Computing and network acceleration
LightBoat-2300 LightBoat series FPGA accelerator card
A high-performance FPGA accelerator card based on AMD UltraScale+™ architecture and Virtex UltraScale+ XCVU23P chip, specially designed for scenarios such as intelligent networks, data center acceleration, AI inference, hardware security, and industrial edge computing.
corecharacteristic
Pure hardware acceleration
IO command parsing and verification operations are implemented through FPGA hardware logic, and the CPU usage is reduced to zero.
ultra low latency
End-to-end latency can be controlled below 10 microseconds, making it suitable for scenarios such as high-frequency trading and real-time data analysis.
High throughput performance
The measured 4KB random read can reach 4.89 million IOPS, and the sequential read bandwidth is as high as 21.8GB/s.
Extreme energy efficiency
Typical power consumption is only 9.3W, which saves more than 85% of energy compared with traditional solutions.
Industrial grade reliability
Compatible with industrial-grade environments, taking into account energy efficiency and performance.
Product details
Product overview
The LightBoat-2300 series FPGA accelerator card is a high-performance hardware acceleration product independently developed by LUISUAN Technology and will be officially released in February 2025. This series adopts AMD UltraScale+™ architecture and is equipped with Virtex UltraScale+ XCVU23P high-performance FPGA chip. It is specially designed for scenarios such as intelligent networks, data center acceleration, AI inference, hardware security, and industrial edge computing. It can provide chip-level hardware acceleration capabilities for various computing and data applications.
Detailed explanation of application scenarios
1. IP Core development and verification platform
With its rich hardware resources and complete debugging interface, the LightBoat-2300 series has become an ideal platform for ASIC/IP Core prototype verification:
- Hardware prototype verification: Utilizing 2.252 million logic cells and 99Mb UltraRAM to support hardware simulation and verification of complex digital circuits
- Algorithm hardware verification: verify the functional correctness and performance indicators of the algorithm hardware implementation before tape-out
- Interface protocol verification: supports protocol consistency testing of high-speed interfaces such as PCIe Gen4 and 100G Ethernet
- Debugging and testing: Equipped with USB Type-C debugging interface (JTAG + UART×4) and status LED indicators to facilitate collaborative debugging of software and hardware
- Rapid iterative development: FPGA reprogrammable features support rapid design iteration and shorten the IP Core development cycle
2. Encrypted computing and hardware security
Based on the hardware-level parallel computing capabilities of FPGA, the LightBoat-2300 series has excellent performance in the field of cryptography acceleration:
- National secret algorithm hardware acceleration: integrated SM2/SM4 national secret algorithm hardware engine, performance improved several times compared with pure software solutions
- Symmetric/asymmetric encryption: supports hardware-level acceleration of mainstream cryptographic algorithms such as AES, DES, RSA, and ECC.
- Hash operation acceleration: supports high-speed calculation of hash algorithms such as SHA256 and SM3, suitable for blockchain and data integrity verification
- Homomorphic encryption acceleration: supports homomorphic encryption operations in privacy computing, which can increase efficiency by more than 50 times compared to CPU implementation
- SSL/TLS offload: As a 200G SmartNIC, it implements hardware encryption and decryption of network data packets, significantly reducing server CPU load.
- Quantum safe encryption: Supports anti-quantum algorithms such as Lattice-based Cryptography to meet future security needs
3. Password management and key security
The LightBoat-2300 series provides hardware-level key management capabilities to ensure the security of the entire life cycle of the cryptographic system:
- Hardware Security Module (HSM): Build a trusted execution environment based on FPGA to achieve secure generation, storage and use of keys
- Key life cycle management: supports full-process hardware-level protection of key generation, distribution, rotation, and destruction
- True random number generation: Use the physical characteristics of FPGA to generate high-quality random numbers to meet cryptographic security requirements
- Key isolation storage: ensure the separation of keys and business data through hardware isolation mechanism to prevent side channel attacks
- Access control: Hardware-based identity authentication and rights management to prevent unauthorized access
- Audit and Compliance: Record all cryptographic operation logs to meet compliance requirements such as GDPR, Class A Safeguard 2.0, and cryptography laws.
Other typical application scenarios
- data center network: 200G smart network card (SmartNIC), NVMe-oF acceleration - reduce CPU load by 85%+, and delay to less than 10μs
- AI Inference and LLM: KVCache extension, large model inference acceleration - solves the video memory bottleneck and supports the "unlimited video memory" solution
- NVMe RAID: 16-bay NVMe SSD hardware RAID - bandwidth increased by 34 times, supports RAID0/1/5/10
- Fintech: High-frequency trading, real-time risk control, option pricing - trading order generation time compressed to microseconds
- privacy computing: Multi-party secure computing - hardware-level privacy protection, data does not leave the domain
- industrial edge: Machine vision, PLC control, real-time quality inspection - millisecond response, sub-pixel detection accuracy
Product form and ecological support
- Physical form: Standard half-height, half-length (HHHL) PCIe accelerator card, no additional cables required, plug and play
- Software ecosystem: supports Vivado development environment, provides complete SDK and drivers, and is compatible with mainstream AI frameworks and cloud platforms
- Security certification: Supports the national security level two qualification requirements and meets the security standards of sensitive industries such as finance and government.
- Expansion capability: A single card supports up to 16 NVMe SSDs, which can be upgraded to 24/32
The LightBoat-2300 series is not only a high-performance computing accelerator card, but also a full-featured hardware security platform. Its IP Core development and verification capabilities facilitate rapid iteration of chip design, hardware-level encryption calculations ensure data processing security, and key management functions meet compliance requirements. With the powerful performance of XCVU23P and the in-depth optimization of LUISUAN Technology, this series has become the core infrastructure in fields such as intelligent computing centers, financial technology, and privacy computing.
technologySpecification
| parameter | LightBoat-2300 |
|---|---|
| core chip | AMD Virtex UltraScale+ XCVU23P |
| logical unit | 2,252K (2.252 million) |
| DSP slice | 1,320 |
| UltraRAM | 99Mb |
| Computing power | 4.1 TOPS |
| high speed transceiver | 58 (34-way GTY@32.75Gb/s + 4-way GTM@58Gb/s) |
| Onboard memory | 4GB DDR4(1600MHz)+ 2Gb QSPI Flash |
| network interface | Dual 100G Ethernet (QSFP28) |
| PCIe interface | PCIe Gen4x8×2 / PCIe Gen3x16 |
| Debug interface | USB Type-C(JTAG、UART×4) |
| Power supply design | PCIe slot power supply (75W Max) + external ATX power supply (120W Max) |