ASIC or FPGA? Selecting the right IC for your application

By Ross Turnbull, Director of Business Development and Product Engineering at Swindon Silicon Systems

To grasp the differences between an ASIC and an FPGA, let's first define these components. An ASIC is a specialised integrated circuit meticulously designed for a specific application. In contrast, an FPGA is a versatile, standard part that users can configure to perform a variety of functions. FPGAs offer the flexibility of reprogramming multiple times to suit different tasks.

While the basic distinction seems straightforward – ASICs are optimised for particular tasks, and FPGAs offer adaptability – the decision between the two for your application involves more intricate considerations. What factors should influence your choice?

What are your performance goals?

First, consider what you want your chip to do. The design of ASICs enables them to achieve superior performance characteristics. Since they are crafted for a single purpose, engineers can eliminate unnecessary features and streamline the design to meet precise requirements, making them particularly well-suited for high-volume products where performance is critical.

Power efficiency is another significant advantage of ASICs, particularly where battery-powered applications are considered. ASICs also excel in area efficiency, meaning they can perform their functions using less silicon real estate. This is particularly important in applications where space is at a premium, such as in automotive systems where multiple functionalities need to be integrated into a compact space.

The tailored design of an ASIC ensures that it can handle the specific demands of its intended use case with superior efficiency compared to a general-purpose FPGA.

What are your project goals?

It’s important to not just think about the end result, but also to consider how you’d like the project to play out during the design and manufacture process. Depending on how much involvement you’d like to have, and how quickly you wish to get to market, either an ASIC or an FPGA could have an edge over the other.

If time-to-market is of high importance, an FGPA may be the preferred route. Due to their specialist nature, ASIC projects involve a rigorous design and testing process. FPGAs, with their standard interface, offer the advantage of being easily reprogrammable to perform specific functions, facilitating quick and seamless product integration. This flexibility makes them particularly well-suited to dynamic industries where the environment and expectations are rapidly evolving, and adaptability is prioritised over cost considerations.

But it is possible to enjoy the optimised performance of an ASIC without an extensive timeframe. ASIC designers have access to libraries of foundry and third-party IP. These IP blocks can be used to form the basis of a custom IC design, shortening development time without losing the optimisation benefits of an ASIC.

When thinking of the investment you’re making with an IC solution, budget will inevitably come into question. The cost factor largely depends on the scale of production required. ASICs typically have a higher non-recurring engineering (NRE) cost than FPGAs. They will, however, have a lower cost per unit due to the removal of any functionalities that are not required for that application. As a result, while the initial cost and time investment for ASIC design is higher, it typically offers a much better ROI for larger or longer production volumes.

What are your long-term goals?

Don’t just think about the solution you want right now – it’s also important to consider its longevity. Navigating obsolescence can be challenging and is often a significant concern for purchasing and procurement professionals, as well as the engineers seeking the IC solution.

FPGA supply, like many other off-the-shelf solutions, is determined by the manufacturer. If supply falls below a certain threshold, the manufacturer may cease production of new FPGA chips and issue a notification of obsolescence. At this point, it becomes the customer's responsibility to find an alternative, which could involve purchasing remaining stock or sourcing a similar product from another supplier.

ASICs operate differently. Instead of the manufacturer halting production, the equivalent issue in ASICs is the silicon process becoming obsolete. When designing a custom IC, the supplier ensures that the chosen silicon process will last for the product's expected lifetime. However, in the rare event that obsolescence occurs sooner than anticipated, a reputable ASIC company will collaborate with the customer to find an alternative solution. This may include porting the design to a newer silicon process or purchasing and storing wafers for several decades in dry nitrogen cupboards.

At first glance, the differences between an ASIC and an FPGA may appear straightforward. However, determining which is best suited for your specific application requires a more nuanced approach. Factors such as cost, time-to-market, obsolescence and other critical considerations must be carefully evaluated to ensure you select a chip that meets all your product's requirements, both for the present and future iterations.

This article originally appeared in the August'24 magazine issue of Electronic Specifier Design – see ES's Magazine Archives for more featured publications.