mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2025-10-17 22:21:48 +08:00
[Backend] Add RKNPU2 backend support (#456)
* 10-29/14:05 * 新增cmake * 新增rknpu2 backend * 10-29/14:43 * Runtime fd_type新增RKNPU代码 * 10-29/15:02 * 新增ppseg RKNPU2推理代码 * 10-29/15:46 * 新增ppseg RKNPU2 cpp example代码 * 10-29/15:51 * 新增README文档 * 10-29/15:51 * 按照要求修改部分注释以及变量名称 * 10-29/15:51 * 修复重命名之后，cc文件中的部分代码还用旧函数名的bug * 10-29/22:32 * str(Device::NPU)将输出NPU而不是UNKOWN * 修改runtime文件中的注释格式 * 新增Building Summary ENABLE_RKNPU2_BACKEND输出 * pybind新增支持rknpu2 * 新增python编译选项 * 新增PPSeg Python代码 * 新增以及更新各种文档 * 10-30/14:11 * 尝试修复编译cuda时产生的错误 * 10-30/19:27 * 修改CpuName和CoreMask层级 * 修改ppseg rknn推理层级 * 图片将移动到网络进行下载 * 10-30/19:39 * 更新文档 * 10-30/19:39 * 更新文档 * 更新ppseg rknpu2 example中的函数命名方式 * 更新ppseg rknpu2 example为一个cc文件 * 修复disable_normalize_and_permute部分的逻辑错误 * 移除rknpu2初始化时的无用参数 * 10-30/19:39 * 尝试重置python代码 * 10-30/10:16 * rknpu2_config.h文件不再包含rknn_api头文件防止出现导入错误的问题 * 10-31/14:31 * 修改pybind，支持最新的rknpu2 backends * 再次支持ppseg python推理 * 移动cpuname 和 coremask的层级 * 10-31/15:35 * 尝试修复rknpu2导入错误 * 10-31/19:00 * 新增RKNPU2模型导出代码以及其对应的文档 * 更新大量文档错误 * 10-31/19:00 * 现在编译完fastdeploy仓库后无需重新设置RKNN2_TARGET_SOC * 10-31/19:26 * 修改部分错误文档 * 10-31/19:26 * 修复错误删除的部分 * 修复各种错误文档 * 修复FastDeploy.cmake在设置RKNN2_TARGET_SOC错误时，提示错误的信息 * 修复rknpu2_backend.cc中存在的中文注释 * 10-31/20:45 * 删除无用的注释 * 10-31/20:45 * 按照要求修改Device::NPU为Device::RKNPU，硬件将共用valid_hardware_backends * 删除无用注释以及debug代码 * 11-01/09:45 * 更新变量命名方式 * 11-01/10:16 * 修改部分文档，修改函数命名方式 Co-authored-by: Jason <jiangjiajun@baidu.com>
This commit is contained in:
Zheng_Bicheng
@@ -20,12 +20,13 @@
|
||||
|
||||
#pragma once
|
||||
|
||||
#include <algorithm>
|
||||
#include <map>
|
||||
#include <vector>
|
||||
#include <algorithm>
|
||||
|
||||
#include "fastdeploy/backends/backend.h"
|
||||
#include "fastdeploy/utils/perf.h"
|
||||
#include "backends/rknpu/rknpu2/rknpu2_config.h"
|
||||
|
||||
/** \brief All C++ FastDeploy APIs are defined inside this namespace
|
||||
*
|
||||
@@ -34,21 +35,23 @@ namespace fastdeploy {
|
||||
|
||||
/*! Inference backend supported in FastDeploy */
|
||||
enum Backend {
|
||||
UNKNOWN, ///< Unknown inference backend
|
||||
ORT, ///< ONNX Runtime, support Paddle/ONNX format model, CPU / Nvidia GPU
|
||||
TRT, ///< TensorRT, support Paddle/ONNX format model, Nvidia GPU only
|
||||
PDINFER, ///< Paddle Inference, support Paddle format model, CPU / Nvidia GPU
|
||||
POROS, ///< Poros, support TorchScript format model, CPU / Nvidia GPU
|
||||
OPENVINO, ///< Intel OpenVINO, support Paddle/ONNX format, CPU only
|
||||
LITE, ///< Paddle Lite, support Paddle format model, ARM CPU only
|
||||
UNKNOWN, ///< Unknown inference backend
|
||||
ORT, ///< ONNX Runtime, support Paddle/ONNX format model, CPU / Nvidia GPU
|
||||
TRT, ///< TensorRT, support Paddle/ONNX format model, Nvidia GPU only
|
||||
PDINFER, ///< Paddle Inference, support Paddle format model, CPU / Nvidia GPU
|
||||
POROS, ///< Poros, support TorchScript format model, CPU / Nvidia GPU
|
||||
OPENVINO, ///< Intel OpenVINO, support Paddle/ONNX format, CPU only
|
||||
LITE, ///< Paddle Lite, support Paddle format model, ARM CPU only
|
||||
RKNPU2, ///< RKNPU2, support RKNN format model, Rockchip NPU only
|
||||
};
|
||||
|
||||
/*! Deep learning model format */
|
||||
enum ModelFormat {
|
||||
AUTOREC, ///< Auto recognize the model format by model file name
|
||||
PADDLE, ///< Model with paddlepaddle format
|
||||
ONNX, ///< Model with ONNX format
|
||||
TORCHSCRIPT, ///< Model with TorchScript format
|
||||
AUTOREC, ///< Auto recognize the model format by model file name
|
||||
PADDLE, ///< Model with paddlepaddle format
|
||||
ONNX, ///< Model with ONNX format
|
||||
RKNN, ///< Model with RKNN format
|
||||
TORCHSCRIPT, ///< Model with TorchScript format
|
||||
};
|
||||
|
||||
FASTDEPLOY_DECL std::ostream& operator<<(std::ostream& out,
|
||||
@@ -58,12 +61,12 @@ FASTDEPLOY_DECL std::ostream& operator<<(std::ostream& out,
|
