This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>
- Enterprise
Enterprise products, solutions & services
-
Huawei Cloud
-
Carrier
-
Consumer
-
Corporate
Winners of The OlympusMons Awards 2023
About the OlympusMons Awards
As a new factor of production, data has become a basic strategic resource in today's digital economy. However, explosive data growth and innovative data applications place higher requirements on data infrastructure. Sufficient storage capacity, premium usability, higher security, and better energy efficiency are now key priorities for industry, academia, and research organizations. In addition to ever-changing requirements for IT architecture, how to build data infrastructure with maximum energy efficiency is a major challenge for the industry.
Building a technological ecosystem, especially in the field of basic technology breakthroughs, requires the collaboration of all parties. That's why Huawei established the annual OlympusMons Awards in 2019 to lead the global research of data storage basic theories, break through key technical problems, accelerate the industrialization of scientific research achievements, and achieve industry-academia-research win-win collaboration.
Learn MoreChallenges of the
OlympusMons Awards 2023
- Challenge 1
Storage Technologies That Deliver Ultimate Per-Bit Cost Efficiency
With the emergence and development of new non-volatile media and high-speed network protocols, further research needs to be done on storage architecture, media application, and storage media to build a storage system that delivers ultimate per-bit cost efficiency.
Learn More
- Challenge 2
Data Enablement and Resilience Technologies for Emerging Services
To handle the diverse range of devices and new application workloads in data centers, high-speed data processing requirements, and new trends in data resilience, research needs to be done into technologies like data enablement and governance for new scenarios such as AI, data resilience and high reliability. This will enable us to build data enablement and resilience technologies for emerging services.
Learn More
OlympusMons Awarding Results
Winners of OlympusMons Pioneer Award 
Winners of OlympusMons Award Winners of OlympusMons Pioneer Award Winners of OlympusMons Award The 2024 Global Data Storage Expert Forum announced the shortlist for the 2023 OlympusMons awards. The OlympusMons Award was presented to Professor Torsten Hoefler and his team from ETH Zurich, and to Professor Guo Minyi and his team from Shanghai Jiao Tong University.
The OlympusMons Pioneer Award was awarded to four teams: Professor You Yang (National University of Singapore), Professor John Kim (Korea Advanced Institute of Science & Technology), Mr. Yang Tong (Peking University), and Mr. Wang Ying (Institute of Computing Technology of the Chinese Academy of Sciences). The winning teams are determined by a review committee, who evaluated all entries based on standards of innovation, advancement, generality, and feasibility.
OlympusMons Award
Research Bonus: 1,000,000 RMB-
- Achievements：
- Network-Accelerated Distributed Storage Systems and Databases
-
- Major team members：
- ETH Zurich
-
-
Torsten Hoefler
Maciej Besta
Salvatore Di Girolamo
Timo Schneider
-
- Achievement introduction：
- Prof. Torsten Hoefler's team's holistic approach covers all layers of the computing stack, driving significant improvements in efficiency and performance. At the forefront of their achievements is the groundbreaking work in network support and acceleration for distributed storage systems. They have pioneered innovative data transfer and processing solutions, such as streaming Processing in the Network (sPIN), that have significantly improved overall system performance. Their efforts in developing middleware and infrastructure for distributed storage systems have laid a foundation for seamless integration and scalability. This facilitates the management of vast volumes of data across distributed environments. The team's contributions extend to the development of models, abstractions, and paradigms for distributed storage systems, providing frameworks for streamlined data management and access. One of their standout innovations is their approach to Storage Architecture Innovation, with which they have successfully offloaded entire disaggregated distributed file systems to processing-enhanced network interface cards. This breakthrough not only optimizes storage performance, but enhances overall system efficiency and scalability. Moreover, their advancements in storage networking, encompassing both endpoint and topology considerations, are poised to revolutionize future systems, paving the way for unprecedented levels of connectivity and data accessibility.
OlympusMons Award
Research Bonus: 1,000,000 RMB-
- Achievements：
- Training/Inference Convergence Data Foundation Technologies
-
- Major team members：
- Shanghai Jiao Tong University
-
-
Guo Minyi
Leng Jingwen
Guo Cong
-
- Achievement introduction：
- To address growing challenges in training efficiency and integrated training and inference of large AI models, this achievement innovatively proposed a research idea centered on efficient data management. It deeply analyzed features such as data access, value distribution, and high-order information during model training, inference, and training/inference conversion. Using GPU memory pool management, model weight encoding, and low-bit-width compression of full-precision models, this team developed a comprehensive technical system for efficient model fine-tuning, fast model compression, and low-bit-width inference deployment for training and inference convergence. This reduced memory training requirements by over 30%, achieved low-bit-width (4-bit) inference in scenarios that do not require fine-tuning of data, and slashed the time needed to convert a full-precision model to a low-bit-width model to seconds, nearly 10,000 times faster. The team members have published multiple papers and journals, and won an IEEE Micro Top Picks Honorable Mention. Their software prototypes and hardware design have been made open-source and widely applied, providing innovative ideas for building easy-to-use and high-performance data foundations for training and inference convergence.
