aclnnReluGradV3【免费下载链接】ops-nn本项目是CANN提供的神经网络类计算算子库实现网络在NPU上加速计算。项目地址: https://gitcode.com/cann/ops-nn产品支持情况产品是否支持Atlas A2 训练系列产品/Atlas A2 推理系列产品√功能说明计算ReLU函数的梯度。前向ReLU定义为 $z \max(0, x)$。本算子实现该激活函数对输入 $x$ 的反向梯度计算$grad_input (x 0) \ ? \ grad_output : 0$。函数原型每个算子分为两段式接口必须先调用aclnnReluGradV3GetWorkspaceSize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aclnnReluGradV3接口执行计算。aclnnStatus aclnnReluGradV3GetWorkspaceSize( const aclTensor* x, const aclTensor* y, aclTensor* z, uint64_t* workspaceSize, aclOpExecutor** executor)aclnnStatus aclnnReluGradV3( void *workspace, uint64_t workspaceSize, aclOpExecutor *executor, aclrtStream stream)aclnnReluGradV3GetWorkspaceSize参数说明参数名输入/输出描述使用说明数据类型数据格式维度(shape)非连续TensorxaclTensor*输入前向ReLU的输入。shape决定输出shape。数据类型需为FLOAT / FLOAT16 / BFLOAT16 / INT32 / UINT8 / INT8 / BFLOAT16。FLOAT / FLOAT16 / BFLOAT16 / INT32 / UINT8 / INT8ND1-8√yaclTensor*输入反向传播的梯度输入grad_output。支持与x shape一致或作为单元素标量广播到x的shape。数据类型需为FLOAT / FLOAT16 / BFLOAT16 / INT32 / UINT8 / INT8 / BFLOAT16。FLOAT / FLOAT16 / BFLOAT16 / INT32 / UINT8 / INT8ND1-8√zaclTensor*输出梯度计算结果输出grad_input。shape与x一致。FLOAT / FLOAT16 / BFLOAT16 / INT32 / UINT8 / INT8ND1-8√workspaceSizeuint64_t*输出返回需要在Device侧申请的workspace大小。-----executoraclOpExecutor**输出返回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。第一段接口完成入参校验出现以下场景时报错返回值错误码描述ACLNN_ERR_PARAM_NULLPTR161001传入的x、y或z是空指针。ACLNN_ERR_PARAM_INVALID161002x、y或z的数据类型不在支持的范围之内。x、y和z的shape不一致。aclnnReluGradV3参数说明参数名输入/输出描述workspace输入在Device侧申请的workspace内存地址。workspaceSize输入在Device侧申请的workspace大小由第一段接口aclnnReluGradV3GetWorkspaceSize获取。executor输入op执行器包含了算子计算流程。stream输入指定执行任务的Stream。返回值aclnnStatus返回状态码具体参见aclnn返回码。约束说明确定性计算aclnnReluGradV3默认确定性实现。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 aclnn_relu_grad_v3.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::vectorint64_t shape) { int64_t shapeSize 1; for (auto i : shape) { shapeSize * i; } return shapeSize; } int Init(int32_t deviceId, aclrtStream* stream) { 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); 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); 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); 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]; } *tensor aclCreateTensor(shape.data(), shape.size(), dataType, strides.data(), 0, aclFormat::ACL_FORMAT_ND, shape.data(), shape.size(), *deviceAddr); return 0; } int main() { int32_t deviceId 0; aclrtStream stream; auto ret Init(deviceId, stream); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(Init failed. ERROR: %d\n, ret); return ret); std::vectorint64_t xShape {4, 4}; std::vectorfloat xHostData {0.5f, -0.3f, 2.0f, -1.5f, 0.8f, -0.7f, 1.2f, 0.1f, 1.5f, -2.0f, 0.3f, -0.9f, 0.0f, 1.0f, -1.0f, 0.6f}; std::vectorint64_t yShape {4, 4}; std::vectorfloat yHostData(16, 0.25f); std::vectorint64_t zShape {4, 4}; std::vectorfloat zHostData(16, 0.0f); void* xDeviceAddr nullptr; void* yDeviceAddr nullptr; void* zDeviceAddr nullptr; aclTensor* x nullptr; aclTensor* y nullptr; aclTensor* z nullptr; ret CreateAclTensor(xHostData, xShape, xDeviceAddr, aclDataType::ACL_FLOAT, x); CHECK_RET(ret ACL_SUCCESS, return ret); ret CreateAclTensor(yHostData, yShape, yDeviceAddr, aclDataType::ACL_FLOAT, y); CHECK_RET(ret ACL_SUCCESS, return ret); ret CreateAclTensor(zHostData, zShape, zDeviceAddr, aclDataType::ACL_FLOAT, z); CHECK_RET(ret ACL_SUCCESS, return ret); uint64_t workspaceSize 0; aclOpExecutor* executor; ret aclnnReluGradV3GetWorkspaceSize(x, y, z, workspaceSize, executor); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclnnReluGradV3GetWorkspaceSize failed. ERROR: %d\n, ret); return ret); 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); } ret aclnnReluGradV3(workspaceAddr, workspaceSize, executor, stream); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclnnReluGradV3 failed. ERROR: %d\n, ret); return ret); ret aclrtSynchronizeStream(stream); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclrtSynchronizeStream failed. ERROR: %d\n, ret); return ret); // 打印结果 auto size GetShapeSize(zShape); std::vectorfloat resultData(size, 0); ret aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), zDeviceAddr, size * sizeof(resultData[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 size; i) { float expected (xHostData[i] 0.0f) ? yHostData[i] : 0.0f; LOG_PRINT(result[%ld]%f, expected%f\n, i, resultData[i], expected); } aclDestroyTensor(x); aclDestroyTensor(y); aclDestroyTensor(z); aclrtFree(xDeviceAddr); aclrtFree(yDeviceAddr); aclrtFree(zDeviceAddr); if (workspaceSize 0) { aclrtFree(workspaceAddr); } aclrtDestroyStream(stream); aclrtResetDevice(deviceId); aclFinalize(); return 0; }【免费下载链接】ops-nn本项目是CANN提供的神经网络类计算算子库实现网络在NPU上加速计算。项目地址: https://gitcode.com/cann/ops-nn创作声明：本文部分内容由AI辅助生成（AIGC），仅供参考