When you have a product that is difficult to explain, it’s often helpful to provide simple examples to help people understand what it does and why it’s important. Such is the case with the specialized type of semiconductor chips known as FPGAs (Field Programmable Gate Arrays), a powerful but little-understood component that is used in everything from PCs and servers to industrial machines, telecommunications networks and military equipment.
With the launch of a new line of mid-range FPGAs from Semiconductor latticethe company took this approach by highlighting a simple visual demonstration of its new chip identifying and tracking vehicles from a connected car’s camera, a critical capability for assisted and autonomous driving functions.
As the name suggests, FPGAs can be configured and programmed to perform certain functions directly in the hardware. This gives them performance advantages for a number of specific applications. What’s more, they can be reprogrammed after installation, making them ideal not only for developing products (which require constant software and hardware updates), but also for adding new features to existing products.
In the world of AI and edge computing, it’s not uncommon to have algorithms for things like identifying and tracking objects that are updated regularly. FPGAs may be ideal for such applications precisely because of their programmability.
Over the past few years, Lattice Semi has focused its efforts on building a complete line of tiny, low-power FPGAs, which it has dubbed Nexus. In addition, the company has expanded its software offerings with a range of development tools such as Radiant and Propel, as well as a full suite of software platforms such as SenseAI, mVision and ORAN that are optimized for specific markets or applications.
As successful as Lattice Semi has been with its low-power Nexus line, this category only represents a third of the entire FPGA market. Another third are so-called mid-range FPGAs, which offer more performance than their low-power counterparts but require more power to run.
Lattice Semi’s latest move brings them into this mid-range market with a new line of FPGAs dubbed Avant. These new chips offer a 5x increase in capacity, a 10x increase in bandwidth, and a 30x performance jump over the Nexus line. Importantly, the new Avant chips work with all of the company’s existing software, making the transition easier for existing customers and allowing them to leverage their software development work.
In terms of competitiveness, Lattice Semi claims Avant performance is 2x faster and 2.5x more power efficient than similar chips from Intel (Altera/Stratix) and AMD (Xilinx). Avant chips are also 6 times smaller than competitors, potentially making them suitable for PCB and design applications that other larger chips would not be able to fit into.
Lattice Semi says they plan to introduce a number of products from their Avant line (along with new additions to the Nexus). The first chip is called Avant-E and it has been optimized for edge computing applications, although the company was quick to point out that it could (and is) used in other types of applications as well.
The premium Avant-E 500 processor delivers 5 TOPS (tera operations per second) performance for AI applications running at 350 MHz. In theory, this makes it well-suited for running AI inference applications on edge computing devices, an area that seems to be finally gaining momentum. Avant FPGAs integrate security-related hardware features, potentially making them more attractive to certain applications or hardware designs than other general purpose AI accelerators without built-in security.
Potentially even more compelling is that deep learning AI and neural networks built with popular machine learning frameworks such as Pytorch, Caffe, TensorFlow, etc. can be ported to run on these new chips using the latest version. software platform Lattice SenseAI.
The Lattice Semi software can basically import the high-level infrastructure output and convert it into the instructions required by the FPGA. This is very important because few AI algorithm creators have much experience with FPGAs – the chips used to be somewhat infamous for being very difficult to write software for them – meaning that there are plenty of new programmers and applications that could get started. use the more powerful Avant line. .
It is important to note that FPGAs are not necessarily suitable for all applications. However, they represent a potential solution for a much wider range of companies, devices and applications than they currently serve, especially in situations where flexibility and upgradeability are key.
With the debut of Avant, Lattice Semi is expanding its potential market and evolving into a more complete FPGA supplier. Established competitors like Intel and AMD certainly aren’t going to shy away from it, so the company needs to keep innovating. However, it’s clear that the future of FPGAs in general is starting to look a lot more interesting.
Bob O’Donnell – Founder and Principal Analyst Technalize Research LLC technology consulting firm that provides strategic consulting and market research services to the technology industry and the financial professional community. You can follow him on Twitter @bobodtech.