by December 17, 2005 0 comments



If you are in the market for an upgrade, you will be in for
a fair bit of shock. Processor names have gone from the simplistic core speed
figures to complex numbered variants which, Intel says,

depend on features and the added features that these processors carry. Intel
has  perhaps come around to the view that final performance is not just
dependent on the clock speeds but also depends on the entire system as a whole
(something AMD has been saying for quite a while, we might add).

Direct Hit!
Applies to:
All PC buyers
USP:
Understand the nomenclature across Intel’s entire range of P4s
Primary Link:
www.intel.com 
Google keywords:
Intel P4 5xx series, 6xx series, 8xx, Conroe, Presler 

If you’ve just started looking at buying a computer, no 
longer can you just ask for a simple 3.0 GHz processor and get it. 
Ever since Intel has decided to revamp its naming conventions, you have
to contend with such confusing nomenclatures as 561, 640, 820,

Prescott


, 

Smithfield


, so on and so forth. The confusion that people face is near traumatic which
makes you think whether or not the renaming exercise was really a smart move.

We can’t change Intel’s nomenclature (although we do
hope they make it better!) but we can analyze and make 
them simpler  to understand so
you don’t face problems when you are out in the market buying an Intel
processor.

Currently Available Processors from Intel
Processor Series Speed  Core EIST/EM64T
XD/HT
Codename
Pentium EE 840 3.2 Dual Y/Y/Y/Y Smithfield
Pentium D 840 3.2 Dual  Y/Y/Y/N Smithfield
Pentium D 830 3 Dual  Y/Y/Y/N Smithfield
Pentium D 820 2.8 Dual  Y/Y/Y/N Smithfield
Pentium 4 670 3.8 Single Y/Y/Y/Y Prescott 2M
Pentium 4 660 3.6 Single Y/Y/Y/Y Prescott 2M
Pentium 4 650 3.4 Single Y/Y/Y/Y Prescott 2M
Pentium 4 640 3.2 Single Y/Y/Y/Y Prescott 2M
Pentium 4 630 3 Single Y/Y/Y/Y Prescott 2M
Pentium 4 571 3.8 Single N/Y/Y/Y Prescott
Pentium 4 570J 3.8 Single N/N/Y/Y Prescott
Pentium 4 561 3.6 Single N/Y/Y/Y Prescott
Pentium 4 560 3.6 Single N/N/N/Y Prescott
Pentium 4 551 3.4 Single N/Y/Y/Y Prescott
Pentium 4 550 3.4 Single N/N/N/Y Prescott
Pentium 4 541 3.2 Single N/Y/Y/Y Prescott
Pentium 4 540 3.2 Single N/N/N/Y Prescott
Pentium 4 531 3 Single N/Y/Y/Y Prescott
Pentium 4 530 3 Single N/N/N/Y Prescott
Pentium 4 521 2.8 Single N/Y/Y/Y Prescott
Pentium 4 520 2.8 Single N/N/N/Y Prescott
Note: All these processors are based on 90 nm fabrication process. Also these processors except the Pentium EE run at 800 FSB. Pentium EE runs at 1066 FSB

If the names were as simple as a 5 series or 7 series or 8
series, we would’ve somehow managed to figure out and learn all the various
names and codes, but the situation with Intel is such that we don’t remember
the processor names without having a big fat manual at hand or a high capacity
storage drive for  memory! 
So there are modifications to each series which may or may not be
consistent across different series.

Before we look at Intel’s future processors, let’s
first take a look at Intel’s present offerings and their complex names.

Intel 5xx Series
The 5xx processor series is based on the

Prescott


core which in turn is based on the very heat-inefficient Netburst architecture.
It’s a 90 nm fabrication process which means that the chip should
theoretically run cooler. The 5xx series   
features only 1 MB of L2 cache and requires a massive 130 W of power. The
6xx series is based on the updated

Prescott


2M core with 2 MB of L2 cache. We will take a look at the 6xx series later in
ths article but for now, let us  get
back to understanding the intricacies of the 5xx series.

