下载
中文
注册

aclStft

支持的产品型号

  • Atlas A2训练系列产品/Atlas 800I A2推理产品。

接口原型

每个算子分为两段式接口,必须先调用“aclStftGetWorkspaceSize”接口获取计算所需workspace大小以及包含了算子计算流程的执行器,再调用“aclStft”接口执行计算。

  • aclnnStatus aclStftGetWorkspaceSize(const aclTensor *self, const aclTensor *windowOptional, aclTensor *out, int64_t nFft, int64_t hopLength, int64_t winLength, bool normalized, bool onesided, bool returnComplex, uint64_t *workspaceSize, aclOpExecutor **executor)
  • aclnnStatus aclStft(void* workspace, uint64_t workspaceSize, aclOpExecutor* executor, aclrtStream stream)

功能描述

  • 算子功能: 计算输入在滑动窗口内的傅里叶变换。
  • 计算公式:X[w,m]=k=0winLength1window[k]self[mhopLength+k]exp(j2πwknFft)X[w,m]=\sum_{k=0}^{winLength-1}window[k]self[m*hopLength+k]exp(-j*\frac{2{\pi}wk}{nFft})

其中,ww为FFT的频点;mm为滑动窗口的index;selfself为1维或2维tensor,当selfself是1维时,其为一个时序采样序列,当selfself是2维时,其为多个时序采样序列;hopLengthhopLength为滑动窗口大小;windowwindow为1维tensor,是STFT的窗函数(例如hann_window),其长度为winLengthwinLengthexp(j2πwknFft)exp(-j*\frac{2{\pi}wk}{nFft})为旋转因子。

aclStftGetWorkspaceSize

  • 参数说明:

    • self(aclTensor*,计算输入):必选参数,待计算的输入,要求是一个1D/2D的Tensor,shape为(L)/(B, L),其中,L为时序采样序列的长度,B为时序采样序列的个数。数据类型支持FLOAT32、DOUBLE、COMPLEX64、COMPLEX128,支持非连续的Tensor数据格式要求为ND。

    • windowOptional(aclTensor*,计算输入):可选参数,要求是一个1D的Tensor,shape为(winLength),winLength为STFT窗函数的长度。数据类型支持FLOAT32、DOUBLE、COMPLEX64、COMPLEX128,且数据类型与self保持一致,数据格式要求为ND。

    • nFft(int64_t,计算输入):必选参数,Host侧的int,FFT的点数(大于0)。

    • hopLength(int64_t,计算输入):必选参数,Host侧的int,滑动窗口的间隔(大于0)。

    • winLength(int64_t,计算输入):必选参数,Host侧的int,window的大小(大于0)。

    • normalized(bool,计算输入):必选参数,Host侧的bool,是否对傅里叶变换结果进行标准化。

    • onesided(bool,计算输入):必选参数,Host侧的bool,是否返回全部的结果或者一半结果。

    • returnComplex(bool,计算输入):必选参数,Host侧的bool,确认返回值是complex tensor或者是实部、虚部分开的tensor。

    • workspaceSize(uint64_t*,出参):Device侧的整型,返回需要在Device侧申请的workspace大小。

    • executor(aclOpExecutor**,出参):Device侧的aclOpExecutor,返回op执行器,包含了算子计算流程。

    • out(aclTensor*,计算输出):self在window内的傅里叶变换结果,要求是一个2D/3D/4D的Tensor,数据类型支持FLOAT32、DOUBLE、COMPLEX64、COMPLEX128,支持非连续的Tensor数据格式要求为ND。

      • 如果returnComplex=True,out是shape为[N, T]或者[B, N, T]的复数tensor。
      • 如果returnComplex=False,out是shape为[N, T, 2]或者[B, N, T, 2]的实数tensor。

      其中,N=nFft(onesided=False)或者(nFft // 2 + 1)(onesided=True); T是滑动窗口的个数,T = (L - nFft) // hopLength + 1。

  • 返回值:

    返回aclnnStatus状态码,具体参见aclnn返回码

    返回161001(ACLNN_ERR_PARAM_NULLPTR):1. 传入的self、out是空指针
    返回161002(ACLNN_ERR_PARAM_INVALID):1. self的格式不在支持的范围之内;
                                        2. self、windowOptional、out的数据类型不在平台的支持范围之内;
                                        3. nFft、hopLength、winLength输入无效值;
                                        4. self、windowOptional、out的维度不在支持的范围之内;

aclStft

  • 参数说明:
    • workspace(void*,入参):在Device侧申请的workspace内存地址。
    • workspaceSize(uint64_t,入参):在Device侧申请的workspace大小,由第一段接口aclStftGetWorkspaceSize获取。
    • executor(aclOpExecutor*,入参):op执行器,包含了算子计算流程。
    • stream(aclrtStream,入参):指定执行任务的AscendCL stream流。

