Guangzhou GRG Wuzhou Technology Co., Ltd.

Background ​

• Complex production line equipment and technologies
  1. There are many core control points and the process is complex.
  2. It is difficult to identify core control factors and control requirements (from materials to processes to equipment).
• Difficulty in issue identification
  1. The R&D test period is long and the cost is high.
  2. It is difficult to conduct innovative scientific experiments.

Responding to the preceding pain points, an intelligent production line system is built based on openFuyao for the advanced manufacturing production line industry represented by film production.

Solution ​

Using openFuyao, the most powerful computing power scheduling base made in China, an intelligent production line system for composite current collectors is built to integrate computing power, algorithms, and data.

• Aggregated computing power: centralized management and control, remote and heterogeneous resource pooling.
• Intelligent O&M: one-stop O&M and usage monitoring of various resources, integrated DevOps.
• Computing power scheduling: software-defined compute (SDC), software-defined networking (SDN), and software-defined storage (SDS); unified and intelligent scheduling and management of computing power
• Out-of-the-box AI tools: AI development tools, AI large models and algorithms, and out-of-the-box AI tools.

Advantages ​

With the convergence and management capabilities of the container platform, the solution implements automatic AI computing power scheduling and scenario-based AI tool algorithms, greatly improving the convenience of building an intelligent production line system for composite current collectors and improving the development efficiency.

Efficient scheduling and one-stop visualized management and control:

• Scheduling of general-purpose CPU computing resources, such as x86, Arm, and C86.
• Adaptation to GPU and NPU computing power scheduling of multiple mainstream vendors in and outside China, such as NVIDIA, Huawei Ascend, and Iluvatar CoreX.
• One card for multiple purposes, diversified computing power, and GPU partitioning, meeting the requirements of integrated training and inference.
• One-stop heterogeneous resource and visualized monitoring alarm.

Diversified development environments and open ecosystem:

• Production-level cloud-native container openFuyao base.
• Built-in foundation models and AI development environments.

Benefits ​

• Based on the integrated training and inference server of the Chinese architecture and the openFuyao computing power base, the solution supports production line large models and various small models, greatly improving the computing efficiency.
• It perfectly adapts to various mainstream hardware, abstracts away the diversity in computing power, and accelerates and schedules underlying computing power, allowing service developers to focus more on service development and improve development efficiency.