CANN/sip BLAS向量复制操作

Copy

【免费下载链接】sip本项目是CANN提供的一款高效、可靠的高性能信号处理算子加速库，基于华为Ascend AI处理器，专门为信号处理领域而设计。项目地址: https://gitcode.com/cann/sip

产品支持情况

功能说明

• 接口功能：
  asdBlasMakeCopyPlan：初始化该句柄对应的Copy算子配置。
  asdBlasScopy：将实数向量的值复制到另外一个向量中。
  asdBlasCcopy：将复数向量的值复制到另外一个向量中。

• 计算公式：

  • asdBlasScopy的公式

  $$ y= x $$

  示例： 输入“x”为： [3, 4] 调用asdBlasScopy算子后，输出“y”为： [3, 4]

  • asdBlasCcopy的公式

  $$ y = x $$

  示例： 输入“x”为： [3+4i,4-3i,4+3i,3-4i] 调用asdBlasCcopy算子后，输出“y”为： [3+4i,4-3i,4+3i,3-4i]

函数原型

AspbStatus asdBlasMakeCopyPlan( asdBlasHandle handle)

AspbStatus asdBlasScopy( asdBlasHandle handle, const int64_t n, aclTensor * x, const int64_t incx, aclTensor * y, const int64_t incy)

AspbStatus asdBlasCcopy( asdBlasHandle handle, const int64_t n, aclTensor * x, const int64_t incx, aclTensor * y, const int64_t incy)

asdBlasMakeCopyPlan

• 参数说明：

  参数名	输入/输出	描述
  handle（asdBlasHandle）	输入	算子的句柄。

• 返回值：

  返回状态码，具体参见SiP返回码。

asdBlasScopy & asdBlasCcopy

• 参数说明：

  参数名	输入/输出	描述
  handle（asdBlasHandle）	输入	算子的句柄。
  n（int64_t）	输入	表示总的元素个数。
  x（aclTensor *）	输入	对应公式中的'x'。 asdBlasScopy的数据类型支持FLOAT32。 asdBlasCcopy的数据类型支持COMPLEX64。 数据格式支持ND。 shape为[n]
  incx（int64_t）	输入	x相邻元素间的内存地址偏移量（当前约束为1）。
  y（aclTensor *）	输出	对应公式中的'y'。 asdBlasScopy的数据类型支持FLOAT32。 asdBlasCcopy的数据类型支持COMPLEX64。 数据格式支持ND。 shape为[n]
  incy（int64_t）	输入	y相邻元素间的内存地址偏移量（当前约束为1）。

• 返回值：

  返回状态码，具体参见SiP返回码。

约束说明

• 输入的元素个数当前覆盖支持[1,6.71e+06]。
• 算子输入shape为[n]，输出shape为[n]。
• 算子实际计算时，不支持ND高维度运算（不支持维度≥3的运算）。

调用示例

示例代码如下，该样例旨在提供快速上手、开发和调试算子的最小化实现，其核心目标是使用最精简的代码展示算子的核心功能，而非提供生产级的安全保障。不推荐用户直接将示例代码作为业务代码，若用户将示例代码应用在自身的真实业务场景中且发生了安全问题，则需用户自行承担。

• asdBlasScopy

#include <iostream> #include <vector> #include "asdsip.h" #include "acl/acl.h" #include "acl_meta.h" using namespace AsdSip; #define ASD_STATUS_CHECK(err) \ do { \ AsdSip::AspbStatus err_ = (err); \ if (err_ != AsdSip::ErrorType::ACL_SUCCESS) { \ std::cout << "Execute failed." << std::endl; \ exit(-1); \ } \ } while (0) #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，acl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侧数据复制到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(int argc, char **argv) { int 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); int64_t n = 7; int64_t incx = 1; int64_t incy = 1; int64_t xSize = 7; int64_t ySize = 7; std::vector<float> tensorInXData; tensorInXData.reserve(xSize); for (int64_t i = 0; i < xSize; i++) { tensorInXData[i] = 1.0 + i; } std::vector<float> tensorOurtYData; tensorOurtYData.reserve(ySize); std::cout << "------- input x -------" << std::endl; for (int64_t i = 0; i < xSize; i++) { std::cout << tensorInXData[i] << " "; } std::cout << std::endl; std::vector<int64_t> xShape = {xSize}; std::vector<int64_t> yShape = {ySize}; aclTensor *inputX = nullptr; aclTensor *outputY = nullptr; void *inputXDeviceAddr = nullptr; void *outputYDeviceAddr = nullptr; ret = CreateAclTensor(tensorInXData, xShape, &inputXDeviceAddr, aclDataType::ACL_FLOAT, &inputX); CHECK_RET(ret == ::ACL_SUCCESS, return ret); ret = CreateAclTensor(tensorOurtYData, yShape, &outputYDeviceAddr, aclDataType::ACL_FLOAT, &outputY); CHECK_RET(ret == ::ACL_SUCCESS, return ret); asdBlasHandle handle; asdBlasCreate(handle); size_t lwork = 0; void *buffer = nullptr; asdBlasMakeCopyPlan(handle); asdBlasGetWorkspaceSize(handle, lwork); std::cout << "lwork = " << lwork << std::endl; if (lwork > 0) { ret = aclrtMalloc(&buffer, static_cast<int64_t>(lwork), ACL_MEM_MALLOC_HUGE_FIRST); CHECK_RET(ret == ::ACL_SUCCESS, LOG_PRINT("allocate workspace failed. ERROR: %d\n", ret); return ret); } asdBlasSetWorkspace(handle, buffer); asdBlasSetStream(handle, stream); ASD_STATUS_CHECK(asdBlasScopy(handle, n, inputX, incx, outputY, incy)); asdBlasSynchronize(handle); asdBlasDestroy(handle); ret = aclrtMemcpy(tensorOurtYData.data(), ySize * sizeof(float), outputYDeviceAddr, ySize * sizeof(float), ACL_MEMCPY_DEVICE_TO_HOST); CHECK_RET(ret == ::ACL_SUCCESS, LOG_PRINT("copy tensor y from device to host failed. ERROR: %d\n", ret); return ret); std::cout << "------- output y -------" << std::endl; for (int64_t i = 0; i < ySize; i++) { std::cout << tensorOurtYData[i] << " "; } std::cout << std::endl; std::cout << "Execute successfully." << std::endl; aclDestroyTensor(inputX); aclDestroyTensor(outputY); aclrtFree(inputXDeviceAddr); aclrtFree(outputYDeviceAddr); aclrtDestroyStream(stream); aclrtResetDevice(deviceId); aclFinalize(); return 0; }

