All News | Boards | Chips | Devices | Software | Archive | About | Contact | Subscribe
Follow LinuxGizmos:
Twitter Facebook Pinterest RSS feed
*   get email updates   *

Hailo-8 NPU ships on Linux-powered Lanner edge systems

Jun 1, 2021 — by Eric Brown 1,373 views

Lanner is deploying Hailo’s up to 26-TOPS, 3-TOPS per Watt Hailo-8 NPU on two fanless edge computers running Linux: the Coffee Lake and PCIe equipped LEC-2290 and the compact, Apollo Lake based LEC-7242.

Hailo and Lanner Electronics announced a partnership to deploy the Hailo-8 neural processing unit (NPU) on selected Lanner edge computers running Linux, starting with the Intel Coffee Lake based LEC-2290 and Intel Apollo Lake powered LEC-7242. Hailo also announced that the Hailo-8 was named Best Edge AI Processor in the 2021 Edge AI and Vision Product of the Year Awards hosted by the Edge AI and Vision Alliance.

Hailo-8 M.2

The up to 26-TOPS, 3-TOPS per Watt Hailo-8 NPU uses a “proprietary novel structure-driven” Dataflow architecture that differs from the Van Neumann architecture used on most neural processors. The architecture achieves low-power memory access by implementing a distributed memory fabric combined with pipeline elements.


The Hailo-8 has been adopted by Foxconn for a BOXiedge edge AI server, which runs Linux on Socionext’s SynQuacer SC2A11 SoC with 24x Cortex-A53 cores. The NPU is also being deployed by Leopard Imaging for its EdgeTuring system, which runs Linux on a quad -A7 and ISP equipped Socionext SC2000 SoC and features dual 4K@30 Sony IMX477 sensors.

On the Lanner systems, the Hailo-8 is deployed using Hailo’s Hailo-8 M.2 AI Acceleration Module, which is also used on the EdgeTuring system. The M.2 M-key 2242 add-on, which is billed as the world’s highest performance AI M.2 module, provides a PCIe Gen3 x4 interface.

Applications for the Lanner systems with Hailo-8 include smart city, smart retail, and industrial operations “that require hundreds of cameras generating video streams that need to be processed locally, quickly, and efficiently with minimal latency,” says Hailo. “Both solutions process multiple video streams in real-time on a single device, while securely transmitting curated metadata and insights with robust onboard network connectivity. such as video analytics, traffic management, and access control.”

LEC-2290A and detail view
(click images to enlarge)

The Coffee Lake based, Ubuntu ready LEC-2290 is available in two models, a compact LEC-2290Z and a larger LEC-2290A. The latter adds a base unit with dual, removable SATA bays, dual PCIe x16 slots, and a second M.2 slot with dual SIM card slots. The LEC-2290A also adds 4x PoE ports.

The Ubuntu-ready LEC-2290 models support up to 32GB of DDR4 and provide 2x GbE, 2x HDMI, 6x serial, 4x USB 3.0, and single DP, M.2, and mini-PCIe connections. The ruggedized, -20 to 45°C tolerant system has a 9-30VDC input.

LEC-7242 and detail views
(click images to enlarge)

The more compact LEC-7242 runs Linux or Windows on a dual- or quad-core Intel Apollo Lake SoC. The system supports up to 8GB LPDDR4 and ships with 64GB eMMC. Other features include 2x GbE, 2x USB 3.0, and single RS-232/422/485, DP, and HDMI ports.

The LEC-7242 supplies a mini-PCIe socket for LTE and an M.2 B+M-key 3042 for LTE or WiFi. Dual external SIM card slots are available along with WiFi and LTE modules that offer CE, FCC, and PTCRB certification.

“Our expertise in creating effective purpose-built hardware platforms, combined with Hailo’s unparalleled AI module performance, will provide industries with a reliable vision-based solution that is low power, low latency and cost-effective,” stated Geoffrey Egger, VP and GM of Lanner Intelligent Edge BU.

Further information

No pricing or availability information was provided for the Lanner LEC-2290 and LEC-7242 featuring the Hailo-8 M.2 AI Acceleration Module. More information may be found in the Lanner’s announcement and Hailo-8 bundle page, and more on Hailo-8 may be found at Hailo’s website.



(advertise here)

Print Friendly, PDF & Email

Please comment here...