Clocks are at the heart of synchronous digital systems and processors. CPUs require clocks to run to ensure the orderly fetching and execution of instructions. In PCs, the selection of clock speeds is determined by various factors. Unless you are overclocking, you will never deal with them directly. Microcontrollers, on the other hand, are usually very flexible with clocks and require that you to specify what clocks will be used and at what speeds. This means you have to understand your application and what clocks are needed to satisfy your requirements.
One of the most obvious effects of clocks is the speed at which the CPU is processing data. If the clock is slow, algorithms will take longer to complete and the application can violate timing requirements. Another good example is the timing required for UART. UART is covered in a later chapter, but being asynchronous means that the reliability of the data being transmitted depends on the input clock, as is the speed at which it transmits data. If the clock is unstable and the baudrate is inaccurate, errors will be detected at the receiver.
The MSP430 clock system is very flexible and is built both to enable a fast CPU while at the same time supporting slow clocks that allow low power. With low power being the cornerstone of the MSP430, reducing the frequency of the CPU is critical to reducing power consumption. The dynamic power consumption can be shown to be: 1/2 C*V^2*f where V is the voltage, C the switching capacitance and f is the frequency of the circuit. Reducing the voltage and frequency can provide tremendous reductions in power consumption which makes products last longer on batteries.