oneAPI DevSummit China 2023 Autumn, with OpenVINO

November 3, 2023
November 3rd: 9:30-17:00 CST, November 4th: 13:00-18:30 CST
Register See agenda


Whether you’re an AI developer, C++ or GPU programmer, IoT developer, or data scientist in HPC and AI, if you’re interested in high-performance AI and accelerated computing across architectures, you should definitely check out this event. Intel will be holding its oneAPI DevSummit China 2023 jointly with OpenVINO DevCon, on November 3-4. In this conference, you will be able to learn about the latest trends in the industry, hear guests from famous companies and institutions share various cases, get new oneAPI Hackathon information and hear winners share their stories. You’ll also be able to participate in practical training courses in which to learn and optimize applications, and take full advantage of device capabilities on both CPU and GPU. In addition, we have also prepared a wealth of gifts for participants. Come and join us!


November 3rd
November 4th

November 3rd

9:30 - 9:45 CST Keynote

Welcome Speech: oneAPI State of the Union

Introducing the current status of the oneAPI ecosystem and Intel’s latest related projects. Learn how the ecosystem is expanding and how users are leveraging the power of oneAPI. Understand the current status and development direction of oneAPI, and how to maximize its value for you in Istanbul.

9:45 - 10:15 CST Keynote

Making an Open Future for Accelerators

There is strong momentum behind having open standards and open goverance for accelerator software development. There is a lot of exciting news for open accelerator development, including UXL, new Intel tools (2024.0), a second edition of the SYCL book (released a few weeks ago). Many companies and institutions are coming together to help realize true open standards and software for accelerators. The drive for open…

10:15 - 10:35 CST Tech Talk

Accelerating Medical Large-scale Model Based on SPR and AMX

Introduces the typical application scenarios of medical large models – medical record abstraction and generation, and how to use the CPU to accelerate medical large model inference based on Intel Sapphire Rappids and AMX technology to enhance the implementation capabilities of large models. Download Presentation

10:35 - 10:55 CST Tech Talk

oneAPI Accelerates AI-based Encrypted Traffic Recognition

EverSec is a cybersecurity service company with intelligence systems and digital security protection. Traffic detection software is the core TECH of cybersecurity technology. The latest version of ENTA, V3.0, contains more than 1000 functional modules and more than 40 million lines of code. Intel compiler is the main building tool of ENTA, which provides a high-performance base for ENTA software. Both traditional AI training and…

10:55 - 11:00 CST Break

11:00 - 11:20 CST Tech Talk

Accelerate Crystal Structure Comparison with oneAPI

Our in-house C++ software, Ilum, has been extensively applied to our inner CSP process for structure similarity comparisons. However, the efficiency of this software previously constrained the time efficiency of CSP. Now, with the assistance of oneAPI Toolkits, Ilum boasts a remarkable 3x to 10x increase in speed, substantially reducing the computational cost. Download Presentation

11:20 - 11:40 CST Tech Talk

Improved BSPADMM Based on ADMM Algorithm

Porting and optimization of BSPADMM on Intel GPU and some experience using oneAPI. Download Presentation

11:40 - 12:00 CST Tech Talk

Explore the Combination of Industry and Education of C++/SYCL and Parallel Programming Course

Currently, parallel computer architectures are in full bloom, and workloads at the application level are becoming increasingly diverse. In university computer education, parallel programming courses based on traditional parallel programming models and development tools need to teach students parallel programming and optimization methods on different architectures so that students can be flexible (combination) when facing multiple architectures and be able to use a variety of…

12:00 - 12:05 CST Break

12:05 - 13:00 CST Lunch

13:00 - 13:30 CST

Fireside Chat: OpenVINO 2023.1 Release, Accelerating Generative AI on Client and Edge Devices

OpenVINO™ toolkit’s newest release, 2023.1, was announced at the end of September. Join to hear the most exciting updates articulated by Intel senior technical leaders.

13:30 - 14:30 CST

OpenVINO™ 2023.1 New Version Analysis, GenAl for Everyone

This presentation will introduce the new features of OpenVINO 2023.1 version, aiming to help developers understand how to utilize OpenVINO to lower the barrier for deploying large models and improve their performance on edge devices. Download Presentation

14:30 - 14:35 CST Break

14:35 - 15:05 CST

Hugging face + OpenVINO™ Co-Accelerate to Build GenAl

Content Introduction: 1. Introduction to Hugging Face. 2. How to use the optimum-intel tool to accelerate generative AI.

