SOFTWARE OPTIMIZATION TECHNIQUES - Software Design for Six-Sigma: A Roadmap for Excellence - page 448

Moreover, embedded systems are used in smaller and more portable applications,
making memory space smaller and at a premium. As a result, memory is still an
issue.
One way to classify memory is through the term coined volatility. Volatile mem-
ories only hold their contents while power is applied to the memory device, and as
power is removed, the memories lose their contents. 19 Volatile memories are unac-
ceptable if data must be retained when the memory is switched off. Some examples of
volatile memories include static random access memory (SRAM), and synchronous
dynamic random access memory (SDRAM), which are discussed in greater detail
subsequently.
In contrast, nonvolatile memories retain their contents when power is switched off.
Items such as CPU boot-code typically are stored in nonvolatile memory. Although
nonvolatile memory has the advantage of retaining its data when power is removed, it
is typically much slower to write to than volatile memory and often has more complex
writing and erasing procedures. Moreover, nonvolatile memory is also usually only
erasable for a given number of times. Some types of nonvolatile memories include
flash memory, erasable programmable read only memory (EPROM), and electrically
erasable programmable read only memory (EEPROM), which also are discussed in
greater detail subsequently. Most types of embedded systems available today use
some type of flash memory for nonvolatile storage. Many embedded applications
require both volatile and nonvolatile memories because the two memory types serve
unique and exclusive purposes.
The main types of memory are random access memory (RAM), read only memory
(ROM), and a hybrid of the two different types. The RAM family includes two
important memory devices: static RAM (SRAM) and dynamic RAM (DRAM). 20
SRAM is retained as long as electrical power is applied to the chip, and DRAM has a
short data lifetime of a few milliseconds. When deciding which type of RAM to use,
a system designer must consider access time and cost. SRAM offers fast access times
but are much more expensive to produce. DRAM can be used when large amounts of
RAM are required. Most types of embedded systems include both types of memory
in which a small block of SRAM and a large block of DRAM is used for everything
else.
ROM memory can have new data written. Some types of ROM rewritten, reflect
the evolution of ROM devices from hardwired to programmable to erasable and
programmable. However, all ROM devices are capable of retaining data and programs
forever. The first ROMs contained a preprogrammed set of data or instructions in
which the contents of the ROM had to be specified before chip production. Hardwired
memories still can be used, and are called masked ROM. The primary advantage
of a masked ROM is its low production cost. PROM (programmable ROM or a
one-time programmable device) is purchased in an unprogrammed state. A device
programmer writes data to the PROM one word at a time by applying an electrical
charge to the input pins of the chip. Once a PROM has been programmed in this way,
19 http://www.altera.com/literature/hb/nios2/edh ed51008.pdf
20 http://www.netrino.com/Embedded-Systems/How-To/Memory-Types-RAM-ROM-Flash