banner

Find Offerings

Switch to Partners

By Source
Offering Types
Region
Device Family
Market Segments
Program Level
Sort by
By Source
Offering Types
Region
Device Family
Market Segments
Program Level
Sort by

Heronic AI Accelerator Design Services

Heronic Technologies offers specialized design services for bespoke AI accelerator architectures on FPGAs, utilizing in-house automation and an IP portfolio to meet the unique demands of diverse AI applications like object detection and image classification. This approach directly tackles the inefficiency of generalized hardware in Edge AI and TinyML, enabling the rapid development of highly efficient, streaming-based accelerators. As a result, Heronic delivers unparalleled performance and microsecond latency, evidenced by state-of-the-art results on MLPerf Tiny and superior outcomes compared to GPU offerings.

High Dynamic Range (HDR) IP - using single exposure

Real-time, high-quality HDR IP with virtually zero latency, based on a single exposure. The IP is designed for FPGAs and can be embedded seamlessly in Gidel’s modular vision/imaging grabbing and image processing systems, which include PCIe boards and edge computers supporting GigE Vision, CoaXPress, Camera Link, and user-defined protocols.

High Speed JPEG Video Encoder IP Core

Very low latency Video Encoder supporting high resolutions and frame rates. Compression factor dynamically adjustable. Standard output format. An option allows the standard video stream to be decoded efficiently by GPU (multiple threads).

Offerings Image

High-Reliability FPGA Design Services

Installing functionality against soft error Implementing isolation to reduce error Support for neutron irradiation tests and consulting for soft error measures

HiPrAcc™ CS200D Dual-Agilex™ Computational Storage Module  

Highest performance PCIe Computational Storage Accelerator card with two Intel Agilex AGF027 FPGAs. up to 64TB on-card NVMe storage (8x M.2 SSDs) and Gen4 PCIe switch (Broadcom) interconnect.

HiPrAcc™ NC220 Low Profile PCIe Accelerator Card

Low-profile PCIe accelerator for data centers and edge AI, with up to 400 Gbps Ethernet, x16 PCIe Gen4, Agilex™ 7 FPGAs (006 through 027 density), 24/12 GB DDR4 and SMB PPS sync input

HiPrAcc™ NCS200 Networked Computational Storage Card

FPGA accelerator card featuring Agilex™ 7 Series FPGA with x16 Gen4 PCIe, 4x 100G networking interfaces, four on-card Gen4 M.2 NVMe SSDs (22110 and 2280) and two DDR4 UDIMM/RDIMM/LRDIMM slots — ideal for SmartNIC, computational storage and signal/image processing.

HiPrAcc™ NCS280-I CXL/PCIe Gen5 Accelerator 

High performance CxL/PCIe Gen5 accelerator with R-Tile, dual QSFP28 based Ethernet networking, and dual on-card Gen4 M.2 NVMe SSDs

HSDLC: HDLC & SDLC Protocol Controller Core

The HSDLC IP core implements HDLC and SDLC protocols, based on the Intel® 8XC152 GSC in SDLC mode with added HDLC and proprietary frame support. It connects as a peripheral to a host processor via APB or 80C51-like interfaces, with full interrupt support for efficient operation. Flexible design allows two independent TX/RX interfaces with support for full- or half-duplex, hardware flow control (RTS/CTS), collision detection, and programmable baud rates. Receive clock is derived from incoming data or supplied externally. Available in Normal and Safety-Enhanced (TMR, DO-254 DAL-A) versions, the HSDLC core is fully synchronous, scan-ready, verified, and delivered in Verilog RTL or FPGA netlist. Deliverables include scripts, testbench, and complete documentation.