Post Syndicated from Digilent original https://spectrum.ieee.org/computing/networks/breaking-down-barriers-in-fpga-engineering-speeds-up-development
It’s hard to reinvent the wheel—they’re round and they spin. But you can make them more efficient, faster, and easier for anyone to use. This is essentially what Digilent Inc. has done with its new Eclypse Z7 Field-Programmable Gate Array (FPGA) board. The Eclypse Z7 represents the first host board of Digilent’s new Eclyspe platform that aims at increasing productivity and accelerating FPGA system design.
To accomplish this, Digilent has taken the design and development of FPGAs out of the silo restricted to highly specialized digital design engineers or embedded systems engineers and opened it up to a much broader group of people that have knowledge of common programming languages, like C and C++. Additional languages like Python and LabVIEW are expected to be supported in future updates.
FPGAs have been a key tool for engineers to tailor a circuit board exactly the way it is needed to be for a particular application. To program these FPGAs specialized development tools are needed. Typically, the tool chain used for Xilinx FPGAs is a programming environment known as Vivado, provided by Xilinx, one of the original developers of FPGAs.
“FPGA development environments like Vivado really require a very niche understanding and knowledge,” said Steve Johnson, president of Digilent. “As a result, they are relegated to a pretty small cadre of engineers.”
Johnson added, “Our intent with the Eclypse Z7 is to empower a much larger number of engineers and even scientists so that they can harness the power of these FPGAs and Systems on a Chip (SoCs), which typically would be out of their reach. We want to broaden the customer base and empower a much larger group of people.”
Digilent didn’t just target relatively easy SoC chip devices. Instead, the company jumped into the deep end of the FPGA pool and focused on the development of a Zynq 7020 FPGA SoC from Xilinx, which has a fairly complex combination of a dual-core ARM processor with an FPGA fabric. This complex part presents even more of challenge for most engineers.
To overcome this complexity, Johnson explains that they essentially abstracted the complexity out of the system level development of Input/Output (I/O) modules by incorporating a software layer and FPGA “blocks” that serve as a kind of driver.
“You can almost think of it as when you plug a printer into a computer, you don’t need to know all of the details of how that printer works,” explained Johnson. “We’re essentially providing a low-level driver for each of these I/O modules so that someone can just plug it in.”
With this capability, a user can configure an I/O device that they just plugged in and start acquiring data from it, according to Johnson. Typically, this would require weeks of work involving the pouring over of data sheets and understanding the registers of the devices that you’ve plugged in. You would need to learn how to communicate with that device at a very low-level so that it was properly configured to move data back and forth. With the new Eclypse Z7 all of that trouble has been taken off the table.
Beyond the software element of the new platform, there’s a focus on high-speed analog and digital I/O. This focus is partly due to Digilent’s alignment with its parent company—National Instruments—and its focus around automated measurements. This high-speed analog and digital I/O is expected to be a key feature for applications where FPGAs and SoCs are really powerful: Edge Computing.
In these Edge Computing environments, such as in predictive maintenance, you need analog inputs to be able to do vibration or signal monitoring applications. In these types of applications you need high-speed analog inputs and outputs and a lot of processing power near the sensor.
The capabilities of these FPGA and SoC devices in Edge Computing could lead to applying machine learning or artificial intelligence to these devices, ushering in a convergence between two important trends – Artificial Intelligence (AI) and the Internet of Things (IoT) that’s coming to be known as the Artificial Intelligence of Things (AIoT), according to Johnson.
Currently, the FPGA and SoC platforms used in these devices can take advantage of 4G networks to enable Edge devices like those envisioned in AIoT scenarios. But this capability will be greatly enhanced when 5G networks are mature. At that time, Johnson envisions you’ll just have a 5G module that you can plug into a USB or miniPCIe port on an Edge device.
“These SoCs—these ARM processors with the FPGAs attached to them—are exactly the right kind of architecture to do this low-power, small form factor, Edge Computing,” said Johnson. “The analog input that we’re focusing on is intended to both sense the real world and then process and deliver that information. So they’re meant exactly for that kind of application.”
This move by Digilent to empower a greater spectrum of engineers and scientists is in line with their overall aim of helping customers create, prototype and develop small, embedded systems—whether they are medical devices or edge computing devices.