This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>

Enterprise
Worldwide
Huawei Global - English
Search
About OlympusMons Awards

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 More

OlympusMons Awards
2025 Challenges

OlympusMons Awarding Results

Winners of OlympusMons Pioneer Award Winners of OlympusMons Pioneer Award
Winner of OlympusMons Award Winner of OlympusMons Award

The 2026 Global Data Storage Expert Forum unveiled its shortlist for the 2025 OlympusMons Awards. The OlympusMons Award was presented to Prof. Shen Zhirong and Prof. Zhang Yiming and their team from the Key Laboratory of Intelligent Storage and Computing.

The OlympusMons Pioneer Award went to six teams: Mr. Chen Lei (Hong Kong University of Science and Technology (Guangzhou), Shanghai Jiao Tong University), Mr. Xu Erci (Shanghai Jiao Tong University), Mr. Chen Zhineng (Fudan University), Mr. Liu Fangxin (Shanghai Jiao Tong University), Mr. Hu Yuchong (Huazhong University of Science and Technology), and Mr. Frank van Harmelen (Vrije Universiteit Amsterdam).

The 2025 OlympusMons Awards received submissions from over 160 experts and scholars from 30 universities and research institutions in five countries and regions. The winning teams were determined by the OlympusMons Review Committee, who evaluated all entries based on the technical and commercial value of their scientific research achievements.

OlympusMons Award

OlympusMons Award

Research Bonus: 1,000,000 RMB
  • Achievements：
    Research on I/O Optimization and Efficient Organization Techniques for Hierarchical Memory Systems
  • Major team members：
    Key Laboratory of Intelligent Storage and Computing
  •  
    Shen Zhirong
    Shen Zhirong
    Zhang Yiming
    Zhang Yiming
    Gao Congming
    Gao Congming
    Xiang Qiao
    Xiang Qiao
    Tang Lu
    Tang Lu
  • Achievement introduction：
    This research addresses the memory requirements of agentic AI hybrid workloads, including high bandwidth, low latency, large capacity, and high reliability. It builds hierarchical memory systems based on heterogeneous storage media such as new high-bandwidth storage device, DRAM, and SSD, and offers an efficient I/O access paradigm powered by next-generation xPUs. By developing GeminiFS, the first companion file system for GPUs, and innovative techniques such as the I/O stack of direct GPU access to storage, network-storage-compute convergence, and cross-tier collaborative indexing, this research provides a solution to the problem of low inference efficiency in traditional file systems, and enables high-performance data access and inference acceleration for xPUs. These research insights have been applied in areas such as China's supercomputing and AI computing center, computing card clusters, and data centers of top cloud and Internet vendors. In the past three years, more than 20 papers on this subject have been published in top-tier conferences.
OlympusMons Pioneer Award

OlympusMons Pioneer Award

Research Bonus: 200,000 RMB
  • Achievements：
    Research on Key Techniques for Quantitative Investment Based on Knowledge Graphs and AI Agents
  • Major team members：
    Hong Kong University of Science and Technology (Guangzhou), Shanghai Jiao Tong University
  •  
    Chen Lei
    Chen Lei
    Shen Yanyan
    Shen Yanyan
    Zhu Xinyi
    Zhu Xinyi
    Wang Liping
    Wang Liping
    Wang Hao
    Wang Hao
    Li Xujia
    Li Xujia
    Yu Haolin
    Yu Haolin
  • Achievement introduction：
    This research provides technical insights into the formation of high-quality knowledge foundations for AI agents. It offers a complete technology system covering dynamic and trustworthy knowledge extraction, multimodal knowledge representation, high-reliability retrieval and inference, and agent application. For the challenges of heterogeneous data convergence, delayed knowledge updates, and lack of trustworthy bases for model decision-making, the research team proposed the innovative use of dynamic knowledge graph construction, dynamic graph knowledge representation, multimodal joint inference, and explainable quantitative modeling. This can significantly improve knowledge extraction accuracy, knowledge recall rate after updates, and knowledge-based prediction accuracy. The results of this research have been published in top-tier international conferences such as VLDB, KDD, AAAI, ICLR, and ICDE, and numerous invention patents have been applied for. In addition, the team has developed LingCopilot, a multi-agents no-code quantitative trading system, which won a Gold Medal with Congratulations of the Jury at the 51st International Exhibition of Inventions Geneva in 2026. The research findings have been applied and validated in financial enterprise applications. This has enabled the development of trustworthy knowledge infrastructure for LLM agent applications.
OlympusMons Pioneer Award

