ACACloud Product Introduction
ACACloud integrates globally leading computing power service capabilities, dedicated to providing customers with comprehensive intelligent computing services. It helps developers quickly build high-performance model APIs, enabling them to focus on product innovation without worrying about the high computing power costs associated with large-scale promotion.
Core Product Services
Cloud Computing Power Service
- Provides cloud services based on GPU computing power
- Including elastic GPU cloud servers and GPU bare-metal servers
- Fully supports multimodal AI intelligent computing businesses
- Instantly available for use, billed based on actual usage, effectively reducing costs
Model API Service
Curated and validated models to accelerate enterprise AI application development
Training-Inference Convergence Service
- An integrated service platform
- Meets online service needs such as AI training, inference, online programming, and evaluation
Product Features
Visual Interaction
- Graphical interface for a unified service experience
- Visual interaction for job management, supporting job creation, submission, and scheduling management via the web interface
- Visual management of cluster resources
Infrastructure Convergence
- Computing Power Convergence: CPU and GPU converged clusters, providing a heterogeneous computing power resource pool with CPU + GPU
- Network Convergence: IB network supports parallel computing for cross-node data access and transmission; VPC is used for general computing data transmission or management traffic transmission
Cluster Scheduling
- Heterogeneous computing power support
- Cross-cluster scheduling of computing power resources
- HPC job scheduling, supporting multiple schedulers such as Slurm and PBS Pro
- Support for scheduling GPU resources via virtualization and containerization methods
Fine-Grained Operations
- Multi-Tenant Operation and Billing
- Customizable Billing Units
- Multiple Billing Models: Subscription-by-Period and Pay-as-You-Go
- Visual Bill Management
Multi-Level Governance
- Support for channels and distribution partners
- Multiple subscription and delivery methods
- Multi-role permission management
Security and Compliance
- Tenant Data Isolation
- Large Model Platform Based on Privacy-Preserving Computation
- Compliance Assurance: Product Compliance, Service Transparency, Data Localization
High Efficiency & Performance
- High-performance computing power for premium services, supported by integrated tools
Intelligent Operations and Maintenance (O&M)
- Unified O&M Management of Infrastructure
- Real-Time Monitoring of Computing Power Clusters
- Support from Automated O&M Systems
Intelligent Computing Expansion
- Containerized management of AI training and inference GPU clusters
- GPU virtualization, supporting GPU resource virtualization and resource pooling to enable GPU oversubscription
- K8S containerized architecture for GPU computing power management, supporting elastic scaling to achieve flexible scheduling of GPU computing power