Home Overclocking the Athlon XP-M 2500+ processor
Reviews

Overclocking the Athlon XP-M 2500+ processor

Andy Brown
Disclosure
Disclosure
In our content, we occasionally include affiliate links. Should you click on these links, we may earn a commission, though this incurs no additional cost to you. Your use of this website signifies your acceptance of our terms and conditions as well as our privacy policy.

THOSE OF YOU WHO REGULARLY PERUSE the news here at TR probably saw this item about the Athlon XP-M 2500+ and its recently discovered overclocking potential. The chips have become so popular so quickly that their street price has jumped 33% in recent days. Of course, given the accounts of overclocking to 2700MHz “on air” (using only a heatsink/fan) it’s no wonder they’re so popular.

Needless to say, we couldn’t sit this one out, so last week we ordered up an Athlon XP-M 2500+, and have spent the last 24 hours or so putting it through its paces. Read on to find out more about this overclocking wonder, and see how our particular chip fared.

XP-M: It’s not just for notebooks anymore
The first question in many people’s minds is: What’s up with putting a mobile chip in a desktop system? Well, there are a couple of important factors that make the XP-M such a desirable chip from an overclocking standpoint. The first is the default voltage of the chip. Desktop Athlon XP 2500+ chips run at a frequency of 1.8GHz using a default voltage of 1.65V. That’s fine for desktop systems that don’t have to worry about power consumption or battery life, but it’s a different story for mobile systems.

Because voltage relates directly to power consumption, one way to cut down on the amount of power used in a mobile application would be to lower the core voltage of the processor. Of course, it’s not quite that simple. Some 2500+ chips may continue to run properly at 1.8GHz if their core voltage is set below 1.65V, but others may become unstable or crash outright at anything below 1.65V. Every chip is different in this regard, and you won’t know until you try.

AMD does in fact use this method for the Athlon XP-M. Basically, they test their chips and determine which 2500+ chips can maintain their 1.8GHz clock speed properly with only 1.45V of core voltage instead of 1.65V. Chips that pass this test can then be set to a default voltage of 1.45V and sold as Athlon XP-Ms. Essentially, the XP-M 2500+ is the cream of the crop within its speed grade. Not only can it do 1.8GHz, it can do it with one hand tied behind its back, so to speak.


Up close: our Athlon XP-M 2500+

Typically, overclocking is a crapshoot because you never know how fast a particular chip is going to go. There are many tales of people taking Pentium 4 2.4Cs and overclocking them to 3.2GHz or higher with ease, but we have an example in our labs that refuses to go any higher than 2.6GHz. Overclocking a processor is like a box of chocolates….

In the case of the Athlon XP-M, however, AMD has basically created a part consisting of cherry-picked processors. After all, if a 1.8GHz chip is good enough to stay at 1.8GHz even when undervolted by 0.2V, how much higher than 1.8GHz might it go at default voltage? This is one big reason for the XP-Ms popularity with overclockers; grab an XP-M, and you know you’re getting something good. The only question is: how good?

The second important mark in the XP-M’s favor is what’s missing: a multiplier lock. AMD mobile systems use the PowerNow! power management feature to conserve battery life. Unlike a desktop system, where bus speed times processor multiplier equals processor speed, things in a PowerNow! system are much more fluid. The processor has a minimum and maximum speed, and the system is constantly adjusting CPU speed based on processor load. If your system isn’t working very hard (you’ve gone to lunch, or you’re working on a Word document) the processor will limp along at 800MHz. Fire up a game, and it may jump up to 1500MHz and stay there until you quit. The speed is constantly updated relative to system load, and can be set to any number of intermediate speeds between the minimum and maximum.

One of the ways that the system works this magic is by manipulating the multiplier on the processor. Thus, in order for PowerNow! to work, the processor’s multiplier must be unlocked. Since PowerNow! is an important part of AMD’s mobile strategy, all Athlon XP-M chips are factory unlocked.

Add these two factors together, and you wind up with a guaranteed high-quality, unlocked Athlon XP core for only $25 more (at the time of this writing) than a “regular” Athlon XP 2500+. Whatta bargain!

Putting theory into practice
So now that we’ve looked at why the XP-M should be a good overclocker, let’s see if our hypothesis holds up to the test. Our subject is an Athlon XP-M 2500+, manufactured in week 49 of 2003, with a stepping code of IQZFA.


The markings on each chip betray its origins

We started out on an Asus A7N8X motherboard, and quickly got the CPU up to [email protected] on a 200MHz front-side bus with little drama. Any attempts to push harder, however, were met with failure, possibly because the A7N8X refused to crank the voltage any higher than 1.825V.

