Facebook is working with Intel to develop the next line of Cooper Lake processors. In this sense, the giant of social networks will make its deep learning technology available to improve the performance of this hardware.
According to ITNews, Facebook will use its Float 16 technology, which consists of an FP format that enhances the performance of deep learning by providing a dynamic range that is directly proportional to the FP32 format. This format also enhances the performance of tasks that do not require high 32-bit accuracy, such as face recognition, speech recognition, automatic translation, etc.
In the case of Intel Copper Lake, the processor uses the 14 nanometer (nm) node and is deployed on HEDT chips and servers like its predecessor hardware, Cascade Lake. Based in China, Inspur, a hardware technology company, assisted Intel in developing this new 4 Xeon socket motherboard design that enables the development of 112-core stations in a 2U format.
Hardware manufacturers such as Lenovo, Dell, HP, ZT Systems and Supermicro are also focusing on developing motherboards for servers that use this type of architecture.
During CES 2019, Intel and Facebook announced that they were working with Facebook to develop an Artificial Intelligence (AI) chip that would be ready for implementation later this year. At the moment it’s not clear if this is Cooper Lake, but the truth is that these chips will be available in the next generation of Dell laptops.
In January last year Reuters revealed that the introduction of this technology had suffered several delays, but Navin Shenoy, head of Intel’s data center, said it would be available in December 2019.