OlympusMons Pioneer Award

Research Bonus: 200,000 RMB
  • Achievements：
    Full-Stack I/O Collaborative Optimization for Agentic AI Workloads
  • Major team members：
    Shanghai Jiao Tong University
  •  
    Xu Erci
    Xu Erci
    Xue Guangtao
    Xue Guangtao
    Shen Yanyan
    Shen Yanyan
    Wu Haonan
    Wu Haonan
    Yang Leping
    Yang Leping
  • Achievement introduction：
    This research innovatively proposed the use of a full-stack I/O collaborative optimization system that covers compute, memory, and storage, to address challenges such as I/O bursts, access conflicts, and long-tail latency caused by agentic loads spanning different stages. The system was developed based on a "3 horizontal and 1 vertical" technical roadmap. The horizontal part covers orchestration and governance of high-impact traffic, transformation of hardware-software converged data paths, and collaborative pooling of multimedia resources. The vertical part covers a closed-loop reliability process spanning from observation, to diagnosis and feedback. Key technologies, such as hybrid scheduling, high-performance protocol stack, elastic memory pooling, and cloud-edge fault monitoring enable a comprehensive upgrade in system performance and reliability. The research team has published more than 20 papers in top-tier conferences, and has received the Best Paper Award on three separate occasions at FAST. This technical solution has been validated at scale on top cloud platforms. It has been successfully applied in industries such as finance and telecommunications, generating economic benefits and exerting significant industry influence.
OlympusMons Pioneer Award

OlympusMons Pioneer Award

Research Bonus: 200,000 RMB
  • Achievements：
    Ultra-Lightweight Document Image Parsing and Knowledge Extraction
  • Major team members：
    Fudan University
  •  
    Chen Zhineng
    Chen Zhineng
    Jiang Yugang
    Jiang Yugang
    Du Yongkun
    Du Yongkun
    Wu Zuxuan
    Wu Zuxuan
    Su Yuchen
    Su Yuchen
  • Achievement introduction：
    This research examines accurate, efficient, and lightweight technologies for text recognition, document parsing, and knowledge extraction. By developing a text recognition model based on a unified vision model, the research team achieved industry-leading recognition accuracy and inference speeds. The research results have been applied at national government departments and renowned enterprises. It has been integrated as a core module in the world's leading open-source text recognition system.
    The team also developed OpenDoc-0.1B, an ultra-lightweight document parsing system with only 0.1 billion parameters. It uses unified parsing of text, formulas, and tables to significantly reduce the compute resource overhead and device-side deployment costs of document parsing. Furthermore, the team has continued to conduct research on knowledge extraction using lightweight large models for open scenarios, improving knowledge extraction accuracy across diverse applications. The research results have been published in leading journals and conferences, including IEEE TPAMI, IJCV, CVPR, and ICCV, and have received top honors, including the first prize in the Algorithm Challenge for the 2024 Global AI Innovation Contest (GAIIC).  
OlympusMons Pioneer Award