At that point, we switched over to Abit’s new nForce 2 Ultra 400 board, the AN7 (review forthcoming). The AN7 had no qualms about pushing the voltage up to and past our own comfort level. In the end, though, it didn’t make much difference with this chip. We managed to get [email protected] (yes, 1.96, that’s one of the settings on the AN7), but it wasn’t stable enough to complete our benchmark suite, so we decided to back down to 2.4GHz at 1.8V and run our benchmarks there.


Our open-air test bench with Abit AN7 motherboard and bone-stock air cooling

Abit’s uGuru utils kept us aware of our temps, shown here at 2.4GHz and 1.8V Although we didn’t hit the 2.7GHz+ levels that some others have apparently reached, let’s not lose sight of what we did accomplish. This is a 33% overclock resulting in a 200MHz advantage over an Athlon XP 3200+ (currently going for $195), from a chip that cost $98.

We decided against experimenting with bus overclocking, because our intent was to test the processor’s capabilities, and bus overclocking would’ve thrown too many other variables into the equation, such as non-standard PCI and AGP clock speeds. We said it before, but it bears repeating: One of the big advantages of the XP-M is its unlocked multiplier, which allows for substantial performance gains without having to resort to bus overclocking.

 

Our testing methods
As ever, we did our best to deliver clean benchmark numbers. Tests were run at least twice, and the results were averaged.

Our test systems were configured like so:

Processor Athlon XP-M 2500+ 2.4GHz Athlon XP ‘Barton’ 3200+ 2.2GHz Athlon XP ‘Barton’ 3000+ 2.167GHz AMD Athlon 64 3000+ 2.0GHz
AMD Athlon 64 3200+ 2.0GHz
AMD Athlon 64 3400+ 2.2GHz
AMD Athlon 64 FX-51 2.2GHz Pentium 4 2.8’C’GHz
Pentium 4 3.2GHz
Pentium 4 3.2GHz Extreme Edition
Pentium 4 3.4GHz Extreme Edition
Pentium 4 2.8’E’GHz
Pentium 4 3.0’E’GHz
Pentium 4 3.2’E’GHz
Front-side bus 400MHz (200MHz DDR) 400MHz (200MHz DDR) 333MHz (166MHz DDR) HT 16-bit/800MHz downstream
HT 16-bit/800MHz upstream
HT 16-bit/800MHz downstream
HT 16-bit/800MHz upstream
800MHz (200MHz quad-pumped)
Motherboard Abit AN7 Asus A7N8X Deluxe v2.0 Asus A7N8X Deluxe v2.0 MSI K8T Neo MSI 9130 Abit IC7-G
BIOS revision 1.4 C1007 C1007 1.1 1.1 IC7_21.B00
North bridge nForce2 SPP nForce2 SPP nForce2 SPP K8T800 K8T800 82875P MCH
South bridge nForce2 MCP-T nForce2 MCP-T nForce2 MCP-T VT8237 VT8237 82801ER ICH5R
Chipset drivers ForceWare 3.13 ForceWare 3.13 ForceWare 3.13 4-in-1 v.4.51
ATA 5.1.2600.220
4-in-1 v.4.51
ATA 5.1.2600.220
INF Update 5.1.1002
Memory size 1GB (2 DIMMs) 1GB (2 DIMMs) 1GB (2 DIMMs) 1GB (2 DIMMs) 1GB (2 DIMMs) 1GB (2 DIMMs)
Memory type Corsair TwinX XMS4000 DDR SDRAM at 400MHz Corsair TwinX XMS4000 DDR SDRAM at 400MHz Corsair TwinX XMS4000 DDR SDRAM at 333MHz Corsair TwinX XMS4000 DDR SDRAM at 400MHz Corsair CMX512RE-3200LL PC3200 registered ECC DDR SDRAM at 400MHz Corsair TwinX XMS4000 DDR SDRAM at 400MHz
Hard drive Seagate Barracuda V 120GB ATA/100 Seagate Barracuda V 120GB ATA/100 Seagate Barracuda V 120GB ATA/100 Seagate Barracuda V 120GB SATA 150 Seagate Barracuda V 120GB SATA 150 Seagate Barracuda V 120GB SATA 150
Audio Creative SoundBlaster Live!
Graphics Radeon 9800 Pro 256MB with CATALYST 4.1 drivers
OS Microsoft Windows XP Professional
OS updates Service Pack 1, DirectX 9.0b

All tests on the Pentium 4 systems were run with Hyper-Threading enabled, except where otherwise noted.

