Home Articles Walking the razor's edge: Balance between power and performance...

Walking the Razor's Edge: Balancing Power and Performance in Embedded Systems

power and performance

Miniaturization is a key feature in most embedded systems today. We want more computing power in our pockets. Most of the FPGA based embedded systems are also following the same trend. We want smaller industrial and professional cameras, portable medical devices, smaller PLCs, and driver assistance modules in cars. Miniaturization also poses additional challenges, the biggest of which can be condensed into a single term: "energy efficient performance." Typically, if a system's performance increases, so does its power consumption, which in turn increases heat dissipation. And in smaller modules, heat dissipation is a headache designers face every day. Cooling a module so that it can function in a thermally constrained environment often becomes the performance bottleneck.

This article highlights how FPGAs are enabling the next-generation technology revolution by delivering power-efficient performance in many new high-volume applications in all walks of life.

Small format cameras running AI algorithms are guiding farmers through drone footage, providing video analytics at chain stores, counting passengers on transit, and reading license plates at toll booths.

In the medical arena, portable ultrasound machines are democratizing care delivery in the field. Endoscopes and surgical assistive smart glasses are providing clinicians with much higher resolution images than ever before. Surveillance systems based on thermal imaging and designed to protect borders against intruders are also getting smarter. Usually deployed in remote locations, these systems must work autonomously and remain hidden.

Amateur content creators are all the rage today, driving the need for FPGA-based streaming video converters, giving creators the option to convert 4K video streams between any format such as HDMI, SDI, USB or PCIe.

Industrial automation also benefits from the flexibility of FPGA-based architectures and the longevity of Microchip's FPGAs over 20 years. Currently, automotive driver assistance systems maintain the safety of the driver and passengers in our cars.

FPGA architecture has come a long way. From having to choose between performance and power, and being used solely as a prototyping platform for expensive ASICs, FPGAs are now considered mainstream, offering highly reliable and cost-optimized architectures, along with flexible, easy-to-use software. .

Let's look at several examples of use cases where PolarFire® FPGAs or PolarFire SoCs, with their hardened RISC-V processor system, play a critical role.

Professional drones

Professional drones have strict flight safety requirements:

  • Precise control and positioning, including collision avoidance
  • Secure communication and control frequencies
  • predictable flight time

To succeed in such a vast drone market, drone manufacturers have to differentiate themselves by providing additional features such as high-resolution images and artificial intelligence. Drones often require multiple sensors, pre-processing or fusion of sensor data, and transfer of that data over a wireless connection, making them complex systems.

The range of applications is very wide and includes monitoring the health and growth of crops in agriculture, the detection and possible tracking of objects in the police and military field, or remote evaluation in emergency situations for firefighters. or the police.

Flight control electronics must be able to manage engine control and rotor speed, interface with sensors, and interface with remote equipment, all in an environment of limited size, weight, and power.

The block diagram of such a system may look like the following:

blocks diagram
Figure 1.

Taking advantage of the flexible architecture of the FPGA, the motors are controlled by field-oriented control (FOC) algorithms, whose control can be multiplexed in the time domain thanks to the performance of the FPGA. A common motor control IP controls multiple motors, the exact number of which depends on the chosen FPGA architecture.

The high precision of the FOC allows for constant torque on the motors, resulting in smoother operation with less vibration, less noise generation, and most importantly, extended flight time by approximately 10 percent or more compared to standard motor controllers using simple microcontrollers.

Additional interfaces, such as visual light, motion, or infrared sensors, that are used to support enhanced functions such as computer vision require careful consideration, and have historically required specialized knowledge. Microchip's VectorBlox™ SDK and IP Array Processor help novice FPGA developers deploy complex neural network algorithms on the FPGA fabric. This allows classification or detection with a very low power footprint. The neural networks running on this IP-accelerator are designed using standard frameworks such as TensorFlow or Caffe.

All results are stored in the board's local memory and then transferred to a wireless module on the board. This communicates with the operator, where the collected data is received for storage and later use. The best security features of PolarFire devices protect both the transferred data and the drone itself from unauthorized access.

With a complex drone architecture that requires multiple application domains – motor control, flight control, and imaging – the use of an FPGA provides the advantage of having individual “tasks” run in parallel.

Professional drone systems often have to run on a very tight power budget of 5 watts or less. Using a PolarFire FPGA to drive multiple applications, a power consumption of less than 1,5 W is expected for the FPGA, including neural network operation.

Portable Ultrasounds

Due to the trend towards miniaturization, coupled with low-power edge computing resources and improved thermal considerations, innovation in low-power medical imaging is growing by leaps and bounds. Leading the way are point-of-care diagnostics, such as portable ultrasound machines, which consist of a handheld transducer that reads and sends ultrasound data to a standard smartphone. Transmissions can be done with a simple cable or wirelessly. These systems are revolutionizing and democratizing the diagnostic capabilities of EMS personnel at accident scenes, in less developed regions, and helping medical professionals make diagnostic decisions outside of traditional hospital settings.

