ULL Fintech Accelerations

iWave delivers ultra-low-latency (ULL) FPGA acceleration solutions for modern financial markets, combining hardware-based market-data processing, customizable trading logic, and AI inference. Built on iWave PCIe Accelerator Hardware, these solutions support latency-sensitive trading applications with deterministic performance and reduced host CPU overhead.

CONTACT SALES

Documents

AI Powered Tick to Trade HFT Solution

Integrate iWave’s HFT Framework and Xelera Silva Trading
AI Inference on iWave PCIe Accelerator Hardware for a
unified trading solution combining FPGA-based market
processing, customizable trading logic, and AI inference.

HFT Accelerator Solution

Combine iWave PCIe Accelerator Hardware with the Xelera
Silva Accelerator to offload AI inference from the host CPU
to the FPGA, enabling concurrent strategy execution with
high performance and low latency.

iWave Tick to Trade HFT Framework

iWave’s HFT Framework provides a white-box pricing and
trading engine that can be customized with user-defined
logic or optionally implemented using Xelera Silva logic
running on iWave PCIe Accelerator Hardware.

Features Details
HFT Framework
  • Tick-to-trade latency of < 300 ns
  • Complete trading path implemented in FPGA fabric, with no host CPU or PCIe crossing between market-data ingress and order egress
  • Single-clock-domain, fully pipelined trading datapath
  • Deterministic processing for predictable trading response
Network Interface
  • 10GBASE-R low-latency MAC/PCS implemented on GTY transceivers
  • Direct 32-bit PMA interface with transceiver buffer bypass for reduced datapath latency
  • Ultra-low-latency UDP/IP TX/RX stack optimized for market-data reception and packet processing
  • TCP/IP interface optimized for low-latency order placement and exchange communication
Market Data & Order Book
  • NASDAQ ITCH 5.0 market-data parser
  • Symbol filtering at ingress
  • Per-symbol, price-level order-book reconstruction in FPGA fabric
  • Incremental Best Bid and Offer (BBO) updates without full book rescanning
Trading Strategy & Order Entry
  • User-programmable strategy region with AXI4-Stream market-event input and order-request output
  • Reference market-making and fair-value pricing strategies
  • NASDAQ OUCH 5.0 order encoder
  • SoupBinTCP session layer with hardware TCP offload
  • Parallel CRC32 processing across the datapath
Software & Host Interface
  • Linux host driver and API for configuration and framework management
  • Symbol subscription, risk-limit and strategy-parameter configuration
  • Telemetry for link status, order and risk-rejection counters
  • PCIe DMA for market-data and order-log drop copy
Platform Details
G35-XFibre PCIe Accelerator
  • 3/4 Length PCIe Gen3 x16 Host Interface
  • Dual Front Panel QSFP-DD/QSFP28 Cages (3 x 10GbE) via breakthrough cables
  • 254mm x 111.15mm, 3/4 Length, Full Height, Dual Slot with Active Cooling PCIe Card
  • Quad/Dual Arm Cortex-A53 @1.5GHz, Dual Cortex-R5 @600MHz
  • Up to 1143K Logic cells & 522K LUTs in FPGA programming logic
  • Product Info: Zynq™ Ultrascale+™ SmartNIC – 100G
G63-XFibre PCIe Accelerator
  • PCIe Gen5 x8 Host Interface
  • QSFP-DD Connector x 3 (3 x 10GbE) via breakthrough cables
  • Full height, ¾ Length, Single Slot with Active Cooling PCIe Card
  • Dual-core Arm Cortex-A72 @1.7GHz, Dual-core Arm Cortex-R5F @800MHz
  • Up to 3763K Logic cells and 1720K LUTs in FPGA programming logic
  • Product Info: Versal™ Premium SmartNIC – 400G
G43-AFibre PCIe Accelerator
iG-G58-AFibre PCIe Accelerator
video-thumbnail
youtube-play-icon
video-thumbnail
youtube-play-icon
video-thumbnail
youtube-play-icon
img

JESD204C based ADC with ZU19EG FPGA

Learn More
img

Discover the Scalability of SmartNIC Portfolio

Learn More
img

Multi Board Synchronization on RFSoC Board & Systems

Learn More
img

ZU11EG MPSoC SoM on Quantum Computing

Learn More
img

ZU49/ZU39/ZU29DR RFSoC SoM on Radio Telescope

Learn More
img

Custom Design ZU19EG MPSoC SoM on LiDAR

Learn More