Recent FPGAs Designed for C++ and Python, Harsh Environments, and ASIC PrototypingNovember 23, 2019 by Robin Mitchell
These recently-released FPGAs aim to use accessible programming languages, withstand harsh environments, and support ASIC prototyping.
FPGAs (Field Programmable Gate Arrays) are increasingly becoming a critical component in modern electronics because of their ability to handle mixed-signals, customize IP cores, provide a wide range of I/O, and enable fast operating speeds.
In this round-up, we will look at three recently released FPGAs that could be useful for your next project.
Xilinx's Open-source FPGA Platform
While FPGAs offer developers a configurable platform, they are often difficult to program—especially for those who are new to programming.
This is because FPGAs typically use hardware descriptive languages such as VHDL and Verilog, which are not as intuitive as languages like C++.
To address this issue, Xilinx has announced Vitis, a unified open-source software platform.
Xilinx claims that the Vitis platform allows users to create a range of applications—including IoT, video processing, and neural networks—using common languages, such as C++ and Python.
Xilinx considers Vitis a unified open-source software platform. Image from Xilinx
The Vitis platform offers tools for debugging and bottlenecking designs. It can also target platforms including Xilinx's recently-released accelerator cards, Zynq SoCs, MPScOS, and Versla ACAPs.
With more than 400 open-source applications, the Vitis AI component can support many modern AI neural network technologies, including TensorFlow, Caffe, and PyTorch. Designers who wish to continue using hardware code can do so with the Vivado Design Suite.
However, the goal of Vitis is to help provide an abstraction layer between the hardware and application by creating software callable functions.
Microchip's Low-Power Radiation-Tolerance PolarFire
Although microcontrollers and SoCs provide functionality for most products on the market, some applications require specialized hardware and circuitry; this is where FPGAs become important.
While many specialized applications are located in stable environments, others require operation in some of the harshest conditions (particularly, in aerospace and satellite development), which are subjected to intense amounts of radiation.
In these applications, a standard FPGA will not be reliable enough since incoming radiation can alter configuration data and damage sensitive interconnects and transistor gates on the silicon.
For these reasons, Microchip has announced its latest radiation-tolerant range of PolarFire FPGAs, which are designed to help space designs cope with strenuous launch vehicles and the pressures of space.
The radiation-hardened PolarFire FPGAs are built to consume as little power as possible while offering high-speed data paths.
Microchip's RT PolarFire is designed for harsh environments. Image from Microchip
Microchip claims that in replacing commonly used ASIC-based designs, the RT PolarFire range reduces cost and development time.
Diagram of RT PolarFire ecosystem. Image from Microchip
The RT PolarFire range, geared for survivability, includes multiple mitigation techniques, like protection from single event upsets (SEU), single event latch-ups (SEL), and configuration upsets.
The radiation shielding of the RT PolarFire range is said to handle exposure beyond 100 kilorads that most earth-orbiting satellites and deep space missions experience.
Intel Stratix 10GX 10M FPGA
The previous examples have shown how FPGAs are becoming more accessible and useful for designers, especially those working with devices for harsh environments. But some designs call for shear logic count and processing power; this is where the Intel Stratix 10GX steps in.
Intel claims that its new FPGA, the Stratix 10GX, is the world's highest-capacity FPGA to date with 10.2 million logic elements.
Stratix 10GX is built on Intel's Embedded Multi-die Interconnect Bridge (EMIB) technology. Intel considers its EMIB technology a means to create complex silicon solutions that require multiple silicon dies in the same integrated circuit package.
The Stratix 10GX is made up of two dies that each has 5.1 million elements. These two dies are connected by electrical and logic connections so they work as one single unit.
While this FPGA may seem excessive for many applications, there is one market that may benefit from it: ASIC prototyping.
Intel Stratix 10GX. Image from Intel
When it comes to creating custom solutions, ASICs can offer designers the freedom of various circuit arrangements; but unless ASICs are purchased in large quantities, they can be incredibly expensive.
Small ASIC designs can be easily simulated before fabrication, but larger, more complex designs, may require prototyping.
For example, an ASIC designed to be used in a PCIe bus may need to be fully prototyped to ensure the correct function, and this is something that many FPGAs may struggle to support. The Intel Stratix 10GX 10M is aimed to help ASIC developers to prototype ever-larger ASICs.
Which FPGAs are useful for your designs? What specs do you value most in those FPGAs? Share your thoughts in the comments below.