||||
|
||||
/*! Paddle Lite power mode for mobile device. */
|
||||
enum LitePowerMode {
|
||||
LITE_POWER_HIGH = 0, ///< Use Lite Backend with high power mode
|
||||
LITE_POWER_LOW = 1, ///< Use Lite Backend with low power mode
|
||||
LITE_POWER_FULL = 2, ///< Use Lite Backend with full power mode
|
||||
LITE_POWER_NO_BIND = 3, ///< Use Lite Backend with no bind power mode
|
||||
LITE_POWER_RAND_HIGH = 4, ///< Use Lite Backend with rand high mode
|
||||
LITE_POWER_RAND_LOW = 5 ///< Use Lite Backend with rand low power mode
|
||||
LITE_POWER_HIGH = 0, ///< Use Lite Backend with high power mode
|
||||
LITE_POWER_LOW = 1, ///< Use Lite Backend with low power mode
|
||||
LITE_POWER_FULL = 2, ///< Use Lite Backend with full power mode
|
||||
LITE_POWER_NO_BIND = 3, ///< Use Lite Backend with no bind power mode
|
||||
LITE_POWER_RAND_HIGH = 4, ///< Use Lite Backend with rand high mode
|
||||
LITE_POWER_RAND_LOW = 5 ///< Use Lite Backend with rand low power mode
|
||||
};
|
||||
|
||||
FASTDEPLOY_DECL std::string Str(const Backend& b);
|
||||
@@ -86,7 +89,7 @@ ModelFormat GuessModelFormat(const std::string& model_file);
|
||||
/*! @brief Option object used when create a new Runtime object
|
||||
*/
|
||||
struct FASTDEPLOY_DECL RuntimeOption {
|
||||
/** \brief Set path of model file and parameter file
|
||||
/** \brief Set path of model file and parameter file
|
||||
*
|
||||
* \param[in] model_path Path of model file, e.g ResNet50/model.pdmodel for Paddle format model / ResNet50/model.onnx for ONNX format model
|
||||
* \param[in] params_path Path of parameter file, this only used when the model format is Paddle, e.g Resnet50/model.pdiparams
|
||||
@@ -102,6 +105,9 @@ struct FASTDEPLOY_DECL RuntimeOption {
|
||||
/// Use Nvidia GPU to inference
|
||||
void UseGpu(int gpu_id = 0);
|
||||
|
||||
void UseRKNPU2(fastdeploy::rknpu2::CpuName rknpu2_name = fastdeploy::rknpu2::CpuName::RK3588,
|
||||
fastdeploy::rknpu2::CoreMask rknpu2_core = fastdeploy::rknpu2::CoreMask::RKNN_NPU_CORE_0);
|
||||
|
||||
void SetExternalStream(void* external_stream);
|
||||
|
||||
/*
|
||||
@@ -173,14 +179,14 @@ struct FASTDEPLOY_DECL RuntimeOption {
|
||||
*/
|
||||
void DisableLiteFP16();
|
||||
|
||||
/**
|
||||
* @brief enable int8 precision while use paddle lite backend
|
||||
*/
|
||||
/**
|
||||
* @brief enable int8 precision while use paddle lite backend
|
||||
*/
|
||||
void EnableLiteInt8();
|
||||
|
||||
/**
|
||||
* @brief disable int8 precision, change to full precision(float32)
|
||||
*/
|
||||
* @brief disable int8 precision, change to full precision(float32)
|
||||
*/
|
||||
void DisableLiteInt8();
|
||||
|
||||
/**
|
||||
@@ -325,9 +331,13 @@ struct FASTDEPLOY_DECL RuntimeOption {
|
||||
int unconst_ops_thres = -1;
|
||||
std::string poros_file = "";
|
||||
|
||||
std::string model_file = ""; // Path of model file
|
||||
std::string params_file = ""; // Path of parameters file, can be empty
|
||||
ModelFormat model_format = ModelFormat::AUTOREC; // format of input model
|
||||
// ======Only for RKNPU2 Backend=======
|
||||
fastdeploy::rknpu2::CpuName rknpu2_cpu_name_ = fastdeploy::rknpu2::CpuName::RK3588;
|
||||
fastdeploy::rknpu2::CoreMask rknpu2_core_mask_ = fastdeploy::rknpu2::CoreMask::RKNN_NPU_CORE_AUTO;
|
||||
|
||||
std::string model_file = ""; // Path of model file
|
||||
std::string params_file = ""; // Path of parameters file, can be empty
|
||||
ModelFormat model_format = ModelFormat::AUTOREC; // format of input model
|
||||
|
||||
// inside parameters, only for inside usage
|
||||
// remove multiclass_nms in Paddle2ONNX
|
||||
@@ -388,6 +398,7 @@ struct FASTDEPLOY_DECL Runtime {
|
||||
void CreateTrtBackend();
|
||||
void CreateOpenVINOBackend();
|
||||
void CreateLiteBackend();
|
||||
void CreateRKNPU2Backend();
|
||||
std::unique_ptr<BaseBackend> backend_;
|
||||
};
|
||||
} // namespace fastdeploy
|
||||
} // namespace fastdeploy
|
||||
|
||||
Reference in New Issue
Block a user