The explosive growth of big data and increasing complexity in the modern data center demands hardware accelerators to offload a broad range of critical data center applications from CPUs. Cloud providers and enterprises must continually optimize for complex applications, adapt acceleration and resources across multiple workloads and deliver a broader range of offloads. These challenges require scalable, low-latency hardware acceleration while retaining rapid reconfigurability for changing and emerging workloads. The Alveo SN1000 family of composable SmartNICs meets these challenges with software-defined hardware acceleration. Revolutionary Xilinx composability empowers providers and enterprises to effortlessly support new protocols, build custom offloads, and deploy efficient and fluid application-specific data paths using P4 or high-level synthesis (HLS). SN1000 SmartNICs deliver protocol-level programmability at linerate performance, and are powered by a Xilinx 16nm UltraScale+ architecture FPGA and an A72 Arm processor subsystem with support for up to 16 cores. Starting with the SN1022 100Gb/s composable SmartNIC, the Alveo SN1000 family provides a comprehensive suite of solutions for network, storage, and compute acceleration functions on a single platform.
High Performance NIC
100Gb connectivity with industry leading small packet performance and low-latency on PCIe Gen 4.
Software Defined Infrastructure
Open Virtual Switch (OVS) and Virtio offloads with efficient switching and routing along with a powerful Arm A72 processor for bare metal services and control plane offloads.
NFV Workload Acceleration
Decoupling of network functions and services from dedicated hardware for efficient and high-performance acceleration.
Security
Root-of-Trust secure boot and technology to ensure the integrity of the firmware and hardware.
QoS
Support for traffic shaping and management mechanisms with dedicated independent queues in hardware.
Programmability
P4 and HLS programming allowing the data plane to be fully software-defined for cloud-scale deployments.