15:05 - 15:35 CST

Code with Efficiency | Boost Your Potential with Github Copilot and OpenVINO™

Have you been using the magic of GitHub Copilot to write programs? In this course, we will introduce you to how GitHub Copilot can enhance your productivity in code writing. We will also discuss how to create your own customized Copilot application tailored for enterprise scenarios. Download Presentation

15:35 - 16:05 CST

Accelerate the Deployment Scheme of YOLOv8 Based on OpenVINO™

The YOLOv8 model and its acceleration deployment solution with OpenVINO will be introduced. This includes an overview of the YOLOv8 model framework, key improvements in model structure, model deployment and acceleration techniques, as well as case demonstrations. Download Presentation

16:05 - 16:35 CST

Paddlepaddle and OpenVINO™ Co-Create Open AI Ecosystem

This presentation will introduce the collaborative roadmap between Baidu PaddlePaddle and Intel OpenVINO over the past three years, as well as the best application practices based on PaddleOCR and OpenVINO. Both parties have made significant progress in technology and have jointly launched a series of ecosystem activities, including the establishment of the OpenVINO section on Baidu AI Studio Xinghe Community, online and offline seminars for…

16:35 - 16:50 CST Q&A for Afternoon Sessions

16:50 - 17:00 CST Survey & Lucky Draw

November 4th

13:00 - 13:15 CST Tech Talk

oneAPI &OpenVINO™ Hackathon Introduction

Along with this DevSummit, we will simultaneously launch the oneAPI Hackathon Competition, aiming to encourage more developers to experience and use oneAPI and OpenVINO. The competition includes topics on AI and SYCL. Download Presentation

13:15 - 13:25 CST Lightning Talk

Work Sharing: Object Detection for Autonomous Vehicles

This work sharing is about the research results of object detection technology for autonomous vehicles. In this work, we first conducted an in-depth study of related algorithms and technologies for target detection, including deep learning-based convolutional neural networks (CNN) and target detection algorithms (YOLO, Faster R-CNN, SSD, etc.). Then, we optimized and improved the algorithm for the special needs of autonomous vehicles. By performing a…

13:25 - 13:35 CST Lightning Talk

Sequence Alignment Software Migration and Performance Evaluation Based on DPCT

Use the DPCT conversion tool and GASAL2 sequence alignment software to explore the process of migrating CUDA programs to DPC++, and verify the functional consistency and performance in different architectures before and after migration. Download Presentation

13:35 - 14:30 CST Lightning Talk

OpenVINO™ Community Contributions and Award-Winning Case Studies Sharing

We Invite past winners of the OpenVINO™ competition to share their development experiences and techniques, helping developers enhance their skills. Download Presentation – Haojun Pang Download Presentation – Liqi Wang Download Presentation – Yiqiao Zhang

14:30 - 14:35 CST Break

14:35 - 15:20 CST Tech Talk

oneAPI & OpenVINO™ Joint Hackathon Technical Training

This presentation will be targeted towards potential participants of the oneAPI & OpenVINO Joint Hackathon. It will provide a detailed introduction to the basic requirements of model optimization tasks and offer key technical training related to this task. Download Presentation

15:20 - 15:25 CST Break

15:25 - 16:40 CST Workshop

Unleash the Power of Intel® Xpus to Accelerate AI by Using Intel Optimized AI Frameworks

As AI applications increase their demand for computing power, CPUs, GPUs and dedicated acceleration chips can all work together to enhance system computing power through heterogeneous computing. We collectively call them XPU. Intel continues to add new features to enhance AI computing in newly launched CPUs and GPUs, such as Intel® Deep Learning Boost, Intel® Advanced Matrix Extensions, and Intel® Xe Matrix Extensions technologies. These…

16:40 - 16:45 CST Break

16:45 - 18:15 CST Workshop

oneAPI Programming Pattern for Heterogeneous Computing

Introducing Data Parallel C++ (DPC++), a programming language that supports heterogeneous computing implemented based on the SYCL specification based on open standards, as well as the core components of the basic tool suite in the Intel® oneAPI series and the tool suite for HPC-specific workloads . At the same time, the DPC++ program structure and operation examples and code analysis of several sample codes are…

18:15 - 18:30 CST Survey & Lucky Draw


Learn about joining the UXL Foundation:

Join now