Thanks to Corsair for providing us with memory for our testing. If you’re looking to tweak out your system to the max and maybe overclock it a little, Corsair’s RAM is definitely worth considering.

The test systems’ Windows desktops were set at 1152×864 in 32-bit color at an 85Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests.

We used the following versions of our test applications:

The tests and methods we employ are generally publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them. Before we proceed, I’d like to give mad props to Damage on this one. He helped me out immensely in getting this article done as quickly as possible, by running the benchmarks and taking the pictures, among other things.

 

Memory performance

Overclocked or not, the Athlon XP is getting long in the tooth, and it especially shows in the area of memory bandwidth. The XP-M comes in right where we’d expect, tied with the XP 3200+ (which shares its 400MHz bus speed) and beating only the XP 3000+ with its 333MHZ front-side bus. The XP chips just don’t have the juice to compete with 800MHz front-side busses and on-die memory controllers.

The 2.4GHz XP-M outpaces all of the AMD chips by a good margin, at least until the matrix sizes get outside its 512KB L2 cache. At larger matrix sizes, however, the Hammer chips’ 1MB L2 cache and on-die memory controller lay down a beating on the XP-M.

A slight jump in clock speed over the XP 3200+ gives the XP-M a slight edge in memory access latency, putting it ahead of the 3.4GHz EE and the 2.8GHz Prescott as well. Still, things are pretty bunched up here in the middle of the graph, and the Hammer systems just embarrass everything.

 

Unreal Tournament 2003

The XP-M jumps a couple of spots relative to the XP 3200+, eclipsing the 3GHz Northwood and 3.2GHz Prescott in one of the tests. Ah, the difference 200MHz makes.

Quake III Arena

Quake III Arena loves memory bandwidth. Repeat it like a mantra. The Athlon XPs get romped on by everything here, and the XP-M’s extra clock speed only helps it beat up on its little brothers.

Wolfenstein: Enemy Territory

The XP-M puts in a better showing here, beating out a couple of P4’s, but it’s a long way off the Hammer chips.

 

Comanche 4

Enjoy this video game based around the Comanche helicopter, because it’s not likely you’ll be seeing one in real life anytime soon. At any rate, the XP-M earns a respectable finish here, though it loses to a number of Northwoods and all the Hammer chips. For the record, Comanche 4 really, really hates Prescott.

Serious Sam SE

The XP-M does exceptionally well in Serious Sam, beating out all Intel comers with less than two megs of cache. It basically ties the 3.4GHz Northwood in spite of a one gigahertz clock deficit. Can you say IPC?

3DMark03

The composite 3DMark scores put the XP-M near the bottom, but if you throw out the XP 3000+ and its 333MHz front-side bus, there’s less than 300 points separating worst from first. On the CPU tests, the XP-M makes a respectable showing, even taking down a Hammer in one of the tests.

 

Sphinx speech recognition
Ricky Houghton first brought us the Sphinx benchmark through his association with speech recognition efforts at Carnegie Mellon University. Sphinx is a high-quality speech recognition routine that needs the latest computer hardware to run at speeds close to real-time processing. We use two different versions, built with two different compilers, in an attempt to ensure we’re getting the best possible performance.

There are two goals with Sphinx. The first is to run it faster than real time, so real-time speech recognition is possible. The second, more ambitious goal is to run it at about 0.8 times real time, where additional CPU overhead is available for other sorts of processing, enabling Sphinx-driven real-time applications.

Sphinx loves the memory bandwidth, and the results aren’t pretty for the XP chips.

LAME MP3 encoding
We used LAME to encode a 101MB 16-bit, 44KHz audio file into a very high-quality MP3. The exact command-line options we used were:

lame –alt-preset extreme file.wav file.mp3

LAME is about raw processing power and little else, and the XP-M’s high clock speed and high IPC leave it tied for fourth place with the 3.2GHz Northwood.

DivX video encoding

Ouch. Clearly the Athlon XPs are out of their league here, trailing the rest of the pack by a significant margin. Obviously, Xmpeg likes the P4. A lot. In fact, I saw them kissing under the bleachers. OK, I made that last part up, but it could’ve happened.

 

3ds max rendering
We begin our 3D rendering tests with Discreet’s 3ds max, one of the best known 3D animation tools around. 3ds max is both multithreaded and optimized for SSE2. We rendered a couple of different scenes at 1024×751 resolution, including the Island scene shown below. Our testing techniques were very similar to those described in this article by Greg Hess. In all cases, the “Enable SSE” box was checked in the application’s render dialog.

