OVC company's server adapts to DeepSeek series large models

Wuhan Yangtze Computing Technology Co in the Wuhan East Lake High-tech Development Zone (Optics Valley of China, or OVC) has comprehensively adapted and optimized the DeepSeek series large models on their independently developed G440K V2 servers.

Established in 2020, Yangtze Computing is a domestic computing power service provider. "In sectors such as finance and telecommunications, there is a pressing demand for domestic artificial intelligence (AI) large models," stated Yuan Zhentao, vice-general manager. "Therefore, following the release of DeepSeek-V3, we promptly organized our technical team to analyze the required hardware resources for inference and swiftly completed the adaptation of large models."

Yuan also revealed that deploying the DeepSeek series models on the G440K V2 servers enhances the model's inference speed, response efficiency, and data processing capabilities while enabling multicomputer networked inference.

In collaboration with Ascend and Huawei Cloud teams, Yangtze Computing has deployed the localized inference service of the DeepSeek-V3 model on the G440K V2 server for a large State-owned bank. This enables the bank to swiftly process massive financial data within a secure internal environment.

Yangtze Computing has completed adaptation work for over 100 large models, covering most domestic and international open-source models. "In the future, the company will stay at the forefront of the market, creating a more flexible and reliable AI application environment for customers and driving the deep application of domestic large models across various industries," said Yuan.

Under local computing power and corresponding technical support, the Wuhan Artificial Intelligence Computing Center has also deployed the DeepSeek and Janus Pro series models, accelerating local AI technology's popularization and practical application.