DSP: What’s all the excitement about?

DSP or Digital Signal Processing by itself is quite mind boggling in terms of
technology and range of applications. Devices that will deliver the power of
today's laptop in a device the size of a wristwatch may not be very far off, if
increasing chip performance is anything to go by. In the near future, there
could be a single chip composed of dozens of digital signal processor (DSP)
chips, each with 500 million transistors integrated into it. This level of DSP
performance, combined with reduced power consumption and space requirements, is
vital for applications such as DSP-based residential gateways, Internet TV,
wearable health monitoring devices, autonomous home robots and real-time
videophones

DSP explained

To start at the very beginning, what is DSP? As the name itself indicates,
deals with the various techniques employed in processing digital signals. In
most cases, the initial signal is in the form of an analog electrical voltage or
current which is converted into digital (i.e. numerical) form before DSP
techniques can be applied. An analog signal can be digitized using an
analog-to-digital converter or ADC. This generates a digital output in the form
of a binary number whose value represents the analog input to the device. In an
increasingly digital world, analog signals need to be first converted to digital
signals to take advantage of the many benefits that digital signals provide.
Digital signals allow for easy manipulation and elaborate modification of audio
and video signals, which provides for greater flexibility. These capabilities
are exactly what DSPs exploit. Once a signal has been reduced to numbers, its
components can be isolated, analyzed and rearranged more easily than in analog
form.

The term DSP can be used for the techniques used to process digital signals
or the digital signal processor that is used to perform the number crunching
operations involved in digital signal processing. The processor has specialized
set of instructions, architecture, bus structure and addressing modes that set
it apart from its more well known counterpart, the general purpose
microprocessor. Another key difference between the two is that DSPs have a
dedicated multiplier accumulator function called MAC. The MAC takes care of the
heavy number crunching operations and the memory architecture in turn is
optimized to support this. Another basic difference is that apart from the
general addressing modes, DSPs have bit reverse addressing which is specifically
targeted for Fast Fourier Transforms. Fast Fourier Transforms allow the
frequency spectrum of a signal to be computed rapidly.

When we talk of speed in chips, the first thing that comes to mind is clock
speed but in the case of DSP, the term clock speed can be misleading. Processing
capabilities can be better gauged when you take into account that they are
capable of performing about 10 billion operations per second. This need for
speed is powered by the requirements to provide real-time performance: that is,
the ability to process a signal "live" as it is sampled and then
output the processed signal.

DSP chips are capable of carrying out millions of floating point operations
per second and for most manufacturers it’s a continuous race to manufacture
faster and more powerful ones.