OlympusMons Pioneer Award
Research Bonus: 200,000 RMB-
- Achievements：
- Colossal-AI: A Deep Learning System for Large AI Models
-
- Major team members：
- National University of Singapore
-
-
You Yang
-
- Achievement introduction：
- This achievement has furthered the progress of AI in the large AI model era. The research of Professor You Yang focuses on the Colossal-AI system and addresses the key challenge of efficiently performing large AI model training and inference tasks. The Colossal-AI system has made innovations in algorithm optimization, system architecture design, and real-world application. The system integrates not only advanced AI technologies, but the latest progress in fields such as large language models, image and video generation, and high-performance computing. This significantly improves hardware utilization and reduces the development and application costs of large AI models. It can be used to build foundation models and industry-specific models to fully unleash the power of AI.
OlympusMons Pioneer Award
Research Bonus: 200,000 RMB-
- Achievements：
- Communication-Centric Architecture
-
- Major team members：
- Korea Advanced Institute of Science & Technology
-
-
John Kim
-
- Achievement introduction：
- Communication—or the movement of data—has become a bottleneck in modern digital systems. The amount of computation required continues to increase as workloads grow. This has resulted in the wide adoption of high-performance processors (or accelerators) and high bandwidth (and high capacity) memory/storage systems. This has created a bottleneck in modern systems in the form of communication between the different components (the movement of data), both within a system (i.e., internal data movement) and between different components in a system (i.e., external data movement). This team's research focused on how a communication-centric approach can enable an efficient system architecture in which the movement of data (or communication) is minimized. The team's communication-centric approach reimagines not only communication architecture (interconnection networks), but storage systems and their overall systems. This improves the performance and cost-efficiency of emerging applications. The team's research covered SSD controller architectures/organizations, main memory system organization with processing-in-memory, and node-to-node data movement.
OlympusMons Pioneer Award
Research Bonus: 200,000 RMB-
- Achievements：
- Large-Scale, Dynamic Key-Value Store Solution
-
- Major team members：
- Peking University
-
-
Yang Tong
Wu Yuhan
Liu Zirui
-
- Achievement introduction：
- This achievement proposed the Large-Scale, Dynamic Key-Value Store Solution, a major innovation in dynamic data management. This team designed three key technologies: Color Embeder, MapEmbed, and Mirror Asymmetric Hash Table. These technologies reduce invalid access in slow memory, cut storage costs, and improve system performance. In particular, the MapEmbed technology significantly improves the utilization of physical storage resources through mapping and embedding policies. In addition, the ChameleMon system is used to detect and locate packet loss on a large-scale network, measured in milliseconds. ChameleMon dynamically shifts measurement attention and flexibly allocates memory resources as network states change. This enables packet loss detection and measurement of accumulated packets. The system uses Fermat's little theorem, namely FermatSketch, to support data segmentation, accumulation, and subtraction. This way, it can adapt to network status changes and efficiently detects faults. These technologies provide innovative ideas for improving the performance of Huawei storage products.
OlympusMons Pioneer Award
Research Bonus: 200,000 RMB-
- Achievements：
- Key Technologies of Computational Storage for Unstructured Data Processing
-
- Major team members：
- Institute of Computing Technology of the Chinese Academy of Sciences
-
-
Liang Shengwen
Wang Ying
Liu Cheng
Li Huawei
Li Xiaowei
-
- Achievement introduction：
- This achievement innovatively uses cognitive memory, a computational storage for unstructured data processing. This provides technical solutions to the problems of unstructured data processing, task offloading, and data management. To solve the problem of low data processing efficiency caused by diverse unstructured data and processing algorithms, this achievement provides a dedicated processor design method for computational storage. The team developed a dedicated hardware architecture to accelerate data processing by abstracting common regular/irregular computing modes. To solve the problem of high task offloading complexity, the team developed an automatic processor generation framework for computational storage, improving the development efficiency of computational storage and providing scenario-based acceleration. Cognitive memory uses data semantics to govern unstructured data. It increases the amount of information per bit of data in storage and establishes a key media foundation for unstructured data processing applications, delivering ultimate per-bit cost efficiency.
Thanks to All the Teams Who Participated the OlympusMons Awards!
As we move towards the intelligent era, we must cross the data peak.
OlympusMons represents Huawei Storage's unremitting pursuit and exploration on the road to data peaks.
The integration of industry, education, and research opens the door for innovation in data infrastructure.
