当前位置:网站首页>Performance improvement 2-3 times! The second generation Kunlun core server of Baidu AI Cloud was launched
Performance improvement 2-3 times! The second generation Kunlun core server of Baidu AI Cloud was launched
2022-07-01 08:45:00 【Baidu AI Cloud】
In recent days, , Baidu AI Cloud and Kunlun core technology officially launched the second generation Kunlun core cloud server NKL5 series , It can provide diversified services for users 、 High performance 、 Domestic self-developed cloud with flexibility and easy operation and maintenance AI Computing services .
Canalys According to the report , With the accelerated deployment of China's new infrastructure planning , The demand for digital transformation of enterprises is becoming more and more urgent , The market has put forward new requirements for cloud service providers . In addition to the need for the underlying cloud technology , Under the trend of digital transformation , Enterprises value more AI The service capability provided by technology and the business innovation brought to the industry are accelerated .
The new generation of Kunlun core cloud server will empower more enterprises and government agencies , Meet the needs of different computing capabilities , Accelerate the landing pace of users' artificial intelligence , Realize intelligent upgrade .
The second generation Kunlun core cloud server has the following core advantages :
>> Equipped with a new generation architecture XPU-R, Computing power is significantly enhanced
Improve the computing power of the computing core through self-developed architecture 2-3 times , It greatly enhances the computing power of the product . Flexible support AI The evolution of the algorithm , Improve the effectiveness of resource input . For typical AI load , The measured throughput performance of the second generation Kunlun core is the mainstream in the industry GPU Of 1.5 times .
>> Hardware level virtualization , Marked improvement AI Utilization rate of computing resources
Support the virtualization function of Kunlun core acceleration card , Its computing unit and storage unit can be physically isolated 、 For multiple users . In ensuring the quality of service (Quality of Service,QoS) Under the circumstances , Marked improvement AI Accelerate the resource utilization of the card .
>> Flexible and easy to operate , Keep up with business development requirements
One or more high-performance computing instances can be easily obtained in a few minutes , Flexible selection of appropriate instance specifications on demand , Fast 、 Smooth expansion , Meet the needs of rapid business development .
>> Out of the box development environment , Support rapid deployment of applications
Support the dedicated image of Kunlun core cloud server , contain AI The necessary development environment for the accelerator card .
Kunlun core is the only cloud in China that has experienced the hardening of Internet large-scale core algorithms AI chip , In Baidu search engine 、 Xiaodu and other businesses deployed Kunlun core more than 2 Ten thousand pieces . In Baidu search business , Compared with the original scheme, Kunlun core realizes 2 More than times the performance , The overall TCO Decrease by more than 100 million yuan .
Outside Baidu , In the past year, Kunlun core technology enabled... In different scenes through computing power AI application , In Bank of Jiangsu 、 Jiangsu Yancheng Intelligent Computing Center 、 Chongqing Higher People's court 、 Provincial procuratorate 、 Hubei Yichang Point Military Region Computing Center 、 Zhiyuan Artificial Intelligence Research Institute and other projects have been implemented on a large scale , Help the government and enterprises speed up the intelligent layout of industries .
Baidu AI Cloud continues to polish its core technology , Provide users with the whole scene 、 End to end AI Computing services , Help the government 、 The development of artificial intelligence in enterprises .
Click on Read the original Learn more about Kunlun core cloud server .
边栏推荐
- Software Engineer Interview Question brushing website and experience method
- Shell脚本-case in 和正则表达式
- Matlab [function derivation]
- 19Mn6 German standard pressure vessel steel plate 19Mn6 Wugang fixed binding 19Mn6 chemical composition
- Insert mathematical formula in MD document and mathematical formula in typora
- Redis源码学习(29),压缩列表学习,ziplist.c(二)
- 中断与其他函数共享变量、临界资源的保护
- [JS reverse] MD5 encryption parameter cracking
- 1. Connection between Jetson and camera
- factory type_id::create过程解析
猜你喜欢
What is 1cr0.5mo (H) material? 1cr0.5mo (H) tensile yield strength
AVL树的理解和实现
个人装修笔记
[MFC development (16)] tree control
Bimianhongfu queren()
MD文档中插入数学公式,Typora中插入数学公式
【js逆向】md5加密参数破解
《微机原理》——微处理器内部及外部结构
Nacos - service discovery
Only in China! Alicloud container service enters the Forrester leader quadrant
随机推荐
Shell脚本-数组定义以及获取数组元素
[untitled]
Brief introduction to AES
VSYNC+三重缓存机制+Choreographer
2022.2.15
The era of low threshold programmers is gone forever behind the sharp increase in the number of school recruitment for Internet companies
内存大小端
性能提升2-3倍!百度智能云第二代昆仑芯服务器上线
Shell脚本-位置参数(命令行参数)
如何一站式高效管理固定资产?
View drawing process analysis
截图小妙招
MATLAB【函数求导】
Computer tips
It is designed with high bandwidth, which is almost processed into an open circuit?
3. Detailed explanation of Modbus communication protocol
电视机尺寸与观看距离
AES简单介绍
V79.01 Hongmeng kernel source code analysis (user mode locking) | how to use the fast lock futex (Part 1) | hundreds of blogs analyze the openharmony source code
【MFC开发(17)】高级列表控件List Control