TekPlus - Development of DSPs and the DSP Market

We are entering a real-time world and the development of the DSP market will change to meet the requirements of the 'real-time' market. We need a clearer indication of how we should develop technology to meet our demands rather than moulding our demands around technology; in this way we should meet expectations rather than continuing a cycle of mismatch in technology supply and demand.

Figure 1.1 shows how technology can wreak havoc on user needs. Technology tends to outperform user requirements for sometime (computer processing power for instance) until users better understand the technology and apply it in new ways, they find new uses for it. Suddenly they find that the technology cannot cope and need to wait for the next technology step-change and the cycle begins again. However, sometimes this trend is exacerbated by other factors in the market where hype surrounds a technology, users begin to expect too much and are eventually 'turned-off' the technology, WAP is a good example. So the lessons we can learn is to ensure that we try to continue to meet user demand by delivering new technology when it is required.

In this way the development of DSP SoC should not get out of control and should limit itself to what is required by the market, rather than continually adding as much power and speed as possible, that may not be needed. The mobile devices that will take-off in the coming years will be those that support high quality audio and visual data in real-time and are easy to use and inexpensive. These are complicated objectives to meet. However, this all comes down to changing the functionality of DSP and the microprocessor. The need to develop single processor architecture is required as DSPs and processors are increasingly used in unison and this is difficult. There are a number of reasons why using DSPs and microprocessors together is complex:

Need for DSPs and microprocessors to communicate with the addition of shared memory systems, adds complexity
Different software development systems make it difficult to manage both DSP and microprocessor tasks
Synchronising DSPs and microprocessors is difficult as they have different clock-rates.

It is therefore necessary to develop real-time operating systems to control this new single processor architecture environment. Figure 1.2 shows the change in how DSPs and microprocessors are defined in a single processor environment.

Source: TekPlus

Mobility is set to become the lifestyle of choice during 2003 and 2004, as enterprises seek to mobilise employees internally and externally. Demand for portable devices is expected to increase and the variety of devices is also expected to expand, as users seek to benefit from more converged or modular devices. This trend in the market is expected to breath new life into the DSP market. The global digital signal processors market is expected to grow to $8 billion market by 2004, growing by 16 percent in 2002. The array of voice and data mobile devices we see now is expected to vastly increase, the addition of ergonomically designed and practical devices such as RIM’s Blackberry and Handspring’s Treo demonstrate the beginning of this trend. Real-time processing is crucial to the mobile world, in addition to multimedia, security and biometrics in their own right; these are all also part of mobile devices. Mobile devices will become more multimedia orientated and they will require additional security features, where biometrics has a role to fulfill. This will increase the reliance of mobile devices on the processing power of DSPs.

DSP vendors will need to meet the challenge of the mobile device market by offering complete solutions and not just the DSP itself. DSPs will continue to meet the demands of many products, but much new growth will come from new mobile devices; this requires a new breed of DSP. Of course the requirements of mobile device manufacturers will force DSP designers to create smaller more cost effective products, which will drive demand for even more integration. It can be inferred that the greater the functionality of a mobile device the greater the convergence of the silicon within the device. Further integration will continue to drive ‘system on-a-chip’ designs where DSP and microprocessor cores are moulded together. For this trend to be fully exploited there will need to be some changes in market dynamics and more co-ompertition will be required. To be able to provide system-on-a-chip (SoC) requires a great deal of IP (Intellectual Proprietary) to be at hand and this may not always be available. Large producers of integrated DSPs will need to work with DSP core licensing/designer companies in order to achieve the goal of meeting the demands of the mobile device market. If they are unable to do this it could hold back the very development of the mobile equipment and telecom services markets. What comes first the chicken or the egg? Well in this case without the egg we definitely will not see any chickens.

The IP that is developed must be re-used as much as possible, too often costs and timelines increase as a result of re-designing from scratch. In deed an increasing trend in the market will see more scalable designs entering the market. In addition, the roadmap for mobile device development should be made clearer and easier to understand. In this way mobile device vendors and DSP (SoC) vendors and designers need to work together to converge ideas and roadmaps for products. Developers need something to aim at in terms of the requirements of the user, hence meeting user demands. The market cannot continue to be purely silicon or technology driven.

The semiconductor industry has tended to focus too hard on what technology can enable and we try to create applications, features etc around silicon, together with business plans, instead of developing technology around our needs now and perceived needs in the future. Far from taking out a crystal ball and looking at the future, we all need to look at the past for a reference point for where to go next. We need to look at the past to see how technology has developed and how we have used that technology; this will give us some clues as to how we will use mobile devices in the future. Technology rarely has a massive impact on our lives as humans, for example we still work long hours when it was anticipated twenty years ago that we thought new technology would mean more leisure time; technology tends to change the way we do things, speeds things up and give us more freedom or choice. The mobile phone means that we can do more things on the fly, we have a portable communications device that enables us to let people know we are late, whereas we would have used a payphone previously, we tell people 'give me a ring at 8 when you get into town and we’ll meet somewhere', whereas previously we would have arranged to meet at a specific place and text messaging has replaced pieces of paper being passed around the playground or a quiet word behind the bike sheds. New technologies add flexibility but do not eradicate what has gone before, well not immediately. TV never killed the radio and the Internet will not kill books or newspapers, not in our lifetime anyway.

Figure 1.1 Mismatches of Technology Development and User Demand

VIEW POINTS

Ref: TP0044A02V01

Development of DSPs and the DSP Market