HPC High-Speed Computing Service
High Performance Computing
Equipped with NVIDIA® Tesla V100 GPUs, the largest, most complete, and classic high-speed computing GPU cluster service in China.
Are you not a management expert in HPC clusters? Is your organization lacking professional IT management personnel?
Don't worry, just provide the computing resources and configurations you need based on your computing requirements, and HPC high-speed computing service can complete the required computing tasks in batches, allowing you to focus on parallel computing applications and accelerate AI training, inference, and high-performance computing.
HPC High-Speed Computing Service
Taiwan AI Cloud's high-speed computing service provides a large number of GPUs paired with a 100Gbps high-performance, low-latency internal interconnect network to help you:
- It meets the needs of large-scale parallel computing.
- Accelerate computational iteration efficiency: For large amounts of computational work, HPC cluster environments can be utilized.
Advantages
High-speed interconnection of a large number of GPUs solves the bottleneck of large-scale data transfer between GPUs, enabling large-scale GPU parallel computing.
Cross-node scheduling of multiple GPUs enables a distributed, high-speed parallel computing environment with performance exceeding 4001 TP3T.
High-bandwidth network connects nodes, enabling high-efficiency transmission of massive amounts of data.
GPU Direct and RDMA architectures deliver extreme acceleration in overall computing efficiency.
Function
- Supports native HPC operating environments.
- Slurm is used to schedule GPUs across nodes, enabling native high-speed parallel computing.
- It operates on physical machines without virtualization, achieving full performance.
- Supports Singularity container environments.
- It allows you to deploy computational tasks in large batches.
- Fully integrated with HFS, a fully managed, high-performance parallel archive system.
- A graphical interface is provided for easy submission of calculations.