Home / Products / FPGA NVMe RAID accelerator card

Computing and network acceleration

FPGA NVMe RAID accelerator card

The NVMe RAID hardware accelerator card based on the LightBoat-2300 FPGA platform independently developed by LUISUAN uses pure FPGA hardware logic to achieve IO parsing and verification acceleration, reducing the CPU occupancy to zero, and the theoretical bandwidth is as high as 128GB/s.

FPGA NVMe RAID accelerator card
128GB/s Theoretical bandwidth
~8μs delay increment
3-4× Bandwidth improvement
0 CPU usage
0/1/5/10 RAID level
Core Features

corecharacteristic

Pure hardware acceleration

IO command parsing and verification operations are implemented through FPGA hardware logic, and the CPU usage is reduced to zero.

Parallel architecture innovation

NVMe commands and data are transmitted separately, reducing data transfer links and greatly reducing transmission delays.

Extreme bandwidth

The theoretical bandwidth under the configuration of 16 PCIe Gen4 NVMe SSDs reaches 128GB/s, far exceeding the 32GB/s limit of the traditional solution.

ultra low latency

The IO latency is only about 8 microseconds higher than that of a bare disk, meeting the needs of high-frequency transactions and real-time computing.

Enterprise-grade reliability

Supports RAID 0/1/5/10, virtual disk is plug-and-play, greatly simplifying deployment and maintenance.

Autonomous and controllable

Authorized by the national invention patent, it has complete source code level development capabilities and can evolve to ASIC tape-out.

Product details

Product overview

On January 20, 2026, LUISUAN Technology officially released its independently developed blockbuster product-FPGA NVMe RAID accelerator card. This product is based on the company's independently developed LightBoat-2300 FPGA accelerator card platform, equipped with AMD Virtex UltraScale+ XCVU23P high-performance FPGA chip, and fully supports PCIe Gen4 high-speed interface, bringing an extremely fast and secure storage experience to key areas such as data centers, cloud computing, and financial technology.

Different from traditional software RAID and serial hardware RAID that rely on CPU computing power, this product uses pure FPGA hardware logic to implement IO command parsing processing and verification operation acceleration, reducing CPU occupancy to zero, allowing valuable computing resources to fully serve customer business applications.

System Architecture

FPGA NVMe RAID system schematic diagram

The product adopts an innovative parallel hardware architecture and cooperates with the NVMe command and data separation transmission mechanism to significantly reduce data transfer links, greatly reduce transmission delays, and achieve the ultimate release of SSD performance.

Performance breakthrough

The innovative architecture adopted by this accelerator card has been authorized by the national invention patent (Patent No.: ZL 2024 1 1505866.8), which reflects the independent research and development capabilities of LUISUAN Technology in the field of hardware acceleration. The measured data is impressive:

  • A RAID 5 array composed of 8 NVMe SSDs, the read and write bandwidth performance is more than 3-4 times that of traditional solutions
  • IO latency only increases by about 8 microseconds compared to bare disk
  • Under the configuration of 16 PCIe Gen4 NVMe SSDs, the theoretical bandwidth can reach 128GB/s, far exceeding the 32GB/s limit of the traditional solution

Flexible configuration and enterprise-grade reliability

From the beginning of product design, we fully consider the diverse needs of enterprise-level applications:

  • Fully supports RAID 0, 1, 5, 10 and other levels, allowing users to flexibly balance performance and data security according to different scenarios.
  • A single card supports up to 16 NVMe SSDs, and can later be upgraded to support 24 or even 32 SSDs.
  • A single card supports the creation of 1-2 virtual disks, which are presented as standard NVMe devices and perfectly adapt to the operating system's native NVMe driver, plug and play.
  • Adopts half-height, half-length AIC standard form, powered through PCIe slot, no need for external power cord, backup battery or supercapacitor
  • No external hard disk interface cable is required, greatly simplifying deployment and maintenance complexity and reducing total cost of ownership.

Application Scenarios

With the comprehensive advantages of high performance, low latency and high energy efficiency ratio, this product shows great application potential in multiple key areas:

Data Center and Cloud Computing

Accelerate big data analysis and distributed storage, significantly improving cloud service IOPS performance.

Fintech

Meet the stringent requirements for extremely low latency for high-frequency trading and real-time risk assessment.

Communication field

Optimize 5G base station efficiency and network traffic deep packet inspection capabilities.

Special industries

It is suitable for scenarios such as civil aviation radar detection and satellite communications that require extremely high real-time performance.

Industrial manufacturing

Support the real-time processing requirements of industrial automation, machine vision and autonomous driving.

scientific research

Accelerate data-intensive research processes such as genetic sequencing and astrophysics simulations.

Xinchuang Industry

Provide a solid high-performance storage base for domestic information technology application innovation.

Autonomy, controllability and future planning

LUISUAN Technology has always adhered to the path of independent research and development of core technologies, and has complete source code-level development capabilities to ensure complete independent control of the technical route. Based on the current NVMe RAID Core core IP implemented in FPGA, LUISUAN Technology has the possibility of tapeout to ASIC chips, which will further reduce solution costs and power consumption and promote the popularization of products to a wider market.

The emergence of the LUISUAN FPGA NVMe RAID accelerator card not only completes the product matrix of LUISUAN Technology in the field of FPGA acceleration, but also brings a more efficient, safer and more reliable storage experience to enterprise customers, helping all walks of life to seize the opportunity in the wave of digitalization.

Technical Specifications

technologySpecification

parameter FPGA NVMe RAID accelerator card
core platform LightBoat-2300 FPGA accelerator card
core chip AMD Virtex UltraScale+ XCVU23P
host interface PCIe Gen4 x16
Support RAID levels RAID 0 / 1 / 5 / 10
Maximum bay 16 NVMe SSDs (upgradeable to 24/32)
virtual disk 1-2 (presented as standard NVMe devices)
Theoretical bandwidth 128GB/s (16-drive PCIe Gen4 configuration)
IO latency Only about 8μs more than bare disk
Product form Half height half length AIC (HHHL)
Power supply design PCIe slot powered, no external power supply/backup battery/supercapacitor required
patent National invention patent ZL 2024 1 1505866.8