Databases Reference
In-Depth Information
linear ways as the system resource becomes saturated. If we model the overhead as
To
(
N
), then the speedup becomes approximately:
13.2
13.1.3 Multicore CPUs
One of the major advances in CPU fabrication in the mid-2000s is the development of
multicore CPUs. These CPUs are a single CPU in appearance and are purchased as such
but internally have multiple CPUs within the chip. This allows parallel processing to
occur within a single CPU fabrication and greatly reduces some of the processing over-
head associated with parallel processing, such as cache coherence across cache lines (see
section 13.3.2, where cache coherence and false sharing are discussed). The following is
from Intel Corp.,
Multicore Processing
, 2006:
Have you ever waited impatiently for your PC to complete a compute-intensive task?
Well those long waits are now a thing of the past. In April 2005, Intel ushered in a new
era of processor architecture by releasing our first dual-core processor. Dual-core proces-
sors are the first step in the transition to multicore computing. Intel is already conduct-
ing research on architectures that could hold dozens or even hundreds of processors on a
single die. So what exactly is multicore?
Intel multicore architecture has a single Intel processor package that contains two or
more processor “execution cores,” or computational engines, and delivers—with appro-
priate software—fully parallel execution of multiple software threads. The operating sys-
tem (OS) perceives each of its execution cores as a discrete processor, with all the associ-
ated execution resources. . . .
We forecast that more than 85 percent of our server processors and more than 70 percent
of our mobile and desktop Pentium family processor shipments will be multicore-based
by the end of 2006.
13.2
Client Server Architectures
The term client server was introduced in the 1980s to describe the relationship between
computers on a network. The term gained broad adoption in the 1990s and is widely
used to refer to processing performed between a series of remote computers with a
larger, more powerful central computer, usually through a message-based communica-
tion protocol, such as TCP/IP. The server provides services to one or more clients, such
as access to database data.
