During the past few years, I've noticed an interesting trend. When a vendor introduces a faster processor, I am flooded with e-mails, phone calls and mailings trumpeting its arrival. In contrast, I may hear nothing from the vendor when the new processor excels in other dimensions, such as low cost. It's almost as if the vendors are ashamed of these not-so-fast processors.

This emphasis on faster processors suggests that blazing speed is the top requirement for signal-processing applications. In reality, most signal-processing applications don't demand maximum processing speed. Rather, they have certain fixed processing requirements. The goal is to meet those requirements while minimizing cost, energy consumption, development effort and risk. This is particularly true of high-volume applications. A portable MP3 player, for example, only needs enough processing power to decode MP3 files in real-time. There's no point in decoding the files at a faster rate — the user can't listen any faster — but there is a strong need to minimize cost, maximize battery life and get the product to market quickly.

This is not to say that signal-processing applications never need more speed. Quite the contrary — processors that target today's communications infrastructure applications must strain for speed. But even in applications that demand top speed, other factors matter. For example, energy efficiency is critical for communications infrastructure equipment, where the cramped quarters make it difficult to dissipate heat.

So why is there so much focus on speed? One possible reason is that processor speed has been a key selling point in personal computer markets for decades. Even nontechnical consumers know the clock speeds of the CPUs in their PCs.

Another consideration is that speed is often easier to measure than other metrics. For example, suppose a processor's main competitive advantage is its on-chip integration of memory, peripherals and other resources. This integration clearly benefits system cost, size and energy efficiency, but quantifying those benefits requires a detailed, highly application-dependent analysis.

In some ways, a fast processor is like a fast car: The speed can be impressive, but purchasing decisions are usually driven by other factors. To make good decisions, designers need access to solid information on other aspects of performance besides speed. And vendors can do their part by expanding their single-minded emphasis on speed to include other important metrics.

Jeff Bier is the general manager of Berkeley Design Technology Inc. (www.BDTI.com), the DSP technology analysis and software development company. Kenton Williston of BDTI contributed to this column.