Page 71 - A Practical Guide from Design Planning to Manufacturing

Computer Components 47

One potential problem is that the other bus clocks on the motherboard
are typically derived from the processor bus frequency. This means
increasing the processor bus frequency can increase the frequency of
not only the processor but of all the other components as well. The
frequency limiter could easily be some component besides the processor.
Some motherboards have the capability of adjusting the ratios between
the various bus clocks to allow the other buses to stay near their nominal
frequency as the processor bus is overclocked.
Processor overclocking is no more illegal than working on your own
car, and there are plenty of amateur auto mechanics who have been able
to improve the performance of their car by making a few modifications.
However, it is important to remember that overclocking will invalidate
a processor’s warranty. If a personally installed custom muffler system
causes a car to break down, it’s unlikely the dealer who sold the car
would agree to fix it.
Overclocking reduces the lifetime of the processor. Like driving a car
with the RPM in the red zone all the time, overclocked processors are
under more strain than the manufacturer deemed safe and they will
tend to wear out sooner. Of course, most people replace their computers
long before the components are worn out anyway, and the promise and
maybe more importantly the challenge of getting the most out of their
computer will continue to make overclocking a rewarding hobby for some.

Main Memory
The main memory store of computers today is always based on a partic-
ular type of memory circuit, Dynamic Random Access Memory (DRAM).
Because this has been true since the late 1970s, the terms main memory
and DRAM have become effectively interchangeable. DRAM chips provide
efficient storage because they use only one transistor to store each bit of
information.
The transistor controls access to a capacitor that is used to hold an
electric charge. To write a bit of information, the transistor is turned on
and charge is either added to or drained from the capacitor. To read, the
transistor is turned on again and the charge on the capacitor is detected
as a change in voltage on the output of the transistor. A gigabit DRAM
chip has a billion transistors and capacitors storing information.
Over time the DRAM manufacturing process has focused on creating
capacitors that will store more charge while taking up less die area. This
had led to creating capacitors by etching deep trenches into the surface
of the silicon, allowing a large capacitor to take up very little area at
the surface of the die. Unfortunately the capacitors are not perfect. Charge
tends to leak out over time, and all data would be lost in less than a
second. This is why DRAM is called a dynamic memory; the charge in
all the capacitors must be refreshed about every 15 ms.