The XQRKU060 additionally brings excessive efficiency machine studying (ML) to house for the primary time. A various portfolio of ML growth instruments supporting business normal frameworks, together with TensorFlow and PyTorch, allow neural community inference acceleration for real-time on-board processing in house with a whole “process and analyze” resolution. The XQRKU060’s dense, power-efficient compute with scalable precision and huge on-chip reminiscence, gives 5.7 tera operations per second (TOPs) of peak INT8 efficiency optimized for deep studying, a virtually 25X improve in comparison with the prior technology.
Building on Xilinx’s house heritage and extremely profitable 65 nm space-grade units, the launch of the primary 20 nm half for house purposes advances the house business by three course of node generations. It delivers a major discount in measurement, weight and energy, and is engineered with sturdy radiation tolerant options. The XQRKU060 gives prospects with a space-resilient machine outfitted to deal with each brief and lengthy length missions in harsh house environments.
“With our extensive history in developing leading-edge, radiation tolerant technology and deploying this in reliable space-grade solutions, Xilinx continues its lead with the launch of the world’s most advanced process node for space,” mentioned Minal Sawant, house methods architect, Aerospace and Defense Vertical Marketing, at Xilinx. “The 20 nm RT Kintex UltraScale FPGA is breaking industry standards and setting a new benchmark for meeting the high compute requirements of high bandwidth payloads, space exploration and research missions.”
Reconfigurable Processing
The XQRKU060 is the business’s solely true limitless on-orbit reconfigurable resolution. The on-orbit reconfiguration capabilities, along with real-time on-board processing and ML acceleration, permits satellites to replace in real-time, ship video-on-demand, and carry out compute “on-the-fly” to course of complicated algorithms. The ML capabilities are appropriate to a wide range of issues spanning scientific evaluation, object detection, and picture classification – reminiscent of cloud detection – enabling improved processing effectivity and diminished choice latency each in house and on the bottom. As protocols and purposes progressively change, the adaptive compute structure of the XQRKU060 permits limitless on-orbit reconfiguration to allow prospects to carry out last-minute product updates previous to launch, in addition to after it has been deployed in orbit.
Performance and Resiliency for Space
The XQRKU060 presents wealthy DSP capabilities optimized for dense power-efficient compute. It is provided with 2,760 UltraScale DSP slices and gives as much as 1.6 TeraMACs of sign processing compute, greater than a 10X improve in comparison with the prior technology, in addition to dramatic effectivity good points for floating level computations. The elevated compute functionality in house is paired with huge I/O bandwidth from 32 high-speed transceivers (SerDes) that may run as much as 12.5 Gbps to ship 400 Gbps combination bandwidth.
“SEAKR Engineering, Inc. has collaborated with Xilinx for 15 years to consistently achieve challenging mission objectives for advanced space communications applications,” mentioned Paul Rutt, CTO at SEAKR. “We have baselined Xilinx’s 20 nm Kintex UltraScale FPGAs with 12.5 Gbps SerDes hyperlinks enabling high-throughput, versatile and reconfigurable modulation, demodulation, channelization and routing functionality in our…
