According to foreign media reports, Google is no longer satisfied with developing artificial intelligence (AI) chips for its own data centers. It is now designing to integrate AI chips into products produced by other companies.
Two years after the release of the Tensor Processing Unit (TPU), Google launched the Edge TPU on Wednesday, local time in the United States, which will enable sensors and other devices to process data faster.
This chip can be used in a variety of scenarios, but one initial use is in industrial manufacturing: consumer electronics manufacturer LG is testing this chip in a system that can detect the presence of glass for the screen Manufacturing defects.
Google's entry into the "custom chip" market is a way for it to try to expand its cloud computing market share and strengthen its competition with Amazon and Microsoft. Since 2015, Google has been using TPU to accelerate certain workloads in its own data centers, rather than relying on commercial hardware provided by vendors such as Nvidia.
In 2017, Google stated that its AI chips are becoming more strategically important. In the field of AI, researchers are training models with large amounts of data so that machines can predict when new data arrives.
The initial version of TPU can only make these predictions, while the second version (released in 2017) can be used to train the model. This update enables it to compete with Nvidia graphics cards. The third-generation TPU was released earlier this year.
Now we have Edge TPU, which is a microchip specially used to process the AI ​​prediction part, which is less computationally intensive than the training model. Edge TPU can run calculations on its own, without connecting to multiple powerful computers, so applications can work faster and more reliably. They can handle AI work with standard chips or microcontrollers in sensors or gateway devices.
Samsung's former chief technology officer Injong Rhee said that Google did not let Edge TPU compete with traditional chips, which is very beneficial to all silicon chip suppliers and equipment manufacturers. Edge TPU may "subvert the cloud computing competition", because many calculations can now be performed on the device, rather than all sent to the data center. In terms of cost and energy consumption, Google chips are more efficient than traditional chips in certain types of calculations.
Google is not the only cloud computing service provider interested in the so-called Internet of Things. The core of the Internet of Things is to manage and process data from many small embedded devices. Earlier this year, Microsoft announced the design of its IoT chip. Google's new chip will run a model based on a simplified version of TensorFlow AI software, which was released by the company under an open source license in 2015.
LG's CNS team, which is responsible for helping internal and other companies handle IT services, is already testing Edge TPUs and plans to start using them to inspect equipment on internal production lines.
Currently, in the process of producing glass for display panels, the inspection equipment can process more than 200 glass images per second. Hyun Shingyoon, chief technology officer of LG's CNS team, said that any problems that arise need to be manually checked, and the accuracy of the existing system is about 50%. The accuracy rate of Google AI can reach 99.9%.
Hyun Shingyoon also said: "My expectation is to save money in finding anomalies and defects that really affect our quality." His team had previously studied a computing system from Nvidia.
Google has built a toolkit that includes Edge TPU, NXP chip and Wi-Fi connection for developers to try. The company is working with manufacturers such as Arm, Harting, Hitachi Vantara, Nexcom, Nokia and NXP.
Injong Rhee did not disclose whether Google plans to build a more powerful Edge TPU for training models.
Heating Eye Mask,Self-warming Eye Mask,Portable Heating Eye Mask,Rechargeable Heating Eye Mask
Ningbo Sinco Industrial & Trading Co., Ltd. , https://www.newsinco.com