每个算子有两段接口,必须先调用“aclnnXxxGetWorkspaceSize”接口获取入参并根据计算流程计算所需workspace大小,再调用“aclnnXxx”接口执行计算。两段式接口如下:
aclnnStatus aclnnConvolutionGetWorkspaceSize( const aclTensor *input, const aclTensor *weight, const aclTensor *bias, const aclIntArray *stride, const aclIntArray *padding, const aclIntArray *dilation, const bool transposed, const aclIntArray *outputPadding, const int64_t groups, aclTensor *output, int8_t cubeMathType, uint64_t *workspaceSize, aclOpExecutor **executor)
返回aclnnStatus状态码,具体参见aclnn返回码。
第一段接口完成入参校验,出现以下场景时报错:
aclnnStatus aclnnConvolution(const void * const workspace, const uint64_t workspaceSize, aclOpExecutor *executor, const aclrtStream stream)
返回aclnnStatus状态码,具体参见aclnn返回码。
对于Atlas 训练系列产品,Cube单元不支持FLOAT32计算。当输入为FLOAT32,可通过设置cubeMathType=1(ALLOW_FP32_DOWN_PRECISION)来允许接口内部cast到FLOAT16进行计算。
#include <iostream> #include <vector> #include "acl/acl.h" #include "aclnnop/level2/aclnn_convolution.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 shape_size = 1; for (auto i : shape) { shape_size *= i; } return shape_size; } int Init(int32_t deviceId, aclrtContext* context, aclrtStream* stream) { // 固定写法,acl初始化 auto ret = aclrtSetDevice(deviceId); CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSetDevice failed. ERROR: %d\n", ret); return ret); ret = aclrtCreateContext(context, deviceId); CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtCreateContext failed. ERROR: %d\n", ret); return ret); ret = aclrtSetCurrentContext(*context); CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSetCurrentContext 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); ret = aclInit(nullptr); CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclInit 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_NCHW, shape.data(), shape.size(), *deviceAddr); return 0; } int main() { // 1. (固定写法)device/context/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; aclrtContext context; aclrtStream stream; auto ret = Init(deviceId, &context, &stream); // check根据自己的需要处理 CHECK_RET(ret == 0, 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> shapeInput = {2, 2, 2, 2}; std::vector<int64_t> shapeWeight = {1, 2, 1, 1}; std::vector<int64_t> shapeResult = {2, 1, 4, 4}; std::vector<int64_t> convStrides; std::vector<int64_t> convPads; std::vector<int64_t> convOutPads; std::vector<int64_t> convDilations; void *deviceDataA = nullptr; void *deviceDataB = nullptr; void *deviceDataResult = nullptr; aclTensor* input = nullptr; aclTensor* weight = nullptr; aclTensor* result = nullptr; std::vector<float> inputData(GetShapeSize(shapeInput) * 2, 1); std::vector<float> weightData(GetShapeSize(shapeWeight) * 2, 1); std::vector<float> outputData(GetShapeSize(shapeResult) * 2, 1); convStrides = {1, 1, 1, 1}; convPads = {1, 1, 1, 1}; convOutPads = {1, 1, 1, 1}; convDilations = {1, 1, 1, 1}; // 创建self aclTensor ret = CreateAclTensor(inputData, shapeInput, &deviceDataA, aclDataType::ACL_FLOAT16, &input); CHECK_RET(ret == ACL_SUCCESS, return ret); // 创建other aclTensor ret = CreateAclTensor(weightData, shapeWeight, &deviceDataB, aclDataType::ACL_FLOAT16, &weight); CHECK_RET(ret == ACL_SUCCESS, return ret); // 创建out aclTensor ret = CreateAclTensor(outputData, shapeResult, &deviceDataResult, aclDataType::ACL_FLOAT16, &result); CHECK_RET(ret == ACL_SUCCESS, return ret); aclIntArray *strides = aclCreateIntArray(convStrides.data(), 2); aclIntArray *pads = aclCreateIntArray(convPads.data(), 2); aclIntArray *outPads = aclCreateIntArray(convOutPads.data(), 2); aclIntArray *dilations = aclCreateIntArray(convDilations.data(), 2); // 3. 调用CANN算子库API,需要修改为具体的HostApi uint64_t workspaceSize = 0; aclOpExecutor* executor; // 调用aclnnConvolution第一段接口 ret = aclnnConvolutionGetWorkspaceSize( input, weight, nullptr, strides, pads, dilations, false, outPads, 1, result, 1, &workspaceSize, &executor); CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnConvolutionGetWorkspaceSize 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aclnnConvolution第二段接口 ret = aclnnConvolution(workspaceAddr, workspaceSize, executor, stream); CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnConvolution 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(shapeResult); std::vector<float> resultData(size, 0); ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), deviceDataResult, size * 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); for (int64_t i = 0; i < size; i++) { LOG_PRINT("result[%ld] is: %f\n", i, resultData[i]); } // 6. 释放aclTensor和aclScalar,需要根据具体API的接口定义修改 aclDestroyTensor(input); aclDestroyTensor(weight); aclDestroyTensor(result); aclDestroyIntArray(strides); aclDestroyIntArray(pads); aclDestroyIntArray(outPads); aclDestroyIntArray(dilations); return 0; }