Baidu and Samsung announced on Tuesday that Baidu Kunlun, Baidu’s first cloud-to-edge artificial intelligence acceleration processor, has completed development. Moreover, it will begin mass production early next year.
The Baidu Kunlun chip is based on Baidu’s independently developed neural processor architecture XPU for the cloud, edge, and artificial intelligence. The chip will use Samsung’s 14-nanometer manufacturing process and I-Cube TM packaging solution.
Also, this chip provides 512GBps of memory bandwidth and achieves 260 TOPS processing power at 150 watts of power. In addition, this new chip supports the pre-trained model Ernie for natural language processing. The inference speed is 3 times faster than traditional GPU/FPGA accelerated models.
With the computing power and energy efficiency of this chip, Baidu can support multiple functions, including large-scale artificial intelligence computing. The latter mainly refers to the deep learning platforms such as search ordering, speech recognition, image processing, natural language processing, autonomous driving, etc.
What’s more interesting, this is also the first chip foundry cooperation between Baidu and Samsung. Baidu will provide an artificial intelligence platform that maximizes the performance of artificial intelligence. At the same time, Samsung will expand its chip manufacturing business to high-performance computing (HPC) chips dedicated to cloud computing and edge computing.
‘We are excited to lead the HPC industry together with Samsung Foundry,’ said OuYang Jian, Distinguished Architect of Baidu. ‘Baidu KUNLUN is a very challenging project since it requires not only a high level of reliability and performance at the same time, but is also a compilation of the most advanced technologies in the semiconductor industry. Thanks to Samsung’s state of the art process technologies and competent foundry services, we were able to meet and surpass our goal to offer superior AI user experience.’