The XP-M does better than one might expect here, as the 200MHz clock speed advantage puts it over not only the XP 3200+, but also the 2.8GHz Prescott and two Athlon 64 chips.

 

Lightwave rendering
NewTek’s Lightwave is another popular 3D animation package that includes support for multiple processors and is highly optimized for SSE2. Lightwave can render very complex scenes with realism, as you can see from the sample scene, “A5 Concept,” below.

To test the effects of Hyper-Threading, we’ve tested the Hyper-Threaded processors with one, two, and four rendering threads. For non-Hyper-Threaded processors, we just tested with one and two threads.

Ouch. The Athlon XP chips just take a beating here; there’s no other way to put it. The XP-M just takes, umm, a slightly less severe beating.

 

POV-Ray rendering
POV-Ray is the granddaddy of PC ray-tracing renderers, and it’s not multithreaded in the least. Don’t ask me why—seems crazy to me. POV-Ray also relies more heavily on x87 FPU instructions to do its work, because it contains only minor SIMD optimizations.

POV-Ray is all about old school x87 math, and the results certainly show it. The mighty 3.4GHz P4 EE is laid low by the Athlon XP 3000+, as AMD just runs the table. For its part, the XP-M comes within a second of the top spot.

 

Cinebench 2003 rendering and shading
Cinebench is based on Maxon’s Cinema 4D modeling, rendering, and animation app. This revision of Cinebench measures performance in a number of ways, including 3D rendering, software shading, and OpenGL shading with and without hardware acceleration. Cinema 4D’s renderer is multithreaded, so it takes advantage of Hyper-Threading, as you can see in the results.

The XP-M actually ties with the FX-51 here, but that’s not saying much; the P4 and its Hyper-Threading just dominate here.

The other Cinebench tests aren’t multithreaded, so the Athlons do much better. The XP-M finishes in the middle of the pack.

 

SPECviewperf workstation graphics
SPECviewperf simulates the graphics loads generated by various professional design, modeling, and engineering applications.

The XP-M bounces around on the SPECviewperf tests, finishing as high as sixth in one test and third worst in another.

 

ScienceMark
ScienceMark 2.0 is optimized for SSE, SSE2, 3DNow! and is multithreaded, as well. The molecular dynamics simulation models “the thermodynamic behaviour of materials using their forces, velocities, and positions”, according to the ScienceMark documentation.

Primordia “calculates the Quantum Mechanical Hartree-Fock Orbitals for each electron in any element of the periodic table.” In our case, we used the default element, Argon.

The XP-M does extremely well here, finishing fourth place or better in all tests and winning the Cipher benchmark outright.

These last two tests, SGEMM and DGEMM, measure matrix math performance using several different codepaths optimized with several instruction set extensions, including SSE, SSE2, and 3DNow!

The Athlon XPs lag the rest of the entrants in Blas SGEMM, but redeem themselves nicely in Blas DGEMM.

 

picCOLOR image analysis
We thank Dr. Reinert Muller with the FIBUS Institute for pointing us toward his picCOLOR benchmark. This image analysis and processing tool is partially multithreaded, and it shows us the results of a number of simple image manipulation calculations. The overall score is indexed to a Pentium III 1GHz system based on a VIA Apollo Pro 133. In other words, the reference system would score a 1.0 overall.

The new version of picCOLOR we’re using today is optimized for SSE and SSE2, so it should perform differently than past revisions.

The XP-M does pretty well, beating out all of the non-EE Northwood chips. Here are the individual test scores for some of our key participants.

The XP-M gets whacked in GraphCopy, but finishes respectably in the other tests. For whatever reason, it absolutely owns the Video+Trgn test.

 
Conclusions
So to sum up, what we have here is a $98 CPU with a full 600MHz of overclocking headroom. What’s not to like? Sure, usually when we’re talking about overclocking, we issue the standard disclaimer: Every chip is different, your results may vary, yadda yadda yadda.

In this case, however, we feel like we should issue a disclaimer for the disclaimer, because everything about the XP-M suggests that the odds are in your favor. You don’t have to worry about getting a poor overclocker, because you’re guaranteed to get an undervolted chip. You don’t have to worry about a multiplier lock because you’re guaranteed to get an unlocked chip.

The only question, then, is how good is it going to get? Anecdotal evidence (based on our results as well as the Overclockers.com CPU database) suggests 2400-2700MHz is typical. AMD may not have intended it this way, but if you asked an overclocker what they would want in the perfect overclocking chip, they might very well describe the Athlon XP-M 2500+.