Shulou(Shulou.com)11/24 Report--

In recent years, the availability and advantages of GPU database are becoming more and more obvious because of its strong parallel computing ability, large memory broadband and capacity. Domestic Internet giants and traditional database enterprises have laid out the GPU database market one after another.

In 2013, the R & D team of Shanghai AGB Information Technology Co., Ltd. (hereinafter referred to as "AGB") began the research and development of high-performance distributed database based on GPU, and actively realized the localization of the database. In 2020, AGB and Haiguang Information began to cooperate to explore the combination of heterogeneous acceleration cards with independent intellectual property rights and database products. Thanks to Haiguang's complete tool chain and friendly cross-platform migration capability, with the cooperation of AGB and Haiguang engineers, the migration from foreign platform to domestic Haiguang platform was completed in only two months. After continuous optimization and improvement, the performance of the migrated database is better than that of foreign products of the same level, and the high-performance distributed database based on GPU is gradually taking shape.

Zheng Xueqiang, chairman of AGB and CEO, said that "GPU-based high-performance distributed database" (hereinafter referred to as DataTurbines) has three advantages: first, it is highly innovative, with the rapid development of AI technology, the innovative value of GPU database will be highlighted; second, it is based on the homemade Hai Guang's heterogeneous acceleration chip, which has a strong advantage in system compatibility. The third is to encourage innovation from the national policy level, and the scheme of "domestic high-performance distributed database based on GPU" has no precedent in China.

At present, AGB's DataTurbines database products have achieved good results in both financial and industrial fields. In the financial field, DataTurbines is used in the core business fields such as online analysis system, transaction risk control system and big data data warehouse platform of the financial industry, replacing foreign all-in-one products, not only saving expensive system authorization and maintenance fees for enterprises, but also introducing new hardware and new functions. In the field of industrial security, security simulation algorithms in complex industrial fields have been monopolized by several established foreign algorithm companies. Through the cooperation between AGB and university laboratories, through the self-development of industrial security algorithms, combined with the distributed architecture of DataTurbines database and the acceleration ability of GPU hardware, the running efficiency and accuracy of the algorithms have been greatly improved, which not only fills the gap of complex industrial security software, but also creates a more secure and efficient digital base.

In 2022, AGB further deepened its cooperation with Haiguang and improved the function of its products by participating in the second AI solution Competition held by the Organization of photosynthesis, and officially launched the intelligent data all-in-one machine based on Haiguang and AGB DataTurbines (hereinafter referred to as "all-in-one"). The all-in-one machine focuses on achieving better product optimization through the combination of software and hardware, making the user experience smoother and more secure.

In view of the future development of the product, Zheng Xueqiang said that he will focus on two aspects: one is to carry out more in-depth cooperation with Haiguang in trusted computing and data security, and introduce Haiguang's security platform features into the DataTurbines project to create a new generation of high-performance and multi-functional financial data integrated machine platform. The second is to adapt to domestic independent and advanced landing applications, such as AI large model, AI training or landing applications based on AIGC, to further optimize the product.

At present, with the development of the AI model, AGB has developed a multimodal database version that integrates support vector data and structured data. "after the arrival of the AI era, this will be an important basic platform for generative AI applications." Zheng Xueqiang said.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.