-
Card Specifications
- FPGA Device
- XCU26 device based on 16nm UltraScale+™ architecture
- 1,030K LUTs
- On-Board Processor¹
- Discrete 16-core 64-bit Arm® Cortex®-A72 at 2.0 GHz
- 8 MB cache
- Performance
- 200 Gb/s Full Duplex Throughput
- Network Interface
- 2x QSFP28 cages on front panel connected directly to FPGA via 8 transceivers.
- Each QSFP28 supports 100 GbE, 4X25 GbE, or 4X 10 GbE
- Direct-attach copper or optical transceiver
- Host Interface
- PCIe® Gen3 x16, or 2x Gen4 x8
- On-Board Memory
- 2x 4GB x72 DDR4-2666 (to FPGA)
- 1x 4GB x72 DDR4-266 (to Arm Processor)
- Form Factor
- Full height, half length (FHHL)
- PCIe CEM card
- Single slot
- Full height, half length (FHHL)
- Electrical Power & Thermal
- 75W
- Passive cooling
- Tool Flow
- Vivado™ Design Suite
- OpenNIC project (open source reference design available on GitHub )
- Vitis™ Net for network programming language support
AMD Alveo™ U45N Network Accelerator Card
The AMD Alveo™ U45N network accelerator is an FPGA-based platform that delivers low-latency, 2x 100G line-rate performance for infrastructure workloads in the data center - freeing up precious server CPU cycles from infrastructure tasks to run business-critical applications.
- 价格: $2,371.00
- 产品编号:
- A-U45N-P08G-PQ-G
- 交付周期: 12 weeks
Explore Partner Network
Overview
Hyperscalers, cloud service providers, and telco operators can adapt to their own infrastructure requirements by customizing virtual switching (e.g., OVS), security workloads (e.g., IPsec) and storage protocols. When combined with the Vivado™ tool flow and a comprehensive OpenNIC reference-based design, the Alveo U45N card delivers the hardware platform, tools, and IP for FPGA developers to build and differentiate their own network acceleration solutions.
Key Features & Benefits
Hardware Acceleration at 2x100G Line Rate Performance
- 2x100G Ethernet line rate performance that scales with network traffic
- Deterministic, low latency for reliable performance and QoS
- Provides redundancy for data center workloads
Adaptable for Custom and Evolving Workloads
- 1 million LUTs of FPGA fabric for full customization
- Accelerate custom workloads in networking, security, and storage
- Discrete Arm® Cortex®-A72 processor for optional control plane functions
Vivado™ Tools and Reference Design for Ease of Development
- Vivado Design Suite for RTL design and hardware customization
- Vast catalog of building block IP for networking
- Comprehensive OpenNIC reference design with DPDK drivers