• asdBlasCcopy

#include <iostream> #include <vector> #include <cmath> #include <complex> #include "asdsip.h" #include "acl/acl.h" #include "acl_meta.h" using namespace AsdSip; #define ASD_STATUS_CHECK(err) \ do { \ AsdSip::AspbStatus err_ = (err); \ if (err_ != AsdSip::ErrorType::ACL_SUCCESS) { \ std::cout << "Execute failed." << std::endl; \ exit(-1); \ } else { \ std::cout << "Execute successfully." << std::endl; \ } \ } while (0) #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，acl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侧数据复制到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; } void printTensor(const std::complex<float> *tensorData, int64_t tensorSize) { for (int64_t i = 0; i < tensorSize; i++) { std::cout << tensorData[i] << " "; } std::cout << std::endl; } int main(int argc, char **argv) { int 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); int64_t n = 6; int incx = 1; int incy = 1; int64_t xSize = 6; int64_t ySize = 6; std::vector<std::complex<float>> tensorInXData; tensorInXData.reserve(xSize); for (int64_t i = 0; i < xSize; i++) { tensorInXData[i] = {(float)(1.0 + i), (float)(2.0 + i)}; } std::vector<std::complex<float>> tensorOutYData; tensorOutYData.reserve(ySize); std::cout << "------- input TensorIn -------" << std::endl; printTensor(tensorInXData.data(), xSize); std::vector<int64_t> xShape = {xSize}; std::vector<int64_t> yShape = {ySize}; aclTensor *inputX = nullptr; aclTensor *outputY = nullptr; void *inputXDeviceAddr = nullptr; void *outputYDeviceAddr = nullptr; ret = CreateAclTensor(tensorInXData, xShape, &inputXDeviceAddr, aclDataType::ACL_COMPLEX64, &inputX); CHECK_RET(ret == ::ACL_SUCCESS, return ret); ret = CreateAclTensor(tensorOutYData, yShape, &outputYDeviceAddr, aclDataType::ACL_COMPLEX64, &outputY); CHECK_RET(ret == ::ACL_SUCCESS, return ret); asdBlasHandle handle; asdBlasCreate(handle); size_t lwork = 0; void *buffer = nullptr; asdBlasMakeCopyPlan(handle); asdBlasGetWorkspaceSize(handle, lwork); std::cout << "lwork = " << lwork << std::endl; if (lwork > 0) { ret = aclrtMalloc(&buffer, static_cast<int64_t>(lwork), ACL_MEM_MALLOC_HUGE_FIRST); CHECK_RET(ret == ::ACL_SUCCESS, LOG_PRINT("allocate workspace failed. ERROR: %d\n", ret); return ret); } asdBlasSetWorkspace(handle, buffer); asdBlasSetStream(handle, stream); ASD_STATUS_CHECK(asdBlasCcopy(handle, n, inputX, incx, outputY, incy)); asdBlasSynchronize(handle); asdBlasDestroy(handle); ret = aclrtMemcpy(tensorOutYData.data(), ySize * sizeof(std::complex<float>), outputYDeviceAddr, ySize * sizeof(std::complex<float>), ACL_MEMCPY_DEVICE_TO_HOST); CHECK_RET(ret == ::ACL_SUCCESS, LOG_PRINT("copy tensor y from device to host failed. ERROR: %d\n", ret); return ret); std::cout << "------- output TensorOut -------" << std::endl; printTensor(tensorOutYData.data(), ySize); aclDestroyTensor(inputX); aclDestroyTensor(outputY); aclrtFree(inputXDeviceAddr); aclrtFree(outputYDeviceAddr); aclrtDestroyStream(stream); aclrtResetDevice(deviceId); aclFinalize(); return 0; }

【免费下载链接】sip本项目是CANN提供的一款高效、可靠的高性能信号处理算子加速库，基于华为Ascend AI处理器，专门为信号处理领域而设计。项目地址: https://gitcode.com/cann/sip