Adam WANG SHANGHAI MEGA
SENIOR MEMBER
- Aug 7, 2017
- 5,079
- -15
- Country
- Location
Defeated Amazon and Google: Alibaba Cloud created a new world record
2018-12-27 10:40:12
https://www.toutiao.com/a6639497224877244942/
On December 25th, Stanford University released the latest DAWNBench deep learning reasoning list. Alibaba Cloud won the image recognition performance and cost double champion, breaking Amazon's record of 8 months, which is the first time the list appeared. Chinese technology company.
Stanford University DAWNBench is one of the most authoritative competitions in the field of artificial intelligence. It requires the participating organizations' computing platforms to accurately identify and classify 50,000 images. The test results show that Alibaba Cloud recognizes pictures 2.36 times faster than Amazon and 5.28 times faster than Google.
Alibaba Cloud's artificial intelligence reasoning performance breaks records, surpassing AWS-based, Google hardware-based results
Behind Alibaba Cloud's record is the super computing power of the Alibaba Cloud computing platform. In this contest, the breakthrough point of Alibaba Cloud computing platform is to develop an acceleration framework for image recognition tasks, which can automatically compress the deep learning inference model from 32 bits to 8 bits, reducing the occupied computing resources by 4 times. It has also increased by 4 times. At present, the Alibaba Cloud computing platform can achieve 1000 trillion levels of deep learning performance per second.
Six years ago, Google brain identified the cat in the video for the first time through deep learning technology, which is considered to open a new era of artificial intelligence. But now, Alibaba Cloud ET brain can not only sort the massive pictures at the speed of Superman, but also quickly and finely identify fine-grained images of fine-grained images under the big picture categories. For example, the machine can not only recognize cats, but also quickly distinguish them. Different varieties of cats. Aliyun ET brain can now recognize 450,000 kinds of flowers and 10,000 kinds of birds in the world...
At present, the super computing power of Alibaba Cloud Computing Platform has been applied to large-scale applications in artificial intelligence scenarios such as image recognition, speech recognition, and content recommendation. For example, Pat Li Tao provides more than 20 million users every day in Taobao and Tmall. Figure search for goods.
Yu Feng, head of Alibaba Cloud's flexible computing, said, "With the popularity of computational scenarios such as artificial intelligence, computing chips will have a specialization in the industry, and a substantial increase in computing power will become the cornerstone of business innovation."
2018-12-27 10:40:12
https://www.toutiao.com/a6639497224877244942/
On December 25th, Stanford University released the latest DAWNBench deep learning reasoning list. Alibaba Cloud won the image recognition performance and cost double champion, breaking Amazon's record of 8 months, which is the first time the list appeared. Chinese technology company.
Stanford University DAWNBench is one of the most authoritative competitions in the field of artificial intelligence. It requires the participating organizations' computing platforms to accurately identify and classify 50,000 images. The test results show that Alibaba Cloud recognizes pictures 2.36 times faster than Amazon and 5.28 times faster than Google.
Alibaba Cloud's artificial intelligence reasoning performance breaks records, surpassing AWS-based, Google hardware-based results
Behind Alibaba Cloud's record is the super computing power of the Alibaba Cloud computing platform. In this contest, the breakthrough point of Alibaba Cloud computing platform is to develop an acceleration framework for image recognition tasks, which can automatically compress the deep learning inference model from 32 bits to 8 bits, reducing the occupied computing resources by 4 times. It has also increased by 4 times. At present, the Alibaba Cloud computing platform can achieve 1000 trillion levels of deep learning performance per second.
Six years ago, Google brain identified the cat in the video for the first time through deep learning technology, which is considered to open a new era of artificial intelligence. But now, Alibaba Cloud ET brain can not only sort the massive pictures at the speed of Superman, but also quickly and finely identify fine-grained images of fine-grained images under the big picture categories. For example, the machine can not only recognize cats, but also quickly distinguish them. Different varieties of cats. Aliyun ET brain can now recognize 450,000 kinds of flowers and 10,000 kinds of birds in the world...
At present, the super computing power of Alibaba Cloud Computing Platform has been applied to large-scale applications in artificial intelligence scenarios such as image recognition, speech recognition, and content recommendation. For example, Pat Li Tao provides more than 20 million users every day in Taobao and Tmall. Figure search for goods.
Yu Feng, head of Alibaba Cloud's flexible computing, said, "With the popularity of computational scenarios such as artificial intelligence, computing chips will have a specialization in the industry, and a substantial increase in computing power will become the cornerstone of business innovation."
