Getting to the 'Core' of Processors

The advancements in technologies such as very large scale integration (VLSI),
ultra large scale integration (ULSI) and wafer scale integration (WSI) have made
it possible to integrate millions of transistors on a single chip with great
precision. The microprocessor as we know is an integrated circuit on which all
the components of a computer are present on a single chip. These are capable of
controlling devices ranging from microwave ovens, computers, cell phones to
robotic arms of space ships and have become ubiquitous in the fields of
computing, communications, manufacturing, etc. Allowing more circuitry to be
packed on each chip, processors have consistently been migrating to smaller
feature sizes, resulting in becoming more efficient in terms of both performance
and energy consumption.

Server class processors

Intel and AMD both have been battling it out in the enterprise consumer
market of servers, blade servers and workstations with their Xeon and Opteron
brand of dual processor and multi processor configurations. Designed for better
performance than their desktop counterparts, these processors have more cache
with better multiprocessing capabilities and have been maintained over several
generations of x86, x86-64 bit processors.

In 2005, Intel and AMD released their first dual core Xeon and Opteron
processors respectively, and since then both consumer and server markets have
constantly seen rise in popularity of multi-core processors, effectively
increasing the processing power on each socket two to four times, and as the
cost of motherboards increase dramatically with the increase in number of
sockets, building a more powerful machine has become a little more economical
with multi-core processors.

AMD's Opteron

First released in April 2003 with the sledgehammer core, Opteron is the
server line of x86-64 processors from AMD and were initially made on 135 nm
process. Opterons had gained popularity when it comes to multiprocessor
machines, where the CPUs communicated using the Direct Connect architecture over
HyperTransport, a bidirectional serial/parallel high bandwidth, low latency
point to point link. Transparent to the programmer, each CPU can access the main
memory of another processor. The Direct Connect is the I/O architecture which is
also used in AMD's Athlon X2 and Phenom processors.

In May 2005 AMD introduced their first multi-core Opteron CPUs having two
separate processor cores on each die. With multicore Opterons one socket could
delver the performance of two processors and two sockets could deliver of four.

The second generation Opteron processors had three categories, the 1000
series (1processor/2 cores), 2000 series (2 processors/4 cores) and the 8000
series (4 processors/8cores). Based on code named Barcelona core design, the
third generation Quad core Opterons based on a 65 nm process were launched in
2007 AMD is the latest in the line server class processors. AMD claimed that the
new K10 architecture based processors had better power and thermal management
and incorporated a number of improvements, mainly in memory prefetching,
speculative loads, SIMD execution and branch prediction yielding performance
improvements over K8 architecture Opterons. Featuring their Powernow and
Coolcore technologies, AMD claims that the Quad core Opterons are the most power
efficient server processors they have ever produced. These processors also
supported the NX bit and AMDs virtualization technology AMD-V.

Intel's 45 nm quad-core MCM: Two dual core dies are placed next to each other to form a four core configuration.

Future

A core codenamed Shanghai should be seen in the future Opteron processors which
will be based on a 45nm process. Also rumours are that AMD is working on their 6
core Sao Paolo and12 core Magny Cours processors which will be based on the MCM
technique. Further, the server line of processors will incorporate a code named
Buldozer processor with 4 or more cores on a 32 nm process, each supporting SSE5
aimed at better HPC and cryptographic computations.

Intel's Xeon

The Xeon branded 5200 series codenamed Woodcrest, based on the Intel's
core-micro architecture was the server and workstation version of the Intel Core
2 processor. The fastest processor in this category operated at 3.0 GHz,
claiming better performance and also less energy consumption than previous
processors. In Jan 2007, Intel launched its quad-core, core 2 quad, as the 3200
series which comprised of two separate dual core dies placed next to each other
in one CPU package and was targeted for the blades. The 3300 series was similar
to 3200 but was manufactured using 45 nm process and featured XD bit and
virtualization technology.

True to Intel's tick-tock release cycle of processors where a tick means a
refresh of the current architecture and tock means a brand new architecture, the
clock ticked and the Harpertown Xeons were released in late 2007. This family of
processors consisted of dual die quad core processors manufactured on a 45 nm
process and featured 1333 to 1600 MHz front side bus with lesser TDPs rated
between 50W to 150W depending upon the model.

With Intel becoming an undisputed leader in the quad core arena, the last in
this segment was the 7300 series code named Tigerton, consisted of two dual core
Core2 architecture silicon chips on a single ceramic module. Claiming greater
processing capabilities, the Tigerton was based on the Intel's Caneland (Clarksboro)
platform.

Intel's 45 nm dual core die

Intel's latest offering

This month Intel has launched its new server class processor codenamed
Dunnington. Dunnington is a 6 core (yes you got it right, 6 cores) processor and
a first of its kind. It means that on a four socket motherboard installing four
Dunnington processors would give you a staggering 24 cores. It features a
single-die six core design and is based on Intel's 45 nm Penryn process. It is
the first Xeon processor to feature 16 MB of L3 cache on board which helps
improve performance by reducing latency in accessing frequently used data.

Desktop processors

A couple of years back, dual or multi-core processors changed the way
desktop processors would function. Intel and AMD both have offerings in this
segment under their Core2 and Athlon X2 brands respectively. Introduced in mid
2006, the brand Core2 refers to Intel's consumer range of 64-bit dual core and
quad core microprocessors based on the x86-64 instruction set. It is based on
Intel's core micro architecture and is an advanced version of the dual core
Yonah microprocessor. The brand Core2 consists of Duo (double core), Quad (four
cores) and Extreme (dual and quad cores for serious enthusiasts), on the other
hand AMD has desktop class processors called Phenom is based on the K10
architecture and consists of triple core versions (Toliman) belonging to the
Phenom 8000 series and the Quad core processors (Agena) that fall in the Phenom
9000 series.