约束与限制

  • 输入self与PyTorch接口的不同:PyTorch接口的输入self为原始输入;aclStftGetWorkspaceSize的入参self是原始输入经过前端PyTorch补pad后得到的结果;
  • PyTorch接口调用STFT时,self数据类型仅支持FLOAT32、DOUBLE;
  • nFft <= L;
  • winLength <= nFft;
  • 当normalized=True时,STFT(w,m)=1NX[w,m]STFT(w,m)=\frac{1}{\sqrt{N}}X[w,m]
  • self、windowOptional、returnComplex以及out数据类型对应关系如下
Dtype FLOAT32, FLOAT32 DOUBLE, DOUBLE COMPLEX64, COMPLEX64 COMPLEX128, COMPLEX128
returnComplex=True COMPLEX64 COMPLEX128 COMPLEX64 COMPLEX128
returnComplex=False FLOAT32 DOUBLE FLOAT32 DOUBLE

调用示例

#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/acl_stft.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) {
  // 固定写法,AscendCL初始化
  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);
  // 调用aclrtMalloc申请device侧内存
  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);
  // 调用aclrtMemcpy将host侧数据拷贝到device侧内存上
  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);
  // 计算连续tensor的strides
  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];
  }
  // 调用aclCreateTensor接口创建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. (固定写法)device/stream初始化,参考acl对外接口列表
  // 根据自己的实际device填写deviceId
  int32_t deviceId = 0;
  aclrtStream stream;
  auto ret = Init(deviceId, &stream);
  // check根据自己的需要处理
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);
  // 2. 构造输入与输出,需要根据API的接口自定义构造
  std::vector<int64_t> selfShape = {5};
  std::vector<int64_t> windowShape = {4};
  std::vector<int64_t> outShape = {3, 1, 2};
  void* selfDeviceAddr = nullptr;
  void* windowDeviceAddr = nullptr;
  void* outDeviceAddr = nullptr;
  aclTensor* self = nullptr;
  aclTensor* window = nullptr;
  aclTensor* out = nullptr;
  std::vector<float> selfHostData = {1, 6, 8, 5, 7};
  std::vector<float> windowHostData = {1, 1, 1, 1};
  std::vector<float> outHostData = {0, 0, 0, 0, 0, 0};
  // 创建self aclTensor
  ret = CreateAclTensor(selfHostData, selfShape, &selfDeviceAddr, aclDataType::ACL_FLOAT, &self);
  // 创建window aclTensor
  ret = CreateAclTensor(windowHostData, windowShape, &windowDeviceAddr, aclDataType::ACL_FLOAT, &self);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建out aclTensor
  ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_FLOAT, &out);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  int n_fft = 4;
  int hop_length = 2;
  int win_length = 4;
  bool normalized = false;
  bool onesided = true;
  bool returnComplex = false;
  // 3. 调用CANN算子库API,需要修改为具体的Api名称
  uint64_t workspaceSize = 0;
  aclOpExecutor* executor;
  // 调用aclStft第一段接口
  ret = aclStftGetWorkspaceSize(self, window, out, n_fft, hop_length, win_length, normalized, onesided, returnComplex, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclStftGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);
  // 根据第一段接口计算出的workspaceSize申请device内存
  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);
  }
  // 调用aclStft第二段接口
  ret = aclStft(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclStft failed. ERROR: %d\n", ret); return ret);
  // 4. (固定写法)同步等待任务执行结束
  ret = aclrtSynchronizeStream(stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSynchronizeStream failed. ERROR: %d\n", ret); return ret);
  // 5. 获取输出的值,将device侧内存上的结果拷贝至host侧,需要根据具体API的接口定义修改
  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. 释放aclTensor,需要根据具体API的接口定义修改
  aclDestroyTensor(self);
  aclDestroyTensor(window);
  aclDestroyTensor(out);
  // 7. 释放device资源,需要根据具体API的接口定义修改
  aclrtFree(selfDeviceAddr);
  aclrtFree(windowDeviceAddr);
  aclrtFree(outDeviceAddr);
  if (workspaceSize > 0) {
    aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();
  return 0;
}