Advantages
CHESS — Unified HPC and AI Computing Management Platform has the following advantages:
Security and Reliability
PThe platform utilises a domestically developed supercomputing cluster management system with independent intellectual property rights. Its core technology is developed in-house, ensuring a high level of security and reliability.
Converged Scheduling
Flexible, bidirectional resource scaling between HPC scheduling and AI container scheduling enables intelligent self-adaptation for mixed HPC/AI workloads without manual resource pool configuration.
Comprehensive Functionality
First launched in 2001, the platform has matured through over 20 years of iterative enhancements. It offers a complete set of functional modules and intelligent visualisation design, and can be customised on demand for specific application scenarios.
User-Friendly
Featuring a flexible topology structure and a dual-role interface for "Administrator" and "User," the platform ensures clear separation of duties and responsibilities while accommodating personalised usage requirements.
Excellent Compatibility
Supports unified management of cross-vendor, heterogeneous computing resources. Compatible with a wide range of operating systems and software, and supports numerous AI frameworks and models.
Dynamic Scalability
Capable of integrating with public clouds, private clouds, supercomputing clusters, and intelligent computing clusters, enabling the effortless construction of hybrid cloud clusters for on-demand, ready-to-use deployment.
Main Functions
Holistic Operations Monitoring
Provides a "Panoramic Dashboard" for clusters, collecting comprehensive performance data across compute/storage/network dimensions in real time. Enables visual monitoring of cluster status, resource loads, job statistics, and anomaly alerts to ensure cluster availability and business continuity.
Unified Management
Centrally manages heterogeneous computing resources, users, data, permissions, and security policies from a holistic perspective, breaking down management silos. Facilitates granular governance by enabling detailed permission assignments tailored to complex organisational structures.
Intelligent Scheduling
Dynamically matches optimal compute nodes based on workload, application priority, and real-time resource status. Supports millisecond-level scheduling for hybrid loads such as AI training/inference, HPC scientific computing, and engineering simulation, enhancing overall cluster resource utilisation.
Precise Billing
Tracks resource consumption across multiple dimensions (project/department/user) and supports itemised metering for resources including CPU, GPU, and storage. Generates visual cost reports to provide users with clear insights into resource usage and expenditure details, aiding cost optimisation.
User Dashboard
View job status, historical job trends, job type distribution, and computing and storage quotas in one central dashboard, with quick access to frequently used HPC and AI applications. This unified overview helps users track resource usage and task progress, monitor HPC and AI workloads at a glance, and simplify day-to-day task management.
Scientific Computing (HPC)
Bring together notebook development environments, training jobs, inference services, and model and dataset management in one platform. Users can launch interactive development instances online to support the full AI workflow, from model development and tuning to inference. This integrated AI computing environment enables researchers to write code, experiment with models, and deploy services.
Artificial Intelligence (AI)
Pre-configured with optimal compilation parameters and runtime templates for multiple applications and images. Allows quick search and one-click invocation without complex configuration, reducing adaptation cycles, lowering technical barriers, and accelerating business deployment.
Foundation Model APIs
Access a unified model hub supporting a wide range of domestic and international foundation models, including publicly hosted and privately deployed models. The platform provides a consistent API interface, load balancing, usage-based billing, access control, and visual monitoring. By abstracting differences between underlying models, it centralises API usage, cost management, and permissions to support model-powered application development and API service operations.
Integrated Simulation and Design
Consolidates 3D modelling, simulation computation, and post-processing analysis within a single platform. Facilitates sharing of design and computational data for simplified management and usage.
Industry Applications

Fundamental Scientific Research
Fundamental Scientific Research Molecular Dynamics, Structural Mechanics, Fluid Dynamics, Thermodynamic Simulation, Biological Research, Astrophysical Research

Meteorology & Environmental Protection
Weather Forecasting, Climate Simulation, Pollution Prediction, Typhoon Warning, Oceanographic Research

Engineering Simulation
Aircraft / Automotive / Ship Design, Chip Design, Aerospace & Aviation, Weapons & Equipment, Nuclear Fusion / Fission Simulation

Life Sciences
Targeted Drug Research, Protein Reconstruction, Gene Sequencing, Genetic Development, Microbiological Research, Agriculture / Livestock & Poultry Breeding

Energy Development
Seismic Data Processing, Reservoir Simulation, New Energy Research, Wind Power Generation, Hydropower Generation

Intelligent Government Governance
Public Computing Centers, Intelligent Transportation, Emergency Management, Flood Control & Disaster Mitigation, Environmental Protection
AI Industrialisation
Large Model Training / Inference, Detection & Recognition, Voice Interaction, Autonomous Driving, Embodied Intelligence, Intelligent Agent Development
Industrial AI Transformation
AI for Science, Smart Manufacturing, Medical Imaging, Intelligent Customer Service, Smart Logistics, Smart Agriculture & Forestry
Finance & Securities
Quantitative Trading, Investment Simulation, Risk Control, High-Frequency Trading
Other Industries
Animation Rendering, New Materials Research, Remote Sensing & Mapping