OlympusMons Pioneer Award

Research Bonus: 200,000 RMB
  • Achievements：
    Enhancing Compute with Querying: Efficient Inference System of Long-Sequence Large Models Driven by Semantic Information Condensation and Query Reuse
  • Major team members：
    Shanghai Jiao Tong University
  •  
    Liu Fangxin
    Liu Fangxin
    Jiang Li
    Jiang Li
    Guan Haibing
    Guan Haibing
  • Achievement introduction：
    This research examines key bottlenecks in long-sequence inference, such as KV cache expansion and limited memory access bandwidth. It proposes an innovative "enhancing compute with querying" approach that is based on information entropy theory. This approach uses entropy-based error analysis to identify semantic information that can be compressed and reused, transforming repeated computing into low-latency queries and enabling end-to-end performance gains through storage-compute collaboration.
    This approach has four major features. First, cross-dimensional semantic clustering based on product quantization enables semantic condensation, reducing KV cache memory consumption. Second, Markov decision modeling enables sensitivity-aware dynamic layer skipping to reduce repeated computing. Third, entropy-aware prefetching and result reuse transform frequently repeated computations into cache queries to reduce data transfer overhead. Finally, semantic indexing operatorization and asynchronous offloading improve inference throughput.
    These research results have been released with MindSpore 2.8 and published at leading conferences such as ASPLOS, ISCA, HPCA, DAC, and ACM MM. The team's paper received the Outstanding Papers in Systems Theme Award at the 2025 ACM MM Conference, the Special Feature Award at the ASP-DAC 2026 University LSI Design Contest, and the Huawei Spark Award.
OlympusMons Pioneer Award

OlympusMons Pioneer Award

Research Bonus: 200,000 RMB
  • Achievements：
    Lightweight Indexing and Offload-Friendly Operator Techniques for High-Performance, Large-Capacity SSDs for Large Models
  • Major team members：
    Huazhong University of Science and Technology
  •  
    Hu Yuchong
    Hu Yuchong
    Li Xiaolu
    Li Xiaolu
    Tong Wei
    Tong Wei
    Fang Peng
    Fang Peng
    Liu Kang
    Liu Kang
  • Achievement introduction：
    This research addresses key bottlenecks faced by large-capacity SSDs in large model training and inference, including index metadata bloat, long data access paths, operator parallelism limitations, and inefficient near-data processing. It proposes the use of key technologies such as label-free hash indexing, semantic-aware pass-through indexing, bit-decoupled operators, and table-based reuse operators. These technologies reduce DRAM overhead for indexing, shorten data access paths, improve parallelism in near-storage computing, and accelerate computing through operator offloading to SSDs. Together, they form an SSD-based storage-compute convergence technology system centered on lightweight indexing and offload-friendly operators.
    These research results have been validated on enterprise-class large-capacity SSDs and the Ascend platform. The research has resulted in more than 50 CCF-A-class papers and 35 invention patents. This has enabled key technical support for the development of large-capacity SSDs, computational storage, and AI infrastructure.   
OlympusMons Pioneer Award

OlympusMons Pioneer Award

Research Bonus: 200,000 RMB
  • Achievements：
    Crossing the Probabilistic-Logic Barrier
  • Major team members：
    Vrije Universiteit Amsterdam
  •  
    Frank van Harmelen
    Frank van Harmelen
    Michael Cochez
    Michael Cochez
    Peter Bloem
    Peter Bloem
    Daniel Daza
    Daniel Daza
    Emile van Krieken
    Emile van Krieken
    Taraneh Younesian
    Taraneh Younesian
  • Achievement introduction：
    Addressing the problems of AI hallucinations and non-explainability, this study innovatively proposes a neuro-symbolic AI framework that combines probability with logic, achieving knowledge enhancement and trustworthy reasoning. To improve the scalability of Graph Neural Networks, advanced sampling techniques are developed to significantly reduce the complexity of expansion while maintaining SOTA accuracy. Further a novel neural link prediction model was introduced into knowledge graphs to reduce the noise interference, surpassing SOTA performance by 22% while requiring two orders of magnitude less training data. Additionally, the research theoretically reveals the zero-gradient defect in fuzzy logic and introduces a unified architectural notation to enhance the reliability of reasoning. By breaking the probability-logic barrier, these innovations enable adaptive scaling and reliability enhancement, supporting the construction of low-cost, high-trust AI data infrastructure.   
Thanks to All the Teams Who Participated the OlympusMons Awards!

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.

HUAWEI eFly App

Copyright © 2026 Huawei Technologies Co., Ltd. All rights reserved.

Ask the Chatbot

Contact Huawei Sales

More Contact Methods

Ask the Chatbot

Contact Huawei Sales

More Contact Methods

TOP