Processors 101 - xeon, prescott, celeron? What's the diff?

So I’m going to build a new computer. When I check with my retailer, they have a whole bunch of intel and AMDs. Not only are there clock speeds to consider, although, I understand that’s not all that important anymore. But there are so many different processors: Xeon, Prescott, Celeron, Itanium, Pentium, Centrino.
What? Why?

Basically, they’re different model types, and usually denote a change in the chip architecture. Some are faster, or more power efficient, or smaller, than others.

Kind of like cars, now that I think about it.

Anyway, if you have a wide variety of processors to choose from, take some time and read up on them. As you mentioned, pure clock speed is not as important anymore, because you can’t always directly compare two chips of different types. Chip A might be billed at 2.0 GHz, but run a good deal faster than chip B at 2.4 GHz, simply because chip A has a more efficient architecture or has a 64-bit instruction set.

They target different segments of consumers. I’m no longer conversant with the different types, so I can’t be much help, but I can give you a few generalizations.

The celeron was a scaled down version of the Pentium (not sure if this is still the case). It had less L2 cache, which made it cheaper at the expense of processing power. The centrino, I believe, is a mobile processor, which means that it uses less power. I further believe that it actually performs better on a clock tick to clock tick basis than a “standard” Intel CPU. Prescott refers to the core, if I recall correctly, but I can’t say what processor lines it spans. Xeon are targeted at servers; never having had to purchase a server, I can’t begin to give you differences.

Hope that helped a little. You should be able to find everything you’d like to know on the Intel website or another website (like Tom’s Hardware or anandtech).

Well, that’s the thing - after going around in circles for two hours on the Intel webpage, I’m non the wiser. I’ve realized that some processors are targeted for servers, others for laptops, but that’s about it. I also realize that Intel makes processors the way car makers work - do I want a smaller car for driving around town or a powerful truck to haul heavy loads?
So intel are now puching processors that “support hyper threading”. All fine and dandy,but there are a number of processors doing that:

So there must be a difference between a 3.40E and a 3.40 Ghz, but I haven’t found an explanation in layman’s terms. Or if there is a discernable difference between a 3.20 and a 3.40 processor.

I came across this on Tom’s Hardware for you – a 65 CPU roundup that might help. It looks to be quite a read, so grab a snack and make sure you’re comfortable before you start.

Poking around on ArsTechnica, I came across an article about Prescott. Give the first page a read; it’s pretty informative, although it has lots of jargon to wade through. I also came across an article about Intel’s new naming convention; it goes into some depth about various CPU topics that you might find useful.

Right now, AMD is kicking Intel’s butt. Their AMD64 models are faster, cheaper AND cooler.

If you’re building a new pc, head over to the forums at anandtech.com and ask the folks there for advice. They know their stuff and they never failed me.

Centrino isn’t a processor. It’s an Intel brand name that applies to the combination of a Pentium M processor, a low power Intel chipset and an Intel wireless network adapter. If it’s not all-Intel, manufacturers are not allowed to put that Centrino sticker on the case.

Oh, yeah - I meant to correct where I said that with a follow-up and forgot. I also wasn’t sure exactly what “Centrino” implied beyond the Pentium M. Thanks.

Here is my quick guide to the major processor families from AMD and Intel.

AMD has 4 major processor familes available: Opterons, Athlon 64s, Athlon XPs, and Semprons, not to mention the lower power mobile versions of the A64s and AXPs.

Opterons: - these chips are designed for workstations & server use, so they are probably not something you would want to buy. They are capable of running 64-bit control, but AMD designed them to be completely backwards compatable with existing 32 bit code, which they also run very quickly. Probably the best value in this area.

Athlon 64’s - these chips are quite similar to the Opteron’s; the main difference is that they don’t support registered memory, and the Socket 754 have only 1 memory channel. They are very quick per clockspeed; a 2.0ghz A64 will generally match a 3.0ghz Pentium 4. That is why they use a number scheme, like the 2.0ghz Athlon 64 3000+. Athlon 64’s come in two flavors - the Socket 754 chips, and Socket 939 chips; the difference being that the 754 chips have only one memory channel, while the 939 chips have two, though this only adds a small bit a performance to the 939 chips in the real world. Note that the Athlon 64 design is extremely good at gaming; if the main heavy duty use of your machine is games, then this is the chip to get; an Athlon 64 2800+ will often match a 3.2-3.4ghz Pentium 4 in gaming.

Athlon XP - these are AMD’s older chips, but they are still found in plenty of systems. 32 bit, they are pretty good performers clock-for clock - for example, the AthlonXP 2800+, which runs at 2.08ghz, will usually match a 2.8ghz Pentium 4 in performance.

