aclnnQr

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: Performs orthogonal decomposition on the input tensor.
Formula:
$A = QR$
Where $A$ is the input tensor with at least two dimensions, and $A$ can be expressed as the product of an orthogonal matrix $Q$ and an upper triangular matrix $R$ .

Example:

A = tensor([[1, 2], [3, 4]], dtype=torch.float)
Q, R = QR(A, some=False)
Q = tensor([[-0.3162, -0.9487],
            [-0.9487, 0.3162]])
R = tensor([[-3.1623, -4.4272],
            [0.0000, -0.6325]])

Function Prototype

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

aclnnStatus aclnnQrGetWorkspaceSize(
  const aclTensor* self,
  bool             some,
  aclTensor*       Q,
  aclTensor*       R,
  uint64_t*        workspaceSize,
  aclOpExecutor**  executor)

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

aclnnQrGetWorkspaceSize

Parameters

Parameter	Input/Output	Description	Instruction	Data Type	Data Format	Dimension (Shape)	Non-contiguous Tensor
self (aclTensor*)	Input	A in the formula.	The shape has at least 2 dimensions and no more than 8, in the format [..., M, N], where ... indicates 0 to 6 dimensions.	FLOAT, FLOAT16, DOUBLE, COMPLEX64, COMPLEX128	ND	2-8	√
some (bool)	Input	A computation attribute that controls the output form of Q and R in the formula.	When set to false, Q is a square matrix. For example, for A[..., M, N], the output is the full Q[..., M, M] and R[..., M, N]. When set to true, Q is a thin matrix. For example, for A[..., M, N], the output is Q[..., M, K] and R[..., K, N], where K is the minimum of M and N.	BOOL	-	-	-
Q (aclTensor*)	Output	Q in the formula, the orthogonal matrix output by the QR decomposition.	For shape constraints, refer to the `some` parameter description. The data format must be consistent with `self` and R.	FLOAT, FLOAT16, DOUBLE, COMPLEX64, COMPLEX128	ND	Deduced from some	√
R (aclTensor*)	Output	R in the formula, the upper triangular matrix output by the orthogonal decomposition.	The shape constraints refer to the `some` parameter description, and the data format must be consistent with `self` and Q.	FLOAT, FLOAT16, DOUBLE, COMPLEX64, COMPLEX128	ND	Derived from some.	√
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.	-	-	-	-	-

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	A null pointer exists in self, Q, or R.
ACLNN_ERR_PARAM_INVALID	161002	The data type and data format of self, Q, and R are not within the supported range.
ACLNN_ERR_PARAM_INVALID	161002	The shapes of self, Q, and R do not meet the constraints.

aclnnQr

Parameters

Parameter	Input/Output	Description
workspace	Input	Address of the workspace memory applied on the Device side.
workspaceSize	Input	Workspace size obtained from the first-phase API `aclnnQrGetWorkspaceSize`.
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 Codes.

Constraints

Determinism description: aclnnQr 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_qr.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: 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);
  // Call aclrtMemcpy to copy host-side data to device-side 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 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 PrepareInputAndOutput(
    std::vector<int64_t>& selfShape, std::vector<int64_t>& qShape, std::vector<int64_t>& rShape, void** selfDeviceAddr,
    aclTensor** self, void** qDeviceAddr, aclTensor** q, void** rDeviceAddr, aclTensor** r)
{
  std::vector<float> selfHostData = {1, 2, 3, 4};
  std::vector<float> qHostData = {0, 0, 0, 0};
  std::vector<float> rHostData = {0, 0, 0, 0};
  // Create the self aclTensor.
  auto ret = CreateAclTensor(selfHostData, selfShape, selfDeviceAddr, aclDataType::ACL_FLOAT, self);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create the q and r aclTensors.
  ret = CreateAclTensor(qHostData, qShape, qDeviceAddr, aclDataType::ACL_FLOAT, q);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  ret = CreateAclTensor(rHostData, rShape, rDeviceAddr, aclDataType::ACL_FLOAT, r);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  return ACL_SUCCESS;
}

void ReleaseTensorAndScalar(aclTensor* self, aclTensor* q, aclTensor* r)
{
    aclDestroyTensor(self);
    aclDestroyTensor(q);
    aclDestroyTensor(r);
}

void ReleaseDevice(
    void* selfDeviceAddr, void* qDeviceAddr, void* rDeviceAddr, uint64_t workspaceSize, void* workspaceAddr, aclrtStream stream,
    int32_t deviceId)
{
    aclrtFree(selfDeviceAddr);
    aclrtFree(qDeviceAddr);
    aclrtFree(rDeviceAddr);
    if (workspaceSize > 0) {
      aclrtFree(workspaceAddr);
    }
    aclrtDestroyStream(stream);
    aclrtResetDevice(deviceId);
    aclFinalize();
}

int main() {
  // 1. (Boilerplate) Initialize the device/stream. See the ACL API manual.
  // Fill in 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 interface.
  std::vector<int64_t> selfShape = {2, 2};
  bool some = false;
  std::vector<int64_t> qShape = {2, 2};
  std::vector<int64_t> rShape = {2, 2};
  void* selfDeviceAddr = nullptr;
  void* qDeviceAddr = nullptr;
  void* rDeviceAddr = nullptr;
  aclTensor* self = nullptr;
  aclTensor* q = nullptr;
  aclTensor* r = nullptr;

  ret = PrepareInputAndOutput(selfShape, qShape, rShape, &selfDeviceAddr, &self, &qDeviceAddr, &q, &rDeviceAddr, &r);
  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 aclnnQr first-phase API.
  ret = aclnnQrGetWorkspaceSize(self, some, q, r, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnQrGetWorkspaceSize 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 aclnnQr.
  ret = aclnnQr(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnQr failed. ERROR: %d\n", ret); return ret);

  // 4. (Boilerplate) Synchronously 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 values and copy the results from the device memory to the host memory. Modify based on the specific API definition.
  auto size = GetShapeSize(qShape);
  std::vector<float> resultQData(size, 0);
  ret = aclrtMemcpy(resultQData.data(), resultQData.size() * sizeof(resultQData[0]), qDeviceAddr,
                    size * sizeof(resultQData[0]), ACL_MEMCPY_DEVICE_TO_HOST);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("copy result Q from device to host failed. ERROR: %d\n", ret); return ret);
  for (int64_t i = 0; i < size; i++) {
    LOG_PRINT("result Q[%ld] is: %f\n", i, resultQData[i]);
  }

  std::vector<float> resultRData(size, 0);
  ret = aclrtMemcpy(resultRData.data(), resultRData.size() * sizeof(resultRData[0]), rDeviceAddr,
                    size * sizeof(resultRData[0]), ACL_MEMCPY_DEVICE_TO_HOST);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("copy result R from device to host failed. ERROR: %d\n", ret); return ret);
  for (int64_t i = 0; i < size; i++) {
    LOG_PRINT("result R[%ld] is: %f\n", i, resultRData[i]);
  }

  // 6. Release aclTensor and aclScalar. Modify based on the specific API definition.
  ReleaseTensorAndScalar(self, q, r);

  // 7. Release device resources. Modify parameters based on the specific API definition.
  ReleaseDevice(selfDeviceAddr, qDeviceAddr, rDeviceAddr, workspaceSize, workspaceAddr, stream, deviceId);

  return 0;
}