The following block diagram shows an example implementation:

implementation block diagram
Figure 2.

Using a PolarFire FPGA in a portable medical device offers the lowest total system power, allowing for efficient thermal control and keeping the transducer head cool, allowing direct skin contact. These efficiencies extend runtime in a compact package of just 11x11mm² that accommodates very small probe enclosures.

Video Converters

Another area where flexibility, along with low power consumption and a small physical footprint, is essential is in the realm of video converters. High-performance professional cameras often offer a single data interface, limiting the selection of post-processing equipment that supports that specific interface. The fact that a video converter offers a bridge to various interface standards allows flexibility in selecting post-processing equipment. Performance is not affected as multiple protocols are supported with many multi-gigabit transceivers and optimized line speeds of up to 12,7 Gbps, supporting HDMI, CoaXPress, SDI and Ethernet protocols. The form factors of the converters are compact, since no heatsinks or fans are required. Video converters built with PolarFire technology are estimated to require less than two watts of power consumption.

Here is an example video converter layout:

polarfire mpf
Figure 3.

Industrial automation

Two different use cases are used as an example, industrial cameras and PLCs.

Industrial cameras often require high frame rates, high resolution, and a small form factor, making thermal design often challenging. Thanks to the optimization of the package design and the efficient thermal characteristics, this challenge can be easily addressed. Low static power consumption allows the device to stay cool, improving thermal management design considerations. Resolution is not compromised, image data up to 4K with 60 frames/second can be easily handled with MIPI CSI-2 receiver interfaces that natively support up to 1,5 Gbps/line.

Despite being physically larger as a complete system, PLCs have the same space and power limitations as cameras.

These rack-based systems are modular, allowing end users to customize their system and offer standard chassis widths. Throughput is still a necessity to support Industrial Ethernet, Human Machine Interfaces, motor control/drivers, and Real Time Operating Systems (RTOS).

The graphic below shows a generic block diagram of such a system, mapped onto the PolarFire SoC, the first FPGA-SoC built on a quad-core RISC-V processor. The PolarFire SoC supports asymmetric multiprocessing (AMP) natively, along with a fixed and granular allocation of cache lanes to individual processors. This native AMP support enables multitasking. For example, a single processor core can be allocated for an Industrial Ethernet protocol stack, while a second core can run a Linux operating system. The corresponding cache is fixed, and Linux is separated from other hardware resources. In addition, the other two available cores can be used to manage the algorithms needed to control the motor or an inverter.

polarfire fpga
Figure 4.

Once again, low power consumption plays an important role in keeping the temperature of the electronics inside the blade modules low, even in harsh thermal environments of 60°C ambient and above.

To learn more about why low power consumption is important in cable-powered systems, see the following article, hosted on Microchip Blogs:

https://www.microchip.com/en-us/about/blog/learning-center/low-power-system-saving-even-in-plug-in-devices

Industrial automation covers a wide range of applications and requirements. Among industrial products, the need to offer support and availability of devices for 20 years or more is common. Microchip is fully dedicated to this longevity requirement and supports it with a robust "Supply Assurance" program.

Automobile

Many different applications in today's automotive market require the flexibility of FPGAs, from sensors like lidar, imaging radars, or cameras, to more hidden functions like tightly synchronized, highly precise driving of electric motors via high-voltage controllers. . One application that is emerging strongly is the use of cameras to warn of collisions. These cameras allow the detection of dangerous situations with feedback to the driver or also with direct control in the vehicle such as automatic activation of the brakes.

These systems have strong requirements for functional safety, security, and low-latency processing combined with the ability to operate reliably in high-temperature environments caused by engine heat and sunlight.

The following diagram shows the configuration of a system using the PolarFire MPF050T, security elements are drawn in yellow, security in green:

Onboard secure non-volatile memory (sNVM) enables storage of fleet keys for authentication within the camera module within the vehicle network. Received image frames are processed in streaming mode using the parallel nature of the FPGA and additionally provided with additional security information such as frame count and CRC for end-to-end protection of the communication. In-stream processing of image data avoids the danger of using "frozen images" of memory and allows processing with a fixed execution time, which translates directly into more reaction time for the system. Depending on the exact requirements of the OEM, the FPGA also provides the flexibility to support interfacing with various established proprietary serializers.

polarfire sensor
Figure 5.

Factors common to all applications, although not detailed above, are the business drivers for bringing a successful product to market. To reduce risk, reach the customer before the competition, and optimize system cost while reaping benefits, you must carefully consider your system architecture and delivery partner. Microchip's broad product portfolio offers a total system solution partnership. Take advantage of key components and reference design solutions to reduce development risk and component count. Designers can also save time and money, as solutions are validated for cross-functionality and offer warranties in many cases.

For more, please visit https://www.microchip.com/en-us/products/fpgas-and-plds/fpgas/polarfire-fpgas