The 5×0 is the low-end model here and between the 520, 530,
…, 560, the only difference  is
that of clock speed. All of them support only HT (hyperthreading) and are not
bestowed with such goodies as the XD (Execute Disable), EM64T or SpeedStep.

The consistency gets broken at the 570, which is marked
with a ‘J’. This shows that the processor supports the ‘Execute Disable’
function which is basically allows the processor to specify memory areas where
codes are allowed to execute and where they aren’t. This takes care of a lot
of buffer overflow attacks.

The 5×1 series is the higher-end version of the 5xx series
and features all the goodies such as EM64T (Intel’s interpretation of 64bit
computing), XD as well as HT. The 5×1 series however, doesn’t get EIST or
Enhanced Intel Speed Step Technology.  This
means that even if your processor is not getting the work load, it will still
run at the maximum speed.  This
generates more heat and consumes more power. With EIST, your system will
automatically under clock your processor depending on the workload and as soon
as the workload increases,  it will
automatically clock it to run at the full speed.

The entire 5xx series uses the soon to be outdated 90 nm
fabrication process, and have an 800 MHz bus speed.

Intel’s Desktop Processor Series (all have different cores)
Processor Series Core name Fab Power Core
Pentium 4 6×2 Prescott 2M 90 nm 95 W Single
Pentium 4 6×1 Cedar Mill 65 nm 86 W Single
Pentium 4 6×3 Cedar Mill(VT) 65 nm 86 W Single
Pentium D 8xx Smithfield 90 nm 130 W Dual
Pentium D/EE 9xx/955 Presler 65 nm NA Dual
NA NA Conroe 65 nm NA Dual
Note: All the above processors have an LGA 775 socket

Intel 6xx series
The 6xx series, as mentioned above, is a step up from the 5xx series (notice
how  the series work in a way similar
to BMW? Series 5, series 6 and so on) The current processors available in this
series are based on the updated

Prescott


2M core  featuring 2 MB of L2 cache
and 800 MHz FSB. All 6xx series processors feature EIST, EM64T, XD and HT. 
Within the 6xx series we have a bifurcation. The 6×2, based on the
currently used Prescott core (with enhanced cache of course) and the 6×1 and 6×3
series which are based on a brand new core code named Cedar Mill.

Intel 6×2
Intel is going to completely revamp the 6xx series beginning 2006. That is
the timewhen Intel is scheduled to launch Cedar Mill processors. However, Intel
still feels there is life in the old Prescott core and it recently launched the
662 and 672 series of processors (not in India yet). These feature the same

Prescott


core with more cache (2MB) and hardware support for virtualization.

Intel 6×1 and 6×3
The 6×1 is when the fun will start. This has a completely revamped
fabrication process and Intel will be launching multiple products in this range.
Codenamed Cedar Mill, Intel’s 6×1 and 6×3 series are based on a 65 nm
manufacturing process and are scheduled to ship in Q1-Q2 2006. The processors
will feature 2MB L2 cache as well as Intel’s virtualization technology VT on
the 6×3 series. Plus, both series will run at an 800 MHz FSB.

However, the basic architecture remains fairly similar to
the Netburst architecture (except for the smaller die) 
hence we don’t really expect it to do any better than the 6×0 series we
have tested. However, due to it’s 65 nm fabrication, 
the heating issues that have long plagued NetBurst will probably be taken
care of. This is perhaps from Intel’s learnings in the Pentium M segment,
which utilizes the ultra efficient Banias architecture which, in-turn, is a
derivative of the PIII!

The interesting thing about Cedar Mill is that apart from
the obvious lower power consumptions, it is being used as the core for the
upcoming upgrade of the

Smithfield


processors (Intel’s current Pentium D processors, 8xx series). The upgraded

version has been codenamed Presler, and is really just two Cedar Mill cores
fitted into the LGA 775 form factor.

