Full-Stack Computing Product System
Based on deep platform management experience, providing a complete product line from underlying scheduling, platform management to upper-level LLM applications.
SCOW Computing Platform
Integrated HPC/AI/Quantum/Cloud computing portal. Lower usage barriers, support VM management in HPC environments, enable flexible Windows/Linux operation without admin privileges for GPU resources.
XSCOW Network Integration
Breaking computing silos. Connect HPC, AI, and cloud computing centers to build a unified computing network marketplace with unified authentication, scheduling, and billing.
CraneSched Scheduler
High-performance computing scheduler. Compatible with Slurm/LSF, supports domestic chips (Kunpeng/Ascend/Hygon). Proprietary algorithms improve resource utilization.
ShadowDesk Remote Desktop
High-performance remote visualization. Supports 4K resolution, 144FPS for EDA chip design and CAE simulation. Browser-based access with data security.
Garlic LLM Agent
Privately deployed local LLM system. Provides RAG knowledge base Q&A and agent applications, ensuring data stays within domain with fine-grained permission control.
MaaS LLM Service
One-stop access to all models. Aggregates mainstream model APIs including DeepSeek, OpenAI, Qwen. Provides unified authentication, billing, and multi-model scheduling.
