aclnnScatterList

Applicable Products

Function

• Description: Applies sparse updates to a variable reference tensor list, scattering values from updates to the corresponding dimension of var based on indices.

• Formula:

  $$var[i][j] = updates[k][j] \quad \text{when} \quad indices[k] = i$$

  Where indices is the index tensor, updates is the value tensor to be updated, var is the tensor list to be updated, and the scatter dimension is specified by the axis parameter.

Function Prototype

Each operator is divided into a two-phase API. The "aclnnScatterListGetWorkspaceSize" API must be called first to obtain the workspace size required for computation and the executor that encapsulates the operator computation flow. Then, the "aclnnScatterList" API is called to perform the computation.

aclnnStatus aclnnScatterListGetWorkspaceSize(
  aclTensorList*       var,
  const aclTensor*     indices,
  const aclTensor*     updates,
  const aclTensor*     mask,
  char*                reduce,
  int64_t              axis,
  uint64_t*            workspaceSize,
  aclOpExecutor**      executor)

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

aclnnScatterListGetWorkspaceSize

• Parameters:

  Parameter	Input/Output	Description	Instruction	Data Type	Data Format	Shape	Non-contiguous Tensor
  var (aclTensorList*)	Input/Output	List of tensors to be updated.	The shape of each tensor in the tensor list and updates satisfy the broadcast relationship. The data type of each tensor must be the same as the data type of updates.	INT8, INT16, INT32, UINT8, UINT16, UINT32, FLOAT16, BFLOAT16, FLOAT	ND	0-8	√
  indices (aclTensor*)	Input	Index tensor.	-	INT32, INT64	ND	0-8	√
  updates (aclTensor*)	Input	The value tensor to be updated to var.	The data type must be the same as the tensor data type in var. The shape must satisfy the broadcast relationship with the shape of each tensor in var.	INT8, INT16, INT32, UINT8, UINT16, UINT32, FLOAT16, BFLOAT16, FLOAT	ND	0-8	√
  mask (aclTensor*)	Optional parameter.	Mask tensor.	-	UINT8	ND	0-8	√
  reduce (const char*)	Attribute	Type of the reduction operation to apply.	Currently, only "update" is supported.	string	-	-	-
  axis (int64_t)	Attribute	Specifies the dimension along which to scatter.	The default value is -2.	int64_t	-	-	-
  workspaceSize (uint64_t*)	Output	Returns the workspace size to be allocated on the Device side.	-	-	-	-	-
  executor (aclOpExecutor**)	Output	Returns the op executor, which contains the operator computation flow.	-	-	-	-	-

• 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 Code	Error Code	Description
  ACLNN_ERR_PARAM_NULLPTR	161001	Indicates that the input tensor or tensorList is a null pointer.
  ACLNN_ERR_PARAM_INVALID	161002	The data type of the tensor in var is not within the supported range.
  		The data type of indices is not within the supported range.
  		The reduce parameter is not within the supported range.
  		The data type of the tensor in var is inconsistent with that of updates.

aclnnScatterList

• Parameters:

  Parameter	Input/Output	Description
  workspace	Input	Specifies the workspace memory address applied for on the Device side.
  workspaceSize	Input	Specifies the workspace size applied for on the Device side, which is obtained through the first-phase API aclnnScatterListGetWorkspaceSize.
  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:

  • aclnnScatterList defaults to a non-deterministic implementation. You can enable deterministic computation through aclrtCtxSetSysParamOpt.

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_scatter_list.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;
}

void PrintOutResult(std::vector<int64_t> &shape, void** deviceAddr) {
  auto size = GetShapeSize(shape);
  std::vector<float> resultData(size, 0);
  auto ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]),
                         *deviceAddr, 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);
  for (int64_t i = 0; i < size; i++) {
    LOG_PRINT("result[%ld] is: %f\n", i, resultData[i]);
  }
}

int Init(int32_t deviceId, aclrtStream* stream) {
  // Boilerplate: 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 allocate device memory.
  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 data from the host 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 a 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];
  }

  // Create an aclTensor by calling aclCreateTensor.
  *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. See the ACL API documentation.
  // Set deviceId based on your actual device.
  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. Construct inputs and outputs based on the API.
  std::vector<int64_t> varShape = {5, 3, 4};
  std::vector<int64_t> indicesShape = {1, 2};
  std::vector<int64_t> updatesShape = {1, 5, 3, 4};

  void* varDeviceAddr = nullptr;
  void* indicesDeviceAddr = nullptr;
  void* updatesDeviceAddr = nullptr;

  aclTensor* varTensor = nullptr;
  aclTensor* indices = nullptr;
  aclTensor* updates = nullptr;

  std::vector<float> varHostData(5 * 3 * 4, 0);
  std::vector<int64_t> indicesHostData = {0, 1};
  std::vector<float> updatesHostData(1 * 5 * 3 * 4, 10);

  ret = CreateAclTensor(varHostData, varShape, &varDeviceAddr, aclDataType::ACL_FLOAT, &varTensor);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  aclTensor* varArray[] = {varTensor};
  auto varTensorList = aclCreateTensorList(varArray, 1);

  ret = CreateAclTensor(indicesHostData, indicesShape, &indicesDeviceAddr, aclDataType::ACL_INT64, &indices);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  ret = CreateAclTensor(updatesHostData, updatesShape, &updatesDeviceAddr, aclDataType::ACL_FLOAT, &updates);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  // 3. Call the CANN operator library API.
  uint64_t workspaceSize = 0;
  aclOpExecutor* executor;

  // Call the first-phase API of aclnnScatterList.
  ret = aclnnScatterListGetWorkspaceSize(varTensorList, indices, updates, nullptr, "update", -2, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnScatterListGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);

  // Allocate 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 aclnnScatterList.
  ret = aclnnScatterList(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnScatterList failed. ERROR: %d\n", ret); return ret);

  // 4. (Boilerplate) Wait for the task execution to finish synchronously.
  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.
  PrintOutResult(varShape, &varDeviceAddr);

  // 6. Release the aclTensor.
  aclDestroyTensorList(varTensorList);
  aclDestroyTensor(indices);
  aclDestroyTensor(updates);

  // 7. Release the device resources.
  aclrtFree(varDeviceAddr);
  aclrtFree(indicesDeviceAddr);
  aclrtFree(updatesDeviceAddr);
  if (workspaceSize > 0) {
    aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();

  return 0;
}