Documentation
Download

aclnnIm2col

📄 View Source Code

Applicable Products

ProductSupported
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:

Image to column, which converts sliding local window data into column vectors and concatenates them into a large tensor. Extracts sliding windows from the batched input tensor.

Consider a batched input tensor of shape (N, C, H, W) or (C, H, W), where N is the batch dimension, C is the channel dimension, and H, W represent the image size. This operation flattens each sliding kernel_size block within the spatial dimensions of the input into columns (i.e., the last dimension) of an output tensor of shape (N, C ×∏\times \prod(kernel_size), L) for 3-D or (C ×∏\times \prod(kernel_size), L) for 2-D, where L is the total number of such blocks.

  • Formula:

    L=∏d⌊spatial_size[d]+2×padding[d]−dilation[d]×(kernel_size[d]−1)−1stride[d]+1⌋L = \prod_{d} \lfloor \frac{spatial\_size[d] + 2 \times padding[d] - dilation[d] \times (kernel\_size[d] -1) -1}{stride[d]} + 1 \rfloor, where spatial_size consists of H and W of the input tensor above.

Function Prototype

Each operator is divided into a two-phase API. The "aclnnIm2colGetWorkspaceSize" API must be called first to obtain the workspace size required for computation and the executor that encapsulates the operator computation flow. Then, the "aclnnIm2col" API is called to perform the computation.

aclnnStatus aclnnIm2colGetWorkspaceSize(
    const aclTensor   *self,
    const aclIntArray *kernelSize,
    const aclIntArray *dilation,
    const aclIntArray *padding,
    const aclIntArray *stride,
    const aclTensor   *out,
    uint64_t          *workspaceSize,
    aclOpExecutor     **executor)
aclnnStatus aclnnIm2col(
    void          *workspace,
    uint64_t       workspaceSize,
    aclOpExecutor *executor,
    aclrtStream    stream)

aclnnIm2colGetWorkspaceSize

  • Parameters:

    Parameter Input/Output Description Instruction Data Type Data Format Dimension (shape) Non-contiguous Tensor
    self (aclTensor*) Input Input parameter for the im2col computation, corresponding to self in the formula. Supports empty tensors. INT8, UINT8, INT16, UINT16, INT32, UINT32, INT64, UINT64, BFLOAT16, FLOAT16, FLOAT, DOUBLE, BOOL, COMPLEX32, COMPLEX64 ND 3-4 √
    kernelSize (aclIntArray*) Input Size of the convolution kernel, corresponding to kernelSize in the formula. kernelSize[0] indicates the 'H' direction.
    kernelSize[1] indicates the 'W' direction.    		 INT64 - 2 -
    dilation (aclIntArray*) Input Dilation parameter, corresponding to dilation in the formula. dilation[0] indicates the 'H' direction.
    dilation[1] indicates the 'W' direction.    		 INT64 - 2 -
    padding (aclIntArray*) Input Padding size of the convolution, corresponding to padding in the formula. padding[0] indicates the 'H' direction.
    padding[1] indicates the 'W' direction.    		 INT64 - 2 -
    stride (aclIntArray*) Input Stride of the convolution, corresponding to stride in the formula. stride[0] indicates the 'H' direction.
    stride[1] indicates the 'W' direction.    		 INT64 - 2 -
    out (aclTensor*) Output Output parameter for the im2col computation, corresponding to out in the formula. Shape derived from the above parameters. INT8, UINT8, INT16, UINT16, INT32, UINT32, INT64, UINT64, BFLOAT16, FLOAT16, FLOAT, DOUBLE, BOOL, COMPLEX32, COMPLEX64 ND 2D (3D input) or 3D (4D input). √
    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. - - - - -
    • Atlas A3 training products/Atlas A3 inference products, Atlas A2 training products/Atlas A2 inference products: Only FLOAT, FLOAT16, and BFLOAT16 are supported.

  • Return value:

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

    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 input self, kernelSize, dilation, padding, stride, or out is a null pointer.
    ACLNN_ERR_PARAM_INVALID 161002 The data type of self is not within the supported range.
    The dimension of self is neither 3D nor 4D.
    The size of kernelSize, dilation, padding, or stride is not 2.
    An element in kernelSize, dilation, or stride has a value less than or equal to 0.
    An element in padding has a value less than 0.
    The data dimension of out differs from that of the infershape parameter.

aclnnIm2col

  • Parameters:

    Parameter Input/Output Description
    workspace Input Memory address of the workspace applied for on the Device side.
    workspaceSize Input Size of the workspace applied for on the Device side, obtained by the first-phase API aclnnIm2colGetWorkspaceSize.
    executor Input Op 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 Codes.

Constraints

  • Deterministic computation:

    • aclnnIm2col 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_im2col.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) {
  // 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);
  // Call aclrtMemcpy to copy data from the host side to the device memory.
  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 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. Custom construction is required based on the API.
  std::vector<int64_t> selfShape = {2, 2, 3};
  std::vector<int64_t> outShape = {8, 4};

  void* selfDeviceAddr = nullptr;
  void* outDeviceAddr = nullptr;
  aclTensor* self = nullptr;
  aclIntArray* kernelSize = nullptr;
  aclIntArray* dilation = nullptr;
  aclIntArray* padding = nullptr;
  aclIntArray* stride = nullptr;
  aclTensor* out = nullptr;

  std::vector<float> selfHostData = {0.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0};
  std::vector<int64_t> kernelSizeData = {2, 2};
  std::vector<int64_t> dilationData = {1, 1};
  std::vector<int64_t> paddingData = {1, 1};
  std::vector<int64_t> strideData = {2, 2};
  std::vector<float> outHostData = {0.0};

  // Create the self aclTensor.
  ret = CreateAclTensor(selfHostData, selfShape, &selfDeviceAddr, aclDataType::ACL_FLOAT, &self);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create the out aclTensor.
  ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_FLOAT, &out);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create the aclIntArray.
  kernelSize = aclCreateIntArray(kernelSizeData.data(), 2);
  CHECK_RET(kernelSize != nullptr, return ret);
  dilation = aclCreateIntArray(dilationData.data(), 2);
  CHECK_RET(dilation != nullptr, return ret);
  padding = aclCreateIntArray(paddingData.data(), 2);
  CHECK_RET(padding != nullptr, return ret);
  stride = aclCreateIntArray(strideData.data(), 2);
  CHECK_RET(stride != nullptr, 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 aclnnIm2col.
  ret = aclnnIm2colGetWorkspaceSize(self, kernelSize, dilation, padding, stride, out, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnIm2colGetWorkspaceSize 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 aclnnIm2col.
  ret = aclnnIm2col(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnIm2col failed. ERROR: %d\n", ret); return ret);

  // 4. (boilerplate) Synchronize and wait for 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 to the host side. Modify based on the specific API definition.
  auto size = GetShapeSize(outShape);
  std::vector<float> 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: %f\n", i, resultData[i]);
  }

  // 6. Release aclTensor and aclIntArray. Modify based on the specific API definition.
  aclDestroyTensor(self);
  aclDestroyIntArray(kernelSize);
  aclDestroyIntArray(dilation);
  aclDestroyIntArray(padding);
  aclDestroyIntArray(stride);
  aclDestroyTensor(out);

  // 7. Release device resources. Modify based on the specific API definition.
  aclrtFree(selfDeviceAddr);
  aclrtFree(outDeviceAddr);
  if (workspaceSize > 0) {
    aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();
  return 0;
}
This site uses cookies to improve your browsing experience. By continuing to browse this site, you agree to our use of cookies.View Details