It has been an exciting couple of weeks for Xilinx, with the Versal and Alveo announcements and all of the activities at XDF Silicon Valley and XDF Beijing. Adding to this momentum, at last week’s Samsung Tech Day, Samsung Electronics announced SmartSSD, a Xilinx-based (FPGA) SSD that enables high-performance accelerated computing closer to the data, bypassing CPU limitations.
Helping to enable future innovations in the next-generation of data centers like high-performance computing, artificial intelligence and emerging applications, Samsung’s SmartSSD powered by Xilinx FPGAs will increase speed and efficiency, and lower operating costs by pushing intelligence to where data lives.
For example, you are looking for an image of a cat. Normally the system would have to transfer the data through the server CPU and then to the storage device and back to the CPU and back to the user. It is significantly more performance and power efficient to have the storage device itself search directly for “cat.” That’s the beauty of SmartSSDs.
Current Xilinx FPGA-based acceleration solutions are targeted towards compute applications like speech recognition, video transcoding and genomics, and are deployed close to the server CPU. In contrast, SmartSSD is focused on accelerating storage services such as compression, de-duplication and encryption, as well as compute functions, like machine learning inference and big data analytics, that can benefit significantly from being accelerated closer to the actual storage component. And similar to Xilinx accelerators that are deployed close to the server CPU, SmartSSD offers a platform approach to an application developer ecosystem where the familiar SDAccel environment can be leveraged to develop and deploy acceleration solutions.
This technology underscores the close technology collaboration between Xilinx and Samsung, and highlights the adaptability and flexibility that Xilinx brings to the data center.