Xilinx, Inc. (Nasdaq: XLNX), a global leader in All Programmable technologies and devices, announced details of the new 16nm Virtex® UltraScale + ™ FPGA with HBM and CCIX technology. . The HBM-enabled FPGA family boasts the highest memory bandwidth, 20x faster memory bandwidth than DDR4 DIMMs, and a 4x reduction in power per bit compared to competing memory technologies. Built to meet the increased memory bandwidth required for compute-intensive applications such as machine learning, Ethernet connectivity, 8K video and radar, these new devices also provide CCIX IP, cache coherency acceleration for any CCIX processor, To meet the requirements of accelerated computing applications.
"Packaging integrated DRAM represents a significant step forward in the development of memory bandwidth for high-end FPGA applications," said Kirk Saban, senior director of FPGA and SoC product management at Xilinx. "HBM is integrated into Xilinx's industry-leading devices, "The clear direction towards multi-terabyte memory bandwidth in the future and our accelerated hardening technology will enable highly efficient heterogeneous computing to meet our customers' demanding workloads and application requirements."
HBM Optimized Virtex UltraScale + products are based on the proven 16nm Virtex UltraScale + FPGA family, which has been sampling in 2015, providing the lowest risk path for HBM integration. Built using the third-generation CoWoS technology jointly developed by TSMC and Xilinx, the series is now the industry benchmark for HBM integration.