Sempron - these are AMD’s newer budget chips. These are tricky, as they come in two different types. The first are renamed AthlonXP chips, and use the same motherboards - the Sempron 2800+ is actually an AthlonXP 2400; note that AMD flipped the numbers so that the number would correspond to a Celeron-D clockspeed, not a Pentium 4. So the Sempron 2800+ is usually about as fast as a 2.8ghz Celeron-D.

The other type of Sempron’s are based on the Socket 754 Athon 64’s, and use the same motherboards. The difference is they don’t support the 64bit extensions, and have less L2 cache. These chips are very good performers, especially in games; the Sempron 3100, which runs at 1.8ghz, will usually match a 3.0ghz Pentium 4 in gaming. These are probably the best chips for someone on a budget.

Intel has 9 major processor families available: Itaniums, Xeons, Northwood Core Pentium 4s, the 5xx Prescott core Pentium 4’s, 6xx Prescott core P4s, Celerons, Celeron-Ds, Pentium-Ms, and Celeron-M chips,

Itaniums (aka Itanics) are Intels heavy duty server chips, and are not something you are likely to buy. They are 64 bit, and use a different instruction set than the other chips; they haven’t been all that succeful, even though Intel sunk a lot of money into them.

Xeons - these are Intels server/workstation chips based off of the Pentium 4 chips. Can be used in multi-processor systems. Also something you are not likely to buy. The latest Xeon chips come with Intel’s copy of AMD’s 64-bit extension.

Northwood core Pentium 4’s - Intel’s older mainstream chips, coming in clockspeeds from 2.4-3.4ghz & having 512kb of L2 cache, and a FSB of 800mhz. These are getting hard to find, as the newer, cheaper to produce Prescott core Pentium 4s have replaced them. Usually, they have a “C” after them - for example, a 2.8Cghz chip.

5xx Prescott core Pentium 4’s - Coming in clockspeeds from 2.8ghz to 3.8ghz & 1MB L2 cache & 800mhz FSB, Intel did a major revision to the Pentium 4 design with these chips, but overall these chips are worse than the Northwood core - clock for clock they are usually slower than Northwood core chips, and they run much hotter to boot. These chips are cheaper to produce than the Northwood core ones (due to Intel using a 90nm manufacturing process, rather than the 130nm one used with Northwoods) and thus these are the easiest to find.

6xx Prescoot core Pentium 4’s - These are pretty similar to the above Prescott core chips, though they have 2MB L2 cache, and have the same 64-bit extensions found on AMD’s Athlon 64 bit chips. They have slightly better performance, and are a bit better on the power consumption/heat output side. These are probably the best chips for someone who needs to do video encoding.

Celeron’s - Intel’s older budget chips, which are basically cut down Northwood core chips, with a slower frontside bus (400mhz), and only 128kb L2 cache. They are pretty poor performers for there clockspeed; a 2.6ghz Celeron will generally be beaten by a 1.6ghz Duron or 1.8ghz Pentium 4.

Celeron-Ds - Intels newer budget chips, which are basically cut down Prescott core chips, with only a 533ghz FSB and 256kb of L2 cache. In a reverse of their parent chips, the Celeron-Ds are much better clock for clock than older Celerons, and hold their own reasonably well against most of AMD’s competing Sempron budget chips.

Pentium-M’s - this is Intel’s best mobile processor; do not confuse it with a Pentum 4-M (basically, a slightly modified Pentium 4) - these chips are very low power, and offer very high performance for their clockspeed; a 1.6ghz Pentium-M will often outmatch a 2.6ghz Pentium 4. Generally, the best chips for Laptops.

Celeron-M’s - based on the Pentium-M chips, they lack some of the power saving features of the Pentium-M chips, and have half the L2 cache. Still good performers, just don’t confuse them with the Pentium 4 based Celerons, which also have found their way into notebooks.

<hijack>

Random bit of trivia that may or may not interst people, but the conversation is in the same direction

When intel or AMD sits down to build a run of chips they dont try to produce a specific clock speed.

For example, when AMD makes a run of 2000 Semprons. At the time of production there is no clock speed set on them (IE sempron 2800) Once the chips are produced AMD then tests each one individually to see how fast it can run before it begins to over heat. So lets say that one chip runs fine at what would be a semprom 2600, but begins to overheat with stock cooling at the 2800 setting. It gets labled as a Sempron 2600 and sold as such.

That is why with beyond stock cooling it is possible to overclock a chip for higher performance

</hijack>