Page 98 - A Practical Guide from Design Planning to Manufacturing

74 Chapter Three

after sales begin. Changes are made to improve performance or reduce
the number of defects. The debugging of initial prototypes and movement
into volume production is called the silicon ramp.
Throughout the design flow, validation works to make sure each
step is performed correctly and is compatible with the steps before and
after. For a large from scratch processor design, the entire design flow
might take between 3 and 5 years using anywhere from 200 to 1000
people. Eventually production will reach a peak and then be gradually
phased out as the processor is replaced by newer designs.

Processor Roadmaps
The design of any microprocessor has to start with an idea of what type
of product will use the processor. In the past, designs for desktop com-
puters went through minor modifications to try and make them suitable
for use in other products, but today many processors are never intended
for a desktop PC. The major markets for processors are divided into
those for computer servers, desktops, mobile products, and embedded
applications.
Servers and workstations are the most expensive products and there-
fore can afford to use the most expensive microprocessors. Performance
and reliability are the primary drivers with cost being less important.
Most server processors come with built-in multiprocessor support to
easily allow the construction of computers using more than one proces-
sor. To be able to operate on very large data sets, processors designed
for this market tend to use very large caches. The caches may include
parity bits or Error Correcting Codes (ECC) to improve reliability.
Scientific applications also make floating-point performance much more
critical than mainstream usage.
The high end of the server market tends to tolerate high power levels,
but the demand for “server farms,” which provide very large amounts
of computing power in a very small physical space, has led to the cre-
ation of low power servers. These “blade” servers are designed to be
loaded into racks one next to the other. Standard sizes are 2U (3.5-in
thick) and 1U (1.75-in thick). In such narrow dimensions, there isn’t
room for a large cooling system, and processors must be designed to con-
trol the amount of heat they generate. The high profit margins of server
processors give these products a much larger influence on the processor
industry than their volumes would suggest.
Desktop computers typically have a single user and must limit their
price to make this financially practical. The desktop market has further
differentiated to include high performance, mainstream, and value
processors. The high-end desktop computers may use processors with per-
formance approaching that of server processors, and prices approaching