How to Create Low-Power FPGA Designs


Low power

Industrial 4.0

Internet of Things

Video Processing / Surveillance

Compute and Storage

Wireless and Infrastructure


Accelerating Your Innovation

A message from our CEO on COVID-19

Our mission is to drive the future of edge computing with our Trion® FPGA silicon platform.

At the heart of Trion FPGAs is our disruptive Quantum™ technology, which delivers a 4X Power-Performance-Area advantage over traditional FPGA technologies. Trion FPGAs, offering 4K to 200K logic elements (LEs), have a small form factor, low power, and are priced for high-volume production. Our Efinity® Integrated Development Environment provides a complete FPGA design suite from RTL to bitstream generation.

With their Power-Performance-Area advantage, Trion FPGAs address applications such as custom logic, compute acceleration, machine learning, and deep learning. With the Efinity software, our customers can migrate seamlessly from a standalone FPGA or full system to Quantum ASIC for ultra-high-volume production.

Register Now

Date: Tuesday June 30, 2020

Time: 9:00 AM PDT

Register for webinar

Webinar: How to Create Low-Power FPGA Designs

Power usage is a critical element of system design. Battery-powered, handheld devices, medical systems, and networking applications are the most obvious power-sensitive systems. However, even power-hungry applications like cloud application servers, hardware accelerators, and crypto currency mining, can receive appreciable operating cost savings by reducing power by just a few percentage points.

This webinar focuses on design techniques for saving power, including system level design, RTL architectural decisions, and low-level techniques, as well as the role of process and power-friendly FPGA features.



In recent years, the automotive industry has been experiencing a paradigm shift in the way it embraces technology. Modern vehicles incorporate hundreds of sensors such as 360 degree surround video, LiDAR sensors for automatic driver assistance, and lane departure assistance sensors. There is even a roadmap that shows a credible path to truly autonomous driving. Inside the vehicle, the driver has an unprecedented wealth of information while the passengers enjoy an immersive in-vehicle entertainment experience. Increasingly, vehicle sensors are connected with automotive Ethernet and feed data back to centralized intelligent compute and storage systems. The modern vehicle has become a data center on wheels.

With embedded MIPI interfaces and the reprogrammable Quantum™ fabric, Efinix® FPGAs are uniquely positioned to address the needs of automotive applications. T13 FPGAs and above feature hardened MIPI interfaces that, alongside a wealth of GPIO, ensure universal connectivity for sensor fusion while the Quantum fabric delivers the compute capability for low latency, intelligent processing of realtime data. From optical and LiDAR sensors to artificially intelligent data processing to display and motor control, Efinix FPGAs deliver the driving power behind automotive applications.


Industrial 4.0

Today’s industrial applications generate unprecedented quantities of data. Sensors monitor not only every parameter of a production line’s output but also the performance of the production line itself. Machines are instrumented to detect any sign of wear or potential failure to allow for preventative maintenance to be performed thus avoiding costly down time. Latency critical processes require that compute and artificially intelligent data processing move closer to the source of the data where it can be analyzed and interpreted in real time.

Efinix® FPGAs have been widely adopted in industrial applications providing sensor fusion, data analysis and real time motor control. They have a cost structure that allows them to be widely deployed and a power and performance envelop that makes them ideally suited for industrial environments. With the growing focus on edge compute and artificially intelligent data processing, Efinix devices will be increasingly deployed to deliver the compute requirements of the next generation of industrial automation.


Video Processing and Surveillance

The past ten years have seen an explosion in the adoption of machine vision and image processing. Once considered too costly to be economically deployable at scale, the decreasing cost of sensor systems and an increase in compute power to process the video has resulted in the broad adoption of vision systems. However, two modern trends threaten to stretch the capabilities of image processing sub systems to their breaking point. The first is the availability of ever higher resolution sensors that are capable of outputting video at higher frame rates resulting in much higher data rates. The second is the proliferation of artificial intelligence and machine learning, which enables computer systems to extract more intelligence from the video, although at the cost of significantly higher processing requirements.

Only the very highest volume applications can support the considerable investment of custom designed video processing SoCs. For other applications, salvation is found with programmable logic. Efinix® FPGAs were designed with video processing applications in mind. They bring inherent parallelism and the possibility to design efficient pipelined architectures. They have embedded MIPI interfaces for natural connection to video sensors and an infinite combination of artificially intelligent processing systems can be instantiated in their programmable fabric. With low power consumption and cost effective silicon efficiency, Efinix Trion® FPGAs have been widely deployed in a broad range of video processing and surveillance applications where they provide image enhancement, feature extraction and artificially intelligent analysis of the video data.

IoT - smart home

Mobile and IoT

With the increased adoption of WiFi and the introduction of revolutionary wireless technologies such as 5G, there is no lack of data bandwidth right to the edge of the network. This, in turn, has led to an explosion in the number and diversity of connected devices. From wireless video doorbells to voice activated personal assistants to connected refrigerators it seems that every consumer device is connected to the internet and promises some level of intelligence. All of these connected devices share similar design challenges. They must operate on low power and in a small footprint, be flexible enough to adapt to changing standards and evolving technologies, and must be released in a timely manner to guarantee their share of this rapidly evolving market.

The Quantum™ fabric lies at the heart of all Efinix® FPGAs and was designed from the ground up to deliver low-power operation with an area efficiency up to four times that of other FPGAs. The Trion® family comprises FPGAs with as few as 4,000 logic elements in the T4 right up 200,000 logic elements in the T200. Their inherent area efficiency gives them a cost structure that can be taken into mass production ensuring a seamless transition from prototype to end product and a corresponding reduction in time to market. Hardened IP blocks such as MIPI for camera sensors reduces system bill of materials while the easily configured Quantum fabric delivers the compute power required for the most demanding artificial intelligence and machine learning applications.


Compute and Storage

Major challenges in the modern data center include an increasing diversity of workloads and an explosion in the volume of data being processed. To counter these trends, data center operators are increasingly turning to customizable hardware accelerators tailored to the particular workload and are attempting increase processing efficiency by moving compute to the data rather than data to the compute.

The Efinix® Quantum™ fabric was designed to address these two trends. By deliberately decoupling the fabric from potential I/O resources, Efinix created a reconfigurable fabric accelerator, one example of which is the Trion® FPGA family. This fabric naturally lends itself to being embedded into other compute structures such as storage controllers and is easily configured as an adaptive compute accelerator. The Efinix Quantum fabric has been deployed as an accelerator in compute applications from consumer IoT to professional video surveillance and has been adapted to deliver compute horsepower in revolutionary in-situ storage architectures.

Connected Infrastructure

Wireless and Infrastructure

The rollout of 5G technology is set to redefine the way we think about wireless infrastructure. Cloud RAN deployments and multi-access edge compute resources running cloud native applications will deliver low latency, geographically relevant services. Whether it be the support infrastructure for self-driving vehicles, content delivery for mobile augmented reality, or low latency networking for industrial automation, the one common theme will be an enormous diversity in the types of workload and a demand for low latency. Hardware acceleration will be key to delivering on these requirements.

By delivering a broad spectrum of compute and reconfigurability options in small footprints and with very low power consumption, Efinix® Trion® FPGA are ideally suited to edge compute applications. From acceleration cards in the edge compute aggregation units themselves to AI and machine learning accelerators in the multitude of IoT devices that will connect to the network, Efinix devices will be driving the revolution in wireless infrastructure.