aclnnSoftmaxCrossEntropyWithLogits

📄 View Source Code

Applicable Products

Function

• Description: Computes the Cross-Entropy loss of softmax and cross entropy, and provides the backward gradient with respect to the input logits.

• Formula:

  $$\text{loss} = - \sum_i y_i \cdot \log(\text{softmax}(x_i))$$

Where $x_i$ corresponds to the input features, and $y_i$ corresponds to the input labels.

Function Prototype

Each operator is divided into a two-phase API. You must first call the "aclnnSoftmaxCrossEntropyWithLogitsGetWorkspaceSize" API to obtain the workspace size required for computation and the executor that contains the operator computation process, and then call the "aclnnSoftmaxCrossEntropyWithLogits" API to perform the computation.

aclnnStatus aclnnSoftmaxCrossEntropyWithLogitsGetWorkspaceSize(
  const aclTensor* features,
  aclTensor*       labels,
  aclTensor*       loss,
  aclTensor*       backprop,
  uint64_t*        workspaceSize,
  aclOpExecutor**  executor)

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

aclnnSoftmaxCrossEntropyWithLogits

• Parameters:

  Parameter	Input/Output	Description	Instruction	Data Type	Data Format	Dimension (Shape)	Non-Contiguous Tensor
  features (aclTensor*)	Input	Represents the model's output logits, denoted as x_i in the formula.	Nullable.	FLOAT, FLOAT16, BFLOAT16	ND	2	√
  labels (aclTensor*)	Input	Represents the input labels, i.e., y_i in the formula.	The shape and data type are consistent with features. Nullable.	FLOAT, FLOAT16, BFLOAT16	ND	2	√
  loss (aclTensor*)	Output	Represents the Cross-Entropy loss of softmax and cross entropy, denoted as loss in the formula.	The data type is the same as that of features. Nullable tensor.	FLOAT, FLOAT16, BFLOAT16	ND	1	√
  backprop（aclTensor*）	Output	Represents the backward gradient of the input logits, denoted as backprop in the formula.	The data type is the same as that of features. Nullable tensor.	FLOAT, FLOAT16, BFLOAT16	ND	2	√
  workspaceSize (uint64_t*)	Output	Returns the workspace size to be allocated on the Device side.	-	-	-	-	-
  executor (aclOpExecutor**)	Output	Returns the operator executor, which contains the operator computation flow.	-	-	-	-	-

  • Atlas training products, Atlas inference products: The supported data types are FLOAT and FLOAT16.

• 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	The passed features, labels, loss, or backprop is a null pointer.
  ACLNN_ERR_PARAM_INVALID	161002	The data types of features and labels are not within the supported range.

aclnnSoftmaxCrossEntropyWithLogits

• Parameters:

  Parameter	Input/Output	Description
  workspace	Input	Address of the workspace memory applied for on the Device side.
  workspaceSize	Input	Size of the workspace applied for on the Device side, obtained through the first-phase API aclnnSoftmaxCrossEntropyWithLogitsGetWorkspaceSize.
  executor	Input	Operator executor, which 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:

  • aclnnSoftmaxCrossEntropyWithLogits defaults to a deterministic implementation.

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_softmax_cross_entropy_with_logits.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<int8_t> 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("mean result[%ld] is: %d\n", i, resultData[i]);
  }
}

int Init(int32_t deviceId, aclrtStream* stream) {
  // Boilerplate: 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 device-side 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 using 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];
  }

  // 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. See the ACL API manual.
  // Enter the 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 the input and output. Customize the construction based on the API.
  std::vector<int64_t> featuresShape = {16, 4};
  std::vector<int64_t> labelsShape = {16, 4};

  std::vector<float> featuresHostData(16*4, 1);
  std::vector<float> labelsHostData(16*4, 2);

  void* featuresDeviceAddr = nullptr;
  void* labelsDeviceAddr = nullptr;

  aclTensor* features = nullptr;
  aclTensor* labels = nullptr;

  ret = CreateAclTensor(featuresHostData, featuresShape, &featuresDeviceAddr, aclDataType::ACL_FLOAT, &features);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  ret = CreateAclTensor(labelsHostData, labelsShape, &labelsDeviceAddr, aclDataType::ACL_FLOAT, &labels);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  std::vector<int64_t> lossShape = {16};
  std::vector<int16_t> lossHostData(16, 8);
  aclTensor* loss = nullptr;
  void* lossDeviceAddr = nullptr;

  std::vector<int64_t> backpropShape = {16,4};
  std::vector<int16_t> backpropHostData(16*4, 9);
  aclTensor* backprop = nullptr;
  void* backpropDeviceAddr = nullptr;

  ret = CreateAclTensor(lossHostData, lossShape, &lossDeviceAddr, aclDataType::ACL_FLOAT, &loss);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  ret = CreateAclTensor(backpropHostData, backpropShape, &backpropDeviceAddr, aclDataType::ACL_FLOAT, &backprop);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

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

  // Call the first-phase API of aclnnSoftmaxCrossEntropyWithLogits.
  ret = aclnnSoftmaxCrossEntropyWithLogitsGetWorkspaceSize(features, labels, loss, backprop,
    &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnDequantBiasGetWorkspaceSize failed. 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 aclnnSoftmaxCrossEntropyWithLogits.
  ret = aclnnSoftmaxCrossEntropyWithLogits(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnDequantBias failed. ERROR: %d\n", ret); return ret);

  // 4. (Boilerplate) Synchronously wait for the task execution to finish.
  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.
  PrintOutResult(lossShape, &lossDeviceAddr);

  auto size = GetShapeSize(lossShape);
  std::vector<float> resultData(size, 0);
  ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), lossDeviceAddr,
                    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. Release aclTensor and aclScalar. Modify based on the specific API definition.
  aclDestroyTensor(features);
  aclDestroyTensor(labels);
  aclDestroyTensor(loss);
  aclDestroyTensor(backprop);

  // 7. Release device resources. Modify based on the specific API definition.
  aclrtFree(featuresDeviceAddr);
  aclrtFree(labelsDeviceAddr);
  aclrtFree(lossDeviceAddr);
  aclrtFree(backpropDeviceAddr);
  if (workspaceSize > 0) {
    aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();

  return 0;
}