Intel releases first Xeon with fpga in early 2016

Intel will make its Xeon chips with field-programmable gate array available in the first quarter of 2016, the manufacturer has announced. The processors are the first to come to specific partners, who can adapt their algorithms accordingly.

Diane Bryant, who leads Intel’s data center branch, told Computerworld at the Structure conference in San Francisco that the processors with fpga fit into the socket of the current Xeons, making it easier for companies to switch. According to Bryant, the Xeons will go to the largest providers of cloud services, among other things.

Those providers can accelerate computations such as those for machine learning and data analysis by deploying the FPGAs in their computing centers. The fpga can, among other things, address the memory on the chip, so that a hybrid chip must be faster than a separate Xeon and fpga. Intel has high hopes for FPGAs to complement its current chip offering. It acquired Altera last summer for $16.7 billion. Intel announced the Xeons with FPGAs in the summer of last year.

Intel expects that by 2020 one-third of the nodes at data centers will contain FPGAs. In addition to machine learning, or neural networks, companies can also use them to accelerate encryption and compression.