By offering the ability to update application functionality, provide a reconfigurable solution and enable easy design customization, Field Programmable Gate Array (FPGA) technology has long been known as a cost-effective design resource. Similarly, x86 processor architectures share many of the same extended ecosystem, installed base, and multi-functionality benefits as FPGAs.
The availability of new x86 processors in combination with an FPGA presents an entirely new design approach thanks to increased performance integrated into smaller form factors that offer very low thermal design power (TDP). The result for embedded applications is that a predetermined feature set for embedded computers practically ceases to exist. Leveraging the inherent advantages of FPGA-controlled I/Os and x86 processor-based boards into new hardware platforms presents an innovative and flexible solution that allows designers to develop dedicated I/Os that support a wider range of application requirements. This single hardware platform also enables OEMs to shrink the overall bill of materials by eliminating chipsets. Designers can reconfigure the FPGA on the same hardware platform with just the exchange of IP cores to suit different protocol requirements and are able to develop different areas of applications without the need for a full board redesign. This hardware-optimized approach provides the ability to do customization on a common platform increasing design differentiation and also reducing time-to-market.
The x86 Architecture Evolution
The continuing evolution of the x86 processor architectures with their high-performance, extended and enhanced features, and lower power consumption ushers in a new realm of platform flexibility that supports a broader variety of design options. This evolution has been a boon for embedded systems, but is not without its pitfalls. The large number of x86-based developers that have created the huge installed base of applications has led to a very long list of pre-determined instruction sets that now limit many types of embedded applications. For this reason, embedded designers have been looking for a better way to meet the specific I/O requirements or have the ability to customize embedded solutions with proprietary I/O or acceleration.
By combining the CPU core with an FPGA, designers have access to pure IP they can use to increase design flexibility and streamline the design process for new applications. Compared to the unwelcome process of managing several different controller components, an added benefit is that only the IP cores need to be maintained. Integrating the latest x86 processors with FPGAs ensures that embedded system OEMs have access to the latest high-performance, open platforms enabling them to utilize dedicated I/Os for proprietary interfaces and other functionality configured to an application’s needs. It is possible to execute the complete chipset functionality directly by an FPGA. Even more importantly, this solution provides design longevity with interface support that is available for as long as it is required.
All the design gains realized from new combined CPU core and FPGA-based platforms meet the demands of today’s more complex systems that have unique I/O requirements that must interface with a broad range of diverse devices. This approach is also well-suited for applications that require compute-intensive calculations. Using a high-performance FPGA-based computing platform gives designers the option of relieving the processor function from the controller and still implement system functionality, all achieved in hardware.
Application Migration and Legacy Support
Adding new features, system upgrades, or migrating from legacy designs typically requires that designers modify or develop a new board with each new application release. With FPGAs, software and IP cores now take on a more predominant role in embedded computing at the hardware level. So, rather than using a chipset with pre-configured I/O support, designers can utilize IP to customize I/O with software in a single board solution. This approach makes it easier to migrate from legacy designs. Designing application upgrades is also streamlined with these new platforms by removing the need to add additional hardware to the application design in order to manage new interface functionality. For example, many applications are migrating to Ethernet-based networks, which has forced designers to use additional hardware devices to supply the mandated networking capabilities.
Legacy designs that are supported by older interfaces such as ISA, RS232 and CAN frequently run into problems because these interfaces are not supported by newer generation chipsets or dedicated hardware-designed I/O add-on cards. The ability to sustain legacy designs is another reason that FPGAs have long been valued as a design tool. Looking to the future, it is foreseeable that PCI will become obsolete and not be supported by standard chipsets. It is becoming more of an issue that current processor generations provide only PCI Express support, which means designers must find other solutions such as PCI switches. Unfortunately, these switches have to steal resources that may be needed elsewhere for the application, making for a less than ideal design. New application designs will always benefit from continuing advancements in technology; however, there is a 20-year installed base of PCI-based applications that are still actively deployed. For this installed base, it would be entirely excessive to migrate to next-generation PCI Express or Gigabit Ethernet for applications that only call for 32-bit/66 MHz performance. The optimal solution in such cases is to use an FPGA instead of a chipset or other switch to execute the PCI interface for the needed I/O.
Support for Proprietary Applications
In the example cited above, the industrial automation market has had to take into consideration the number of industrial Ethernet technologies currently in use, and many of these applications are proprietary. It has become an overwhelming task to develop different fieldbus implementations for all these proprietary protocols. Adding to this issue for proprietary industrial automation applications is how panel PCs and HMIs are supported in these countless different installations. Industrial automation designers have been calling for a common platform that can accommodate a CAN terminal, a PROFIBUS terminal, a LonWorks terminal, or any other fieldbus or industrial Ethernet protocol that is required for their specific application. Using a platform solution that only requires the IP cores to be exchanged demonstrates the value of a flexible approach that allows a wide variety of devices to easily speak with each other.
Many systems for the medical market face the same issues where each piece of diagnostic and patient monitoring equipment has been designed with unique I/O that has special interface requirements. Designers have found it challenging to upgrade these systems with network capabilities due to the huge amount of additional configuration, programming, and hardware that would be required. An advanced CPU core and FPGA-based single board computer is often the answer.
Next-Generation Platform Solutions
Embedded computing platforms have just been introduced based on this combination approach. New boards are now available based on the second generation Intel® Core™ i3/i5/i7 processors and the Intel® Atom™ E6x5C processor series paired with an integrated Altera FPGA into a single package. The new x86-based processor architectures offer extremely high performance, leading-edge graphics and energy efficiency in a small footprint. It also features the integrated, hardware-assisted Intel® Virtualization technology that helps consolidate multiple environments into a single hardware platform.