The 8xx , Pentium EE 840 and Presler (9xx)
The 8xx is Intel’s current dual core

series and is codenamed Smithfied.  We
know it better as Pentium.

The Pentium D 8xx series has an 800 MHz FSB. The Intel EE
840, however, gets a one up here and comes with a 1066 MHz

of FSB.

Even though the 8xx series is 64 bit, only the 830 and 840
come equipped with XD and EIST. As always, only the Extreme Edition has hyperthreading
enabled.

Even though the 820 and 830 versions operate at a maximum
TDP of 95 W, since the architecture is basically two Netburst cores stuck
together, the platform supports TDP up to 135 W. This was essentially for the
Extreme Edition 840 version which obviously has higher power requirements.

For communication between the cores, Intel has provided a
dedicated bus that lets each core access the other’s L2 cache. This method
turned out to be hugely inferior to AMD and Intel plans to fix this in its
upcoming processors Presler and Conroe.

Upcoming Processors From Intel
Processor/
Core
Series Clock Speed L2 Cache/FSB/Fab /Socket/Power/Core
P4/Prescott 2M VT 662/672 3.6/3.8 GHz 2 MB /NA/90 nm/LGA 775/NA/Single
P4/Cedar Mill 641/651/
661/671
3.2/3.4/
3.6/3.8
2 MB/800/65 nm/LGA 775/86W/Single
Pentium D EE/ Presler 955 3.46 4 MB/1066/65 nm/LGA 775/NA/Dual
Pentium D/ Presler 920/930/ 940/ 950/960 2.8/3/3.2/
3.4/3.6
4 MB/800/65 nm/LGA 775/NA/Dual
NA/Conroe Not Available Not Available 4 MB/NA/65 nm/LGA775/ 60-70W/Dual

Presler
Presler is the upgrade from Smithfield and has, thus, been given the 9xx series.
Presler’s might be a stop gap measure or an attempt by Intel to test its 65 nm
processors as the brand new core Conroe is expected to be launched in end of 06
or Q1 07.

There aren’t too many details available about Presler as
of now but it will be made up of two Cedar Mill cores (as mentioned above) and
will feature 2 MB L2 cache per core, giving it a total of 4 MB. Whether or not
it contains the same bus limitations of the Smithfield is not clear and Intel
wouldn’t comment on it.

Presler will support all technologies supported by the
Pentium D 840, namely, EM64T, XD, VT and EIST.

Since Presler will be equipped with 2 MB of L2 cache for
each dice,  we expect it to
(theoretically anyway) peform better than the current Pentium Ds (Smithfield
core).

Both Presler and Cedar Mill are based on the thinner 65 nm
fabrication process. This means that they should run cooler than their 90 nm
brethren. This means that  you can
overclock them to a fair bit without getting into too much heating trouble.

Conroe
Conroe will signal the final departure of the Netburst architecture. It is a
brand new processor which Intel has designed from the ground up.

It reflects Intel’s new philospophy of performance/watt
so even though it is a desktop processor, we can expect it to be extremely
efficient. Intel has used its experience from the the Prescott and the Pentium
Ms. It has combined the performance of the Prescott with the efficiency of the
Pentium Ms. This will result in processors that will consume lesser power, will
run much more efficiently and will, thus,

remain significantly cooler than their predecessors.

Intel did this by reducing the instruction pipeline from 31
to 14 and by making Conroe a four execution core. The processor will be dual
core and will contain 2 MB L2 cache/core, giving it a total of 4 MB of L2 cache
memory.

Talks are also on about a processor which has 8 MB L2 cache
but no confirmations from Intel are available on that. Rest assured, 
whatever happens, we’ll keep you posted.

Varun Dubey

No Comments so far

Jump into a conversation

No Comments Yet!

You can be the one to start a conversation.

<