Always
Always
- 设备管理模块 - 新增OpenCL的设备管理模块 - 新增ROCM的设备管理模块 - 新增OpenGL的设备管理模块 - 内存优化 - `主从内存拷贝优化`:针对统一内存的架构,通过主从内存映射、主从内存地址共享等方式替代主从内存拷贝 - `内存池`:针对nndeploy的内部的数据容器Buffer、Mat、Tensor,建立异构设备的内存池,实现高性能的内存分配与释放 - `多节点共享内存机制`:针对多模型串联场景下,基于模型部署的有向无环图,在串行执行的模式下,支持多推理节点共享内存机制 - `边的环形队列内存复用机制`:基于模型部署的有向无环图,在流水线并行执行的模式下,支持边的环形队列共享内存机制 - 有向无环图的完善和诱惑 - 多端推理模块的优化和完善
## What is this Python project? nndeploy is an open-source AI deployment framework focused on solving the pain points from AI algorithm development to production deployment. Key features include: -...
# nndeploy: An Easy-to-Use and High-Performance AI Deployment Framework Hello, it's our honor to recommend the nndeploy project to you. nndeploy is an easy-to-use and high-performance AI deployment framework, primarily...
# nndeploy: An Easy-to-Use and High-Performance AI Deployment Framework Hello, it's our honor to recommend the nndeploy project to you. nndeploy is an easy-to-use and high-performance AI deployment framework, primarily...
# nndeploy:An Easy-to-Use and High-Performance AI deployment framework Hello, it's our honor to recommend the nndeploy project to you. nndeploy is an easy-to-use and high-performance AI deployment framework. Based on...
# nndeploy:一款简单易用和高性能的AI部署框架 您好,很荣幸跟您推荐 nndeploy 这个项目。 nndeploy是一款简单易用和高性能的AI部署框架,由C++17为主语言。基于可视化工作流和多端推理,开发者可以快速从算法仓库开发出指定平台和硬件所需的SDK,大幅节省开发时间。此外,框架已部署包括 LLM、AIGC 生成、换脸、目标检测、图像分割等众多 AI 模型,开箱即用。