Network Security Internet Technology Development Database Servers Mobile Phone Android Software Apple Software Computer Software News IT Information

In addition to Weibo, there is also WeChat

Please pay attention

WeChat public account

Shulou

Baidu details the new generation of self-developed processor "Kunlun Core 2": support GDDR6 video memory, 128TOPS computing power

2025-02-27 Update From: SLTechnology News&Howtos shulou NAV: SLTechnology News&Howtos > IT Information >

Share

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

CTOnews.com November 29 news, Baidu Apollo and Kunlun Chip Company today held an Apollo Day technology open day to introduce a new generation of self-developed cloud general-purpose chip-Kunlun core second generation AI chip.

According to reports, the chip is based on 7nm technology, equipped with GDDR6 high-speed video memory, memory bandwidth up to 512GB / s, using a new generation of Kunlun core XPU-R architecture, significantly improved versatility and performance, can provide 256TOPS@INT8 and 128 TFLOPS@FP16 computing power.

Kunlun Core Technology CEO Ouyang Jian revealed that Baidu's self-developed AI chip Kunlun Core 2 has completed end-to-end adaptation of self-driving scenes.

In addition, the chip also supports mainstream deep learning development frameworks, such as Tensorflow, Pytorch, PaddlePaddle, etc. In the performance test of the typical perception model, the performance of Kunlun core is ~ 2 times higher than that of the mainstream solutions in the industry.

Features: high performance, strong programmability, strong portability and good stability.

Baidu Apollo also unveiled their roadmap, which they believe will be a key year for smart driving in 2025, when L2 + products will bridge the gap and the L4 business model will initially become available.

CTOnews.com learned that Baidu Apollo has also released its literary heart model "perception 2.0" architecture, which supports multimodal pre-fusion and fisheye camera blindness. Officials say the big model has become a core driver of autopilot.

According to reports, Apollo Lite is currently the only third in the world in the field of visual perception, and the only one in China that supports "pure vision" autopilot on urban roads.

The iterative self-training scheme, which makes full use of unlabeled data, can help small models to improve the training efficiency.

According to reports, the new model will be applied to the following areas:

-long-range visual 3D perception

-Multimodal end-to-end awareness

-Point cloud perception effect

-long-tail data mining based on picture-text weak supervision pre-training model

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.

Views: 0

*The comments in the above article only represent the author's personal views and do not represent the views and positions of this website. If you have more insights, please feel free to contribute and share.

Share To

IT Information

Wechat

© 2024 shulou.com SLNews company. All rights reserved.

12
Report