Webinar Registration: Intel oneAPI HPC with FPGAs
BittWare On-Demand Webinar Using Intel® oneAPI™ to Achieve High-Performance Compute Acceleration with FPGAs Join BittWare and Intel as we look at oneAPI™ with a focus
The First FPGA-Powered CXL NIC
Liquid-Markets-Solutions (LMS) have done something unique: moved the full network stack into an FPGA. What’s even better is that FPGA is Agilex I-Series from Intel, with the groundbreaking bandwidth of PCIe 5.0 and low-latency performance of CXL.
Called ÜberNIC, this family of NIC cards is built on BittWare hardware for enterprise-class deployment in datacenters.
Fact: today’s market feeds regularly pass 850 Mb/s—the point at which even the best competitive NIC’s network stack begin to dramatically lose sufficient latency performance.
Enter ÜberNIC, the first commercially-available FPGA NIC with the entire network stack in accelerated hardware, allowing for consistent low latency—ready for today’s markets and the future.
Raw performance improvement that’s consistent through the critical peak market feed moments when you need it most!
More fiber pairs per card means getting more out of each server. Powerful enough for ancillary services like PCAP and timestamping on the same card.
Support for latest PCIe 5.0, CXL, and link speeds up to 400G.
By moving the complete network stack into FPGA hardware, the ÜberNIC achieves unmatched performance.
Single-slot card supports up to 16× 10/25G fiber pairs
Central to the performance gains of ÜberNIC over competitors is the Intel Agilex 7 FPGA. This chip brings the core features like PCIe 5.0 and CXL—plus I/O giving double (or more) the port density of even top competitors.
At its core, ÜberNIC is a low-latency focused NIC. With the combination of full hardware acceleration and low-latency PCIe transfers thanks to CXL, you get unmatched performance—even during peak market times.
BittWare brings enterprise-class products to market using the latest technology such as Agilex FPGAs. Today’s high-speed interfaces require significant care in the design and manufacturing process to make a reliable, high-quality product at scale.
Pictured is our manufacturing facility at ISI, a Molex company, in California.
Only FPGA-based solutions with bundled hardware network allows for all these extensible functions.
Even up to 3,000 Mb/s ÜberNIC performs with low latency. The cards can run in a variety of servers supporting PCIe 3/4/5 and CXL as shown in the different lines on the performance graph.
Tested By Naros.TaSR
50,000 Frames Per Throughput Rate | 10GbE | 64B UDP Payload (110B Frame) | Transceiver to Transceiver L2 SOF to SOF via Linux User-Space | Excluding Business Logic
Ready to learn more about ÜberNIC? Reach out to BittWare to learn more!
"*" indicates required fields
BittWare On-Demand Webinar Using Intel® oneAPI™ to Achieve High-Performance Compute Acceleration with FPGAs Join BittWare and Intel as we look at oneAPI™ with a focus
PCIe FPGA Card XUP-VV4 UltraScale+ FPGA PCIe Board with VU13P 4x 100GbE and up to 512GB DDR4 Need a Price Quote? Jump to Pricing Form
White Paper Building BittWare’s Packet Parser, HLS vs. P4 Implementations Overview One of the features of both BittWare’s SmartNIC Shell and BittWare’s Loopback Example is
PCIe FPGA Card XUP-VVH UltraScale+ FPGA PCIe Board with Integrated HBM2 Memory 4x 100GbE Network Ports and VU37P FPGA Need a Price Quote? Jump to