aclnnLinalgQr
支持的产品型号
- Atlas 训练系列产品。
- 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“aclnnLinalgQrGetWorkspaceSize”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“aclnnLinalgQr”接口执行计算。
aclnnStatus aclnnLinalgQrGetWorkspaceSize(const aclTensor *self, int64_t mode, aclTensor *Q, aclTensor *R, uint64_t *workspaceSize, aclOpExecutor **executor)
aclnnStatus aclnnLinalgQr(void *workspace, uint64_t workspaceSize, aclOpExecutor *executor, aclrtStream stream)
功能描述
算子功能:对输入Tensor进行正交分解。
计算公式:
其中为输入Tensor,维度至少为2, A可以表示为正交矩阵与上三角矩阵的乘积的形式
示例:
A = tensor([[1, 2], [3, 4]], dtype=torch.float) q,r = linalg_qr(A, mode='reduced') q = tensor([[-0.3162, -0.9487], [-0.9487, 0.3162]]) r = tensor([[-3.1623, -4.4272], [0.0000, -0.6325]])
aclnnLinalgQrGetWorkspaceSize
参数说明:
self(const aclTensor *, 计算输入):公式中的,数据类型支持FLOAT、FLOAT16、DOUBLE、COMPLEX64、COMPLEX128。支持非连续的Tensor,数据格式支持ND,shape维度至少为2且不大于8, 且shape需要与Q,R满足约束条件。
mode(int64_t, 计算输入):计算属性,当mode为0时,使用'reduced'(默认)模式,对于输入A(*, m, n), 输出简化大小的Q(*, m, k), R(*, k, n),其中k为m,n的最小值。 当mode为1时,使用'complete'模式,对于输入A(*, m, n),输出完整大小的Q(*, m, m), R(*, m, n), 当mode为2时,使用'r'模式,仅计算reduced场景下的R(*,k,n),其中k为m,n的最小值,返回Q为空tensor。
Q(aclTensor *, 计算输出):公式中的,数据类型支持FLOAT、FLOAT16、DOUBLE、COMPLEX64、COMPLEX128。支持非连续的Tensor,数据格式支持ND,且数据格式需要与self, R一致。shape为Q(*, m, m)或Q(*, m, k)或为空, 其中k为m, n的最小值。
R(aclTensor *, 计算输出): 公式中的,数据类型支持FLOAT、FLOAT16、DOUBLE、COMPLEX64、COMPLEX128。支持非连续的Tensor,数据格式支持ND,且数据格式需要与self, Q一致。shape为R(*, m, n)或R(*, k, n), 其中k为m, n的最小值。
workspaceSize(uint64_t *, 出参):返回需要在Device侧申请的workspace大小。
executor(aclOpExecutor **, 出参):返回op执行器,包含了算子计算流程。
返回值:
aclnnStatus: 返回状态码,具体参见aclnn返回码。
161001(ACLNN_ERR_PARAM_NULLPTR):1. 传入的self、Q或R是空指针。 161002(ACLNN_ERR_PARAM_INVALID):1. self、Q、R的数据类型和数据格式不在支持的范围之内。 2. self、Q、R的shape不符合约束。 3. mode不在可选范围之内。
aclnnLinalgQr
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aclnnLinalgQrGetWorkspaceSize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。
返回值:
aclnnStatus: 返回状态码,具体参见aclnn返回码。
约束与限制
无。
调用示例
示例代码如下,仅供参考,具体编译和执行过程请参考编译与运行样例。
#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_linalg_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) {
// 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_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 = {1, 1, 4, 4};
std::vector<int64_t> qOutShape = {1, 1, 4, 4};
std::vector<int64_t> rOutShape = {1, 1, 4, 4};
void* selfDeviceAddr = nullptr;
void* qOutDeviceAddr = nullptr;
void* rOutDeviceAddr = nullptr;
aclTensor* self = nullptr;
aclTensor* qOut = nullptr;
aclTensor* rOut = nullptr;
std::vector<float> selfHostData = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15};
std::vector<float> qOutHostData = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15};
std::vector<float> rOutHostData = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15};
int64_t mode = 0;
// 创建self aclTensor
ret = CreateAclTensor(selfHostData, selfShape, &selfDeviceAddr, aclDataType::ACL_FLOAT, &self);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// 创建qOut aclTensor
ret = CreateAclTensor(qOutHostData, qOutShape, &qOutDeviceAddr, aclDataType::ACL_FLOAT, &qOut);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// 创建rOut aclTensor
ret = CreateAclTensor(rOutHostData, rOutShape, &rOutDeviceAddr, aclDataType::ACL_FLOAT, &rOut);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// 3. 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aclnnLinalgQr第一段接口
ret = aclnnLinalgQrGetWorkspaceSize(self, mode, qOut, rOut, &workspaceSize, &executor);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnLinalgQrGetWorkspaceSize 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aclnnLinalgQr第二段接口
ret = aclnnLinalgQr(workspaceAddr, workspaceSize, executor, stream);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnLinalgQr 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 size1 = GetShapeSize(qOutShape);
std::vector<float> resultData1(size1, 0);
ret = aclrtMemcpy(resultData1.data(), resultData1.size() * sizeof(resultData1[0]), qOutDeviceAddr,
size1 * sizeof(resultData1[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 < size1; i++) {
LOG_PRINT("result[%ld] is: %f\n", i, resultData1[i]);
}
auto size2 = GetShapeSize(rOutShape);
std::vector<float> resultData2(size2, 0);
ret = aclrtMemcpy(resultData2.data(), resultData2.size() * sizeof(resultData2[0]), rOutDeviceAddr,
size2 * sizeof(resultData2[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 < size2; i++) {
LOG_PRINT("result[%ld] is: %f\n", i, resultData2[i]);
}
// 6. 释放aclTensor和aclScalar,需要根据具体API的接口定义修改
aclDestroyTensor(self);
aclDestroyTensor(qOut);
aclDestroyTensor(rOut);
// 7. 释放Device资源,需要根据具体API的接口定义修改
aclrtFree(selfDeviceAddr);
aclrtFree(qOutDeviceAddr);
aclrtFree(rOutDeviceAddr);
if (workspaceSize > 0) {
aclrtFree(workspaceAddr);
}
aclrtDestroyStream(stream);
aclrtResetDevice(deviceId);
aclFinalize();
return 0;
}