CANN/sip BLAS Ssyr2算子文档

Ssyr2【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asdBlasMakeSsyr2Plan初始化该句柄对应的Ssyr2算子配置。asdBlasSsyr2用于计算单精度向量的外积并将结果加到一个矩阵上。计算公式 $$ Aalphaxy^Talphayx^TA $$ 示例输入“x”为[ 1, 2 ]输入“A”为[ [ 1,2 ],[ 3,4 ] ]输入“y”为[ 3, 4 ]输入“uplo”为U输入“n”为2输入“lda”为2输入“incx”为1输入“incy”为1。输入“alpha”为2.345。调用“asdBlasSsyr2”算子后输出“A”为[ [ 15.07, 25.45 ], [ 26.45, 41.52 ] ].函数原型AspbStatus asdBlasMakeSsyr2Plan( asdBlasHandle handle)AspbStatus asdBlasSsyr2( asdBlasHandle handle, asdBlasFillMode_t uplo, const int64_t n, const float * alpha, aclTensor * x, int64_t incx, aclTensor * y, int64_t incy, aclTensor * A, const int64_t lda)asdBlasMakeSsyr2Plan参数说明参数名输入/输出描述handleasdBlasHandle输入算子的句柄返回值返回状态码具体参见SiP返回码。asdBlasSsyr2参数说明参数名输入/输出描述handleasdBlasHandle输入算子的句柄uploasdBlasFillMode_t输入指定矩阵A的存储格式。ASDBLAS_FILL_MODE_LOWER:下三角ASDBLAS_FILL_MODE_UPPER:上三角nint64_t输入向量x中的元素个数矩阵A的行列数。alphafloat *输入公式中的alpha标量向量乘积缩放因子。AaclTensor *输入/输出对应公式中的A。数据类型支持FLOAT32。数据格式支持ND。shape为[nn]xaclTensor *输入对应公式中的x。数据类型支持FLOAT32。数据格式支持ND。shape为[n]ldaint64_t输入矩阵A的每列元素的存储步长当前约束为n。incxint64_t输入x相邻元素间的内存地址偏移量当前约束为1。yaclTensor *输入对应公式中的y。数据类型支持FLOAT32。数据格式支持ND。shape为[n]incyint64_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n当前覆盖支持[18192]。算子输入shape为[n]、[n]、[nn]输出shape为[nn]。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。#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::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; } 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 5; asdBlasFillMode_t uplo asdBlasFillMode_t::ASDBLAS_FILL_MODE_LOWER; float alpha 2.0f; int64_t incx 1; int64_t incy 1; int64_t lda 5; const int64_t tensorXSize n; std::vectorfloat tensorInXData; tensorInXData.reserve(tensorXSize); for (int i 0; i tensorXSize; i) { tensorInXData.push_back(1.0 i); } const int64_t tensorYSize n; std::vectorfloat tensorInYData; tensorInYData.reserve(tensorYSize); for (int i 0; i tensorYSize; i) { tensorInYData.push_back(2.0f i); } const int64_t tensorASize n * n; std::vectorfloat tensorInAData; tensorInAData.reserve(tensorYSize); for (int i 0; i tensorASize; i) { tensorInAData.push_back(1.0f); } std::cout alpha static_castint32_t(alpha) std::endl; std::cout uplo static_castint32_t(uplo) std::endl; std::cout ------- input X ------- std::endl; for (int64_t i 0; i tensorXSize; i) { std::cout tensorInXData[i] ; } std::cout std::endl; std::cout ------- input Y ------- std::endl; for (int64_t i 0; i tensorYSize; i) { std::cout tensorInYData[i] ; } std::cout std::endl; std::cout ------- input A ------- std::endl; for (int64_t i 0; i n; i) { for (int64_t j 0; j n; j) std::cout tensorInAData[i * n j] ; std::cout std::endl; } std::vectorint64_t xShape {n}; std::vectorint64_t yShape {n}; std::vectorint64_t matAShape {n, n}; aclTensor *inputX nullptr; aclTensor *inputY nullptr; aclTensor *inputA nullptr; void *inputXDeviceAddr nullptr; void *inputYDeviceAddr nullptr; void *inputADeviceAddr nullptr; ret CreateAclTensorfloat(tensorInXData, xShape, inputXDeviceAddr, aclDataType::ACL_FLOAT, inputX); CHECK_RET(ret ::ACL_SUCCESS, return ret); ret CreateAclTensorfloat(tensorInYData, yShape, inputYDeviceAddr, aclDataType::ACL_FLOAT, inputY); CHECK_RET(ret ::ACL_SUCCESS, return ret); ret CreateAclTensorfloat(tensorInAData, matAShape, inputADeviceAddr, aclDataType::ACL_FLOAT, inputA); CHECK_RET(ret ::ACL_SUCCESS, return ret); asdBlasHandle handle; asdBlasCreate(handle); size_t lwork 0; void *buffer nullptr; asdBlasMakeSsyr2Plan(handle); asdBlasGetWorkspaceSize(handle, lwork); 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); } asdBlasSetWorkspace(handle, buffer); asdBlasSetStream(handle, stream); ASD_STATUS_CHECK(asdBlasSsyr2(handle, uplo, n, alpha, inputX, incx, inputY, incy, inputA, lda)); asdBlasSynchronize(handle); asdBlasDestroy(handle); ret aclrtMemcpy(tensorInAData.data(), n * n * sizeof(float), inputADeviceAddr, n * n * sizeof(float), 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 ------- output A ------- std::endl; for (int64_t i 0; i n; i) { for (int64_t j 0; j n; j) { std::cout tensorInAData[i * n j] ; } std::cout std::endl; } std::cout Execute successfully. std::endl; aclDestroyTensor(inputX); aclDestroyTensor(inputY); aclDestroyTensor(inputA); aclrtFree(inputXDeviceAddr); aclrtFree(inputYDeviceAddr); aclrtFree(inputADeviceAddr); 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