Editor’s Note: This content is contributed by Mike Thompson, Sr. Product Line Manager for High-End ACAPs and FPGAs
VU57P Production FPGAs are Available Now
Last year, Xilinx has added a one-of-a-kind speed-demon to our production-proven Virtex® UltraScale+™ HBM FPGA family, VU57P, which integrates 16GB of HBM DRAM with 58G PAM4 transceivers, our high-performance FPGA fabric, and high-speed connectivity hard IP. The production version of the VU57P devices are shipping now.
Built on Production-Proven Virtex UltraScale+ FPGAs
The Virtex UltraScale+ VU57P FPGA is built with IC and packaging technology Xilinx has mastered for years. Leveraging the modular UltraScale™ architecture and 3rd generation stacked silicon interconnect (SSI technology), we combined the in-production Virtex UltraScale+ 58G PAM4 FPGA and the Virtex UltraScale+ HBM FPGA to bring VU57P to market. The Virtex UltraScale+ VU57P FPGA brings a powerful set of capabilities ideal for the most compute and memory-intensive applications in data center, wired communications, and test and measurement applications.
Features and Benefits
With integrated HBM, the Virtex UltraScale+ VU57P FPGA alleviates crucial memory bottlenecks and power consumption with 20 times higher memory bandwidth at 75% lower power than discrete commodity memories, like DDR4. It integrates power-efficient compute capabilities with the highest memory bandwidth – up to 460GB/s – while leveraging the latest PAM4 high-speed transceivers to double the transmission rate vs. mainstream 25G transceivers.
Unique to the Virtex UltraScale+ HBM FPGAs, the integrated HBM controller and AXI port switch provide contiguous memory access to the entire 16G HBM and save 250 KLUTs, 370K FFs, at 4W lower power vs. soft IP implementations that competing FPGAs require.
Additionally, the VU57P FPGA offers a range of integrated high-speed connectivity such as 100G Ethernet with RS-FEC, 150G Interlaken, PCIe® Gen4 to simplify design efforts and accelerate time to market.
These features make the VU57P FPGA the premier choice for bandwidth-intensive workloads where fast data throughput, high performance processing, and high bandwidth, highly adjacent memory are critical requirements.