CANN/sip卷积滤波算子API文档

asdConvolve【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产品支持情况产品是否支持Atlas 200I/500 A2 推理产品×Atlas 推理系列产品×Atlas 训练系列产品×Atlas A3 训练系列产品/Atlas A3 推理系列产品√Atlas A2 训练系列产品/Atlas A2 推理系列产品√Ascend 950PR/Ascend 950DT×功能说明接口功能对给定的信号进行一维滤波操作。计算公式 $$ w(k)\sum _{j}u(j)v(k-j1) $$ 其中w(k)为输出的k位置的元素u(j)为输入位置为j的一维信号v(k-j1)为位置为k-j1的滤波卷积核。一维信号为复数向量滤波卷积核为实数向量。示例 输入“u”为 [[1.1.j 2.2.j] [1.1.j 2.2.j]] 输入“v”为 [1. 2. 3. 4.] 调用“asdConvolve”算子后输出“result”为 [[4.4.j, 7.7.j], [4.4.j, 7.7.j]]函数原型AspbStatus asdConvolve( const aclTensor * signal, const aclTensor * kernel, aclTensor * output, asdConvolveMode_t mode, void *stream, void * workspace)asdConvolve参数说明参数名输入/输出描述signalaclTensor *输入输入的一维信号。支持的数据类型为COMPLEX32、COMPLEX64。输入信号shape为[BatchCount, n]。kernelaclTensor *输入输入的滤波卷积核。支持的数据类型为FLOAT16、FLOAT32。输入滤波卷积核shape为[k]。outputaclTensor *输入/输出输入/输出信号。支持的数据类型为COMPLEX32、COMPLEX64。输出shape与输入shape保持一致。modeasdConvolveMode_t输入滤波卷积模式当前仅支持ASD_CONVOLVE_SAME即输入和输出的向量维度保持一致。streamvoid*输入算子执行时的stream。workspacevoid*输入算子所需的Workspace指针。返回值返回状态码具体参见SiP返回码。约束说明无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。#include iostream #include asdsip.h #include filter_api.h #include acl/acl.h #include acl/acl_base.h #include acl_meta.h #include complex #include vector using namespace AsdSip; using half op::fp16_t; #define ASD_STATUS_CHECK(err) \ do { \ AsdSip::AspbStatus err_ (err); \ if (err_ ! AsdSip::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::vectorint64_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::vectorT hostData, const std::vectorint64_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::vectorint64_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; } template typename T void printTensor(std::vectorT 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 signalLen 128; // 26208 int64_t kernelLen 32; int64_t batchCount 2; // 768 std::vectorstd::complexhalf tensorSignalData; tensorSignalData.reserve(signalLen * batchCount); std::vectorhalf tensorKernelData; tensorKernelData.reserve(kernelLen); for (int64_t i 0; i signalLen * batchCount; i) { tensorSignalData[i] {(half)1.0, (half)1.0}; } for (int64_t i 0; i kernelLen; i) { tensorKernelData[i] (half)(1.0 i); // tensorKernelData[i] 1.0; } std::vectorstd::complexhalf tensorOutData; tensorOutData.reserve(signalLen * batchCount); for (int64_t i 0; i signalLen * batchCount; i) { tensorOutData[i] {(half)-1.0, (half)-1.0}; } std::vectorint64_t signalShape {batchCount, signalLen}; std::vectorint64_t kernelShape {kernelLen}; std::vectorint64_t resultShape {batchCount, signalLen}; aclTensor *signal nullptr; aclTensor *kernel nullptr; aclTensor *output nullptr; void *signalDeviceAddr nullptr; void *kernelDeviceAddr nullptr; void *outputDeviceAddr nullptr; ret CreateAclTensorstd::complexhalf( tensorSignalData, signalShape, signalDeviceAddr, aclDataType::ACL_COMPLEX32, signal); CHECK_RET(ret ::ACL_SUCCESS, return ret); ret CreateAclTensorhalf( tensorKernelData, kernelShape, kernelDeviceAddr, aclDataType::ACL_FLOAT16, kernel); CHECK_RET(ret ::ACL_SUCCESS, return ret); ret CreateAclTensorstd::complexhalf( tensorOutData, resultShape, outputDeviceAddr, aclDataType::ACL_COMPLEX32, output); CHECK_RET(ret ::ACL_SUCCESS, return ret); size_t lwork 0; AsdSip::asdConvolveGetWorkspaceSize(signalLen, kernelLen, lwork); void *buffer nullptr; std::cout lwork lwork std::endl; if (lwork 0) { ret aclrtMalloc(buffer, static_castint64_t(lwork), ACL_MEM_MALLOC_HUGE_FIRST); CHECK_RET(ret ::ACL_SUCCESS, LOG_PRINT(allocate workspace failed. ERROR: %d\n, ret); return ret); } ASD_STATUS_CHECK(AsdSip::asdConvolve(signal, kernel, output, asdConvolveMode_t::ASD_CONVOLVE_SAME, stream, buffer)); ret aclrtSynchronizeStream(stream); CHECK_RET(ret ::ACL_SUCCESS, LOG_PRINT(aclrtSynchronizeStream failed. ERROR: %d\n, ret); return ret); ret aclrtMemcpy(tensorOutData.data(), signalLen * batchCount * sizeof(std::complexhalf), outputDeviceAddr, signalLen * batchCount * sizeof(std::complexhalf), 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); std::cout ------- result ------- std::endl; for (int batchIdx 0; batchIdx batchCount; batchIdx) { for (int i 0; i signalLen; i) { std::cout ( (float)tensorOutData[batchIdx * signalLen i].real() , (float)tensorOutData[batchIdx * signalLen i].imag() ) ; } std::cout std::endl; } aclDestroyTensor(signal); aclDestroyTensor(kernel); aclDestroyTensor(output); aclrtFree(signalDeviceAddr); aclrtFree(kernelDeviceAddr); aclrtFree(outputDeviceAddr); 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创作声明：本文部分内容由AI辅助生成（AIGC），仅供参考