not supported.
                                  2. The shape of selfRef exceeds 8 dimensions.
                                  3. The shape of src cannot be broadcast to selfRef.
                                  4. The data type of src is not supported or cannot be converted to selfRef.

aclnnInplaceCopy

Parameters:
- workspace (void*, input): address of the workspace to be allocated on the device.
- workspaceSize (uint64_t, input): size of the workspace to be allocated on the device, which is obtained by the first-phase API aclnnInplaceCopyGetWorkspaceSize.
- executor (aclOpExecutor*, input): operator executor, containing the operator computation process.
- stream (aclrtStream, input): stream for executing the task.
Returns:

aclnnStatus: status code. For details, see aclnn Return Codes.

Constraints

Deterministic computing:
- aclnnInplaceCopy defaults to a deterministic implementation.

When the data types of src and selfRef are inconsistent, the conversion rules for floating-point types (HIFLOAT8, FLOAT8_E5M2, and FLOAT8_E4M3FN) are as follows:

HIFLOAT8 -> FLOAT32, BFLOAT16, FLOAT16
FLOAT8_E5M2 -> FLOAT32, BFLOAT16, FLOAT16
FLOAT8_E4M3FN -> FLOAT32, BFLOAT16, FLOAT16
BFLOAT16 -> HIFLOAT8
FLOAT16 -> HIFLOAT8
FLOAT32 -> HIFLOAT8, FLOAT8_E5M2, FLOAT8_E4M3FN

Example

The following example is for reference only. For details, see Compilation and Running Sample.

#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_copy.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 shape_size = 1;
  for (auto i : shape) {
    shape_size *= i;
  }
  return shape_size;
}

int Init(int32_t deviceId, aclrtStream* stream) {
  // (Boilerplate) Initialize resources.
  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);
  // Call aclrtMalloc to allocate memory on the device.
  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);

  // Call aclrtMemcpy to copy the data on the host to the memory on the device.
  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);

  // Compute the strides of the contiguous tensor.
  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];
  }

  // Call aclCreateTensor to create an 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. (Boilerplate) Initialize the device and stream. For details, see the ACL API manual.
  // Set the device ID in use.
  int32_t deviceId = 0;
  aclrtStream stream;
  auto ret = Init(deviceId, &stream);
  // Handle the check as required.
  CHECK_RET(ret == 0, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);
  // 2. Construct inputs and outputs based on the API definition.
  std::vector<int64_t> selfRefShape = {4, 2};
  std::vector<int64_t> srcShape = {4, 2};
  void* selfRefDeviceAddr = nullptr;
  void* srcDeviceAddr = nullptr;
  aclTensor* selfRef = nullptr;
  aclTensor* src = nullptr;
  std::vector<float> selfRefHostData = {0, 1, 2, 3, 4, 5, 6, 7};
  std::vector<float> srcHostData = {1, 1, 1, 2, 2, 2, 3, 3};
  // Create a selfRef aclTensor.
  ret = CreateAclTensor(selfRefHostData, selfRefShape, &selfRefDeviceAddr, aclDataType::ACL_FLOAT, &selfRef);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create an other aclTensor.
  ret = CreateAclTensor(srcHostData, srcShape, &srcDeviceAddr, aclDataType::ACL_FLOAT, &src);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  // 3. Call the CANN operator library API, which needs to be replaced with the actual API.
  uint64_t workspaceSize = 0;
  aclOpExecutor* executor;
  // Call the first-phase API of aclnnInplaceCopy.
  ret = aclnnInplaceCopyGetWorkspaceSize(selfRef, src, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnInplaceCopyGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);
  // Allocate device memory based on workspaceSize computed by the first-phase API.
  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;);
  }
  // Call the second-phase API of aclnnInplaceCopy.
  ret = aclnnInplaceCopy(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnInplaceCopy failed. ERROR: %d\n", ret); return ret);
  // 4. (Boilerplate) Wait until the task execution is complete.
  ret = aclrtSynchronizeStream(stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSynchronizeStream failed. ERROR: %d\n", ret); return ret);
  // 5. Obtain the output value and copy the result from the device memory to the host. Modify the code based on the API definition.
  auto size = GetShapeSize(selfRefShape);
  std::vector<float> resultData(size, 0);
  ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), selfRefDeviceAddr, 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]);
  }

  // 6. Release aclTensor and aclScalar. Modify the configuration based on the API definition.
  aclDestroyTensor(selfRef);
  aclDestroyTensor(src);

  // 7. Release device resources. Modify the configuration based on the API definition.
  aclrtFree(selfRefDeviceAddr);
  aclrtFree(srcDeviceAddr);
  if (workspaceSize > 0) {
    aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();
  return 0;
}