aclnnIndexPutImpl

Applicable Products

Product	Supported
Ascend 950PR/Ascend 950DT	√
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: Replaces or accumulates the data at the coordinates of the input self specified by indices with the input values.
Formula:
- accumulate = False:
  $self[indices] = values$
- accumulate = True:
  $self[indices] = self[indices] + values$

Function Prototype

Each operator is divided into a two-phase API. You must first call the "aclnnIndexPutImplGetWorkspaceSize" API to obtain the input parameters and calculate the required workspace size based on the computation process, and then call the "aclnnIndexPutImpl" API to perform the computation.

aclnnStatus aclnnIndexPutImplGetWorkspaceSize(
 aclTensor*             selfRef,
 const aclTensorList*   indices,
 const aclTensor*       values,
 const bool             accumulate,
 const bool             unsafe,
 uint64_t*              workspaceSize,
 aclOpExecutor**        executor)

aclnnStatus aclnnIndexPutImpl(
 void*             workspace,
 uint64_t          workspaceSize,
 aclOpExecutor*    executor,
 aclrtStream       stream)

aclnnIndexPutImplGetWorkspaceSize

Parameters

Parameter	Input/Output	Description	Instruction	Data Type	Data Format	Dimension (Shape)	Non-contiguous Tensor
selfRef	Input	self in the formula.	Data type is the same as that of values.	FLOAT, FLOAT16, DOUBLE, INT32, INT64, INT16, INT8, UINT8, BOOL, BFLOAT16	ND	1-8	√
indices	Input	Indices in the formula.	-	INT32, INT64, BOOL	ND	-	-
values	Input	The values in the formula.	-	Same as selfRef.	ND	-	-
accumulate	Input	Operation type flag for accumulation or update, a boolean value on the host side.	When accumulate is True, the operation is accumulation; When accumulate is False, the operation is update.	-	-	-	-
unsafe	Input	Flag for checking whether indices are within the valid range.	When unsafe is True, an out-of-bounds index causes an immediate error and terminates execution. When unsafe is False, an out-of-bounds index may cause a runtime exception.	-	-	-	-
workspaceSize	Output	Returns the workspace size to be applied for on the Device side.	-	-	-	-	-
Executor	Output	Returns the operator executor, which contains the operator computation flow.	-	-	-	-	-

Atlas training products: The data type does not support BFLOAT16.
Ascend 950PR/Ascend 950DT: When accumulate is False and indices contain duplicate index values, the result is not order-preserving.

Return Value

aclnnStatus: return code. For details, see aclnn Return Code.

The first-phase API performs input parameter validation and returns an error in the following scenarios:

Return Value	Error Code	Description
ACLNN_ERR_PARAM_NULLPTR	161001	The passed selfRef, indices, or values is a null pointer.
ACLNN_ERR_PARAM_INVALID	161002	The data types of selfRef and values are not within the supported range.
		The data types of selfRef and values are different.
		The data formats of selfRef and values are different.

aclnnIndexPutImpl

Parameters

Parameter	Input/Output	Description
workspace	Input	Address of the workspace memory applied on the Device side.
workspaceSize	Input	Size of the workspace applied on the Device side, obtained through the first-phase API aclnnIndexPutImplGetWorkspaceSize.
Executor	Input	Operator executor that contains the operator computation flow.
Stream	Input	Specifies the stream for task execution.

Return Value

aclnnStatus: return code. For details, see aclnn Return Code.

Constraints

Deterministic computation:
- aclnnIndexPutImpl defaults to a non-deterministic implementation. Deterministic computation can be enabled through aclrtCtxSetSysParamOpt.
The input parameters selfRef, indices, and values generally have the following constraints:
- The number of tensors in indices cannot exceed the dimension of selfRef.
- The dimensions of values must satisfy the following formula or satisfy it after broadcasting:
  - values.Dims() = indices[i].Dims() + (selfRef.Dims() - indices.size())
  - This means that the first half of the values dimensions must match the dimensions of the tensors in indices (the tensors in indices are broadcast to the same shape), and the second half must match the dimensions of selfRef minus the number of tensors in indices.

Example

The following provides example code for reference only. For details about compilation and running, see Compile and Run Samples.

#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_index_put_impl.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 for resource initialization.
  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 apply for memory on the device side.
  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);
  // Copy host data to the device memory by calling aclrtMemcpy.
  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);
  // Calculate 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 the aclCreateTensor API 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/stream. Refer to the ACL API manual.
  // Fill in the deviceId based on the actual device.
  int32_t deviceId = 0;
  aclrtStream stream;
  auto ret = Init(deviceId, &stream);
  // Handle the check as needed.
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);

  // 2. Construct the input and output. Customize the construction based on the API.
  std::vector<int64_t> selfShape = {3, 4};
  std::vector<int64_t> indexShape = {1};
  std::vector<int64_t> valueShape = {1};
  void* selfDeviceAddr = nullptr;
  void* indexOneDeviceAddr = nullptr;
  void* indexTwoDeviceAddr = nullptr;
  void* valueDeviceAddr = nullptr;
  aclTensor* self = nullptr;
  aclTensor* indexOne = nullptr;
  aclTensor* indexTwo = nullptr;
  aclTensor* value= nullptr;
  std::vector<float> selfHostData = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12};
  std::vector<float> valueHostData = {1};
  std::vector<int64_t> indexOneHostData = {0};
  std::vector<int64_t> indexTwoHostData = {2};
  // Create the self aclTensor.
  ret = CreateAclTensor(selfHostData, selfShape, &selfDeviceAddr, aclDataType::ACL_FLOAT, &self);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create the index aclTensor.
  ret = CreateAclTensor(indexOneHostData, indexShape, &indexOneDeviceAddr, aclDataType::ACL_INT64, &indexOne);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  ret = CreateAclTensor(indexTwoHostData, indexShape, &indexTwoDeviceAddr, aclDataType::ACL_INT64, &indexTwo);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  aclTensor* indexs[] = {indexOne, indexTwo};
  auto indexTensorList = aclCreateTensorList(indexs, 2);
  // value aclTensor
  ret = CreateAclTensor(valueHostData, valueShape, &valueDeviceAddr, aclDataType::ACL_FLOAT, &value);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  uint64_t workspaceSize = 0;
  aclOpExecutor* executor;

  // Example of aclnnIndexPutImpl API call.
  // 3. Call the CANN operator library API. Replace with the specific operator API.
  // Call the first-phase API of aclnnIndexPutImpl.
  ret = aclnnIndexPutImplGetWorkspaceSize(self, indexTensorList, value, true, false, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnIndexPutImplGetWorkspaceSizefailed. ERROR: %d\n", ret); return ret);
  // Apply for device memory based on the workspaceSize calculated 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 aclnnIndexPutImpl.
  ret = aclnnIndexPutImpl(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnIndexPutImplfailed. ERROR: %d\n", ret); return ret);

  // 4. (Boilerplate) Wait synchronously for the task execution to 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 side memory to the Host side. Modify based on the specific API definition.
  auto size = GetShapeSize(selfShape);
  std::vector<float> resultData(size, 0);
  ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), selfDeviceAddr, 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. Modify based on the specific API definition.
  aclDestroyTensor(self);
  aclDestroyTensorList(indexTensorList);
  aclDestroyTensor(value);

  // 7. Release device resources. Modify based on the specific API definition.
  aclrtFree(selfDeviceAddr);
  aclrtFree(indexOneDeviceAddr);
  aclrtFree(indexTwoDeviceAddr);
  aclrtFree(valueDeviceAddr);
  if (workspaceSize > 0) {
   aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();

  return 0;
}