A quad core processor is similar to the dual core with the basic difference
that it has four cores instead of two. Intel's multi-chip module (MCM) or the
quad core version had two separate dual core dies placed next to each other to
provide a quad core configuration; on the other hand AMD claims their Phenom X4
series to be 'true' quad cores as they have all four cores on a single piece of
silicon wafer unlike Intel. Though the clock speeds of these processors are
reduced, the main advantage of core architecture lies in more efficient decoding
stages, execution units, increased cache memory and reduced power consumption.

The core 2 branded processors from Intel include dual core processors for
desktops named Conroe, Allendale and Wolfdale, whereas the Quad core for
desktops computers consists of Kentsfield and Yorkfield processors. Most of
these processors support features such as Virtualization Technology, ED bit and
SSE3. The Core2 Extreme has Conroe XE (dual core) and Kentsfield XE and
Yorkfield XE (Quad core) processors. The Yorkfield XE is also the first desktop
processor from Intel with 45 nm technology and high k metal gates and features
SSSE 4.1 instruction set.

Intel's future processors

Nehalem is the codename given to Intel's future micro architecture which is
being developed under the Core i7 brand name, comprising a new series of quad
core processors. Due for release in late 2008, processors based on this new
micro architecture shall target high-end desktop and mobile platforms. They are
the successor to the Core architecture and will be based on the 45nm process. It
appears from Intel's spring IDF 2008 that Nehalem is going to incorporate one of
the most significant changes to the current microprocessor architecture. Intel
claims that Nehalem represents the next step in processor energy efficiency,
performance and dynamic scalability. Dynamic scalability in a processor would
mean managed cores, threads, interfaces, cache for energy efficient power on
demand. These processors shall also feature Intel's Quick path technology which
allows the concept of scalable shared memory between different cores with
integrated memory controllers. Intel shall also produce processors based on
Nehalem on a 32 nm process later.

AMD's future processors

AMD is expected to launch its new line of 45 nm processors code named Deneb
FX for Phenom FX, Deneb for Quad core Phenoms, Heka for triple core and Regor
for Athlon X2 with DDR3 compatibility and probably larger shared L3 cache by the
end of 2008.

Fusion is a future next gen microprocessor that is being developed by AMD and
ATI. It is supposed to combine the general processor execution and 3D geometry
processing and other functions that are performed by a GPU into as single
package and is expected to be first seen in 2009.

'Atom'ic energy

With increase in popularity of ultra mobile personal computers, Intel has come
up with a low power x86 and x86-64 family of microprocessors called Atom. Atom
is Intel's smallest processor ever made, designed for a 45 nm CMOS process and
is intended to target portable and low power applications. It's a low cost
processor with lower clock speeds but powerful enough to run the new wave of
mobile internet devices (MID) and simple entry level low cost laptops. With
devices known as netbooks and nettops gaining popularity, Atom is an ideal
option due to its low power consumption, which results in improved battery life.
Though it's based on an entirely new architecture, Atom was developed to fulfill
the targeted performance with low power consumption while maintaining
compatibility with Intel's Core micro architecture instruction set. It also
features multiple threads for better performance and increased system
responsiveness. Both Indian and other manufacturers have been adopting the
processor in their UMPC and entry level notebooks. These are competitively
priced around 20k or even less depending on the configuration provided. Devices
running on these processors are good enough to keep you connected while on the
move and run your basic productivity apps and capabilities to run entertainment
features.

Processors for notebooks

For notebooks, people are often found confused between platform and processor
naming trends, because laptops are based and advertised on the platform instead
of the processor alone, a little background check on them is necessary. The
platform of a laptop comprises of the main board chipset, the processor and the
wireless network interface tweaked to deliver optimum performance, broader
wireless interoperability and consume minimum power at the same time. This
provides sufficient battery backup while the processor can perform optimally.


There are a number of platforms from Intel including the Napa platform (2006)
that comes with the Centrino logo; the Santa Rosa platform, introduced first in
2007 comes with a Centrino Pro logo. Centrino has been Intel's brand name for
their notebook platform, and in August the company raised curtains over their
fifth generation Centrino2 (Montevina). It claims lesser power consumption than
previously available platforms. These new age platforms boast an Intel Core 2
Duo (code named Penryn) 45 nm processor with clock speeds ranging from 2.26 GHz
to 3.06 GHz, the mobile chipset offers RAM support for DDR2-667, DDR2-800,
DDR3-800, DDR3-1066, DDR3-1333 SODIMM and NAND flash memory caching branded as
Intel turbo memory along with Gigabit Ethernet LAN controllers 82567LM and
82567LF. On the other hand AMD has failed to gain popularity due to less
advertisements and marketing. The latest in platforms from AMD this year is Puma
which features the dual core Turion Ultra or the mobile Sempron processor and
graphics from ATI.

AMD's take on Atom

AMD is expected to launch a very simplified x86 processor codenamed Bobcat
aiming at very low power processing with TDP between 1 and 10W. Bobcat is
expected to make its debut in the UMPC, handheld devices and other small form
factor consumer devices.