When websites compare products from different companies, they tend to mainly focus only on the bottom line performance and not the financial to performance aspect of the products. However, when consumers look to purchase a new component, whether it be an upgrade or for a new system, they look at it from a price to performance point of view rather than simply which is the best overall. With that in mind I've decided to compare a couple of mid-market processors installed in similarly priced systems to see what you can expect performance wise out of the two. 

Back in late Febuary Nate did a comprehensive write up on Intel's 6XX series processors. Today, we'll take a closer look at Intel's 640 processor, their 3.2GHz offering, and compare it to AMD's similarly priced Winchester 3500+ processor, which I reviewed back in November.

When Intel updated the Prescott line late last year with the 5XX series, most people were disappointed. Though the 5XX series processors overclocked as well as their predecessors, the second generation Prescott CPU did not support 64 bit processing, nor did it address the thermal issues that troubled the initial Prescott release just over a year ago.

Recently, everyone's attention has been focused on the "Dual Core Wars" as both Intel and AMD race to release their offerings to the enthusiast market. In the meantime, Intel has quietly released their 6XX series processor. Featuring 64 bit processing, better thermal properties, 2MB of onboard L2 cache, along with Intel's patented "SpeedStep Technology" the 6XX series seems to finally deliver what many of us expected with Prescott's intial release back in Febuaury of 2004.

Does the 6XX series really show the Prescott's true potential? Is it a worthwhile upgrade for those already using a 5XX series processor? Are Prescott's thermal issues finally behind it? And finally, how does Intel's newest desktop processor stack up against AMD's midrange CPU? 

Intel Pentium 4 660 Processor

Model 640 w/ H.T. AMD Athlon 64
Core Prescott Winchester
Operating Frequency 3.2GHz 2.2GHz
FSB 800MHz Integrated
Cache  L1/12K+16K; L2/2MB L1/64K+64K; L2/ 512KB
Voltage 1.4V 1.4V
Process 90nm 90 nm
Socket LGA 775 Socket 939
Multimedia Instruction MMX, SSE, SSE2, SSE3 MMX, SSE, SSE2, 3DNOW!, 3DNOW!+


While both Intel and AMD have their faithful supporters, most will agree that AMD rules the gaming world while Intel rules the multi tasking world. Yet, with each new release, people are always curious as to improvements a given processor might show.

While Nate covered all of the important information regarding the 6XX series processors in his earlier article, I chose to approach this from a different point of view. I took two closely priced systems based on different platforms, using the same video card, case, cooling, PSU, and HDD, to give an example of what they did compared over a range of benches. The AMD 3500+ retails for $270, while the Intel 640 runs about $280. I chose a high end board for both, the DFI SLI-DR is currently pricing about $240, while I purchased the ABIT AA8XE for $249 (Yes Fry's is expensive). So hopefully this will give those of you trying to decide which system to purchase a good indication of what to expect with both systems. 

The Intel systems will use the following components:

The AMD system will use the following components:

All tests were conducted under identical conditions with a fresh installation of WinXP Professional (32 bit) build 2600 with SP2.

Now that you have met the competitors, lets take a look at what we're going to ask of them. First, we'll focus on performance with a variety of 32 bit benchmarks. (After lengthy discussion, we decided against using 64bit WinXP until it is offically released later this month, there is also a lack of quality 64 bit benchmarks available, so we'll focus on what we do have). Lastly, i want you to understand this is more of an "opinion" article rather than a 'scientific" article, my point is that  there are features that are going to affect the overall performance and outcome of these systems and the benchmmarks used. for instance, you will see that no memory testing was done, this is for a couple of reasons. first, Intel's newest chipset, the 925XE uses DDR2 while AMD's integrated memory controller prohibits it from using DDR2. I also did not use a wide variety of gaming benchmarks, this is because i wanted to focus on the processors and their performance, not the overall systems, so I stuck with graphics applications that displayed the CPU's performance. now that we've touched on a few little issues, lets move on to some testing.   




 The 3DMark03 CPU test is a convenient way to measure the performance of the CPU for typical 3D usage. The feature tests isolate the performance of key 3D features primarily relating to shader technologies

3DMark 03


  3DMark05 is a premium benchmark for evaluating the latest generation of gaming hardware. It is the first benchmark to require a DirectX9.0 compliant hardware with support for Pixel Shaders 2.0 or higher. Resolution was set to 1024x768.



PCMark04 supports the complete benchmark cycle ? allowing you to benchmark your PC, view the resulting benchmark details, compare your results to those of others, and finally analyze how to improve your PC performance.  PCMark04?s system score is a single, globally recognized number that represents the overall PC performance for home usage.



 AquaMark3 executes a complete state-of-the-art game engine and generates 3D scenes designed to make the same demands on hardware as a modern game. The utilized game engine, the krass Engine, has been used in Aquanox and AquaNox 2: Revelation as well as in the upcoming RTS Spellforce by Phenomic Game Development. AquaMark3 utilizes recent hardware features of the new DirectX 9 API, such as PixelShader 2.0, while staying fully backward compatible to DirectX 8 and 7 graphics hardware.

AquaMark 3

Super Pi

 SuperPi calculates the number Pi in this raw number crunching benchmark. The benchmark is fairly diverse and allows the user to change the number of digits of Pi that can be calculated. In this benchmark we ran SuperPi to 4 million places.


Sisoft Sandra 2005

SiSoft Sandra 2005 is a synthetic benchmark for testing primary components.
The tests can stress a system's CPU, Memory, or Multimedia capabilities.
Sandra 2005 supports X86 (32-bit) and X86-64 (64-bit) hardware and OS's.

Sisoft Sandra

Though the 3500+ Winchester comes out on top in almost every benchmark, the P4 640 proves to be no slouch. When considering this strictly from a price comparison point of view, Intel should be very happy with the results. At stock speeds, the AMD system displays a healthy advantage in almost every test. but with far better overclocking performance, the Intel 640 closes the gap at higher clock speeds. I'm sure with some more exotic cooling the P4 640 would even be a better performer.        

Graphics Testing


3DMark05 is a premium benchmark for evaluating the latest generation of gaming hardware. It is the first benchmark to require a DirectX9.0 compliant hardware with support for Pixel Shaders 2.0 or higher. Resolution was set to 1024x768.

3DMark05 GPU


3DMark03 is a collection of  four 3D gmae based tests.  Each 3DMark03 game test is a real-time rendering of a 3D scenario. It is important to note that these renderings are not merely animations or a set of recorded events; they are designed to function like 3D games work. As with 3D games, all computations are performed in real time. This is a critical part of FutureMarks philosophy of 3D graphics benchmarking.

3DMark03 GPU


 AquaMark3 is a powerful tool to determine reliable information about the gaming performance of a computer system. Again, resolution was set 1024x768.

AquaMark 3

After running all of the benchmarks over and over, and over....I played CS:Source for hours on end with both systems, and noticed no difference at all. I understand that benchmarks don't lie (most of them anyway), but I saw nothing in either systems performance which left me disappointed (though i wish I could as at least I'd have an excuse for being so horrible at gaming). In the end, both systems performed very well, and i would have been happy having either of them under my desk.

Final Thoughts

Antec Titan 550 Server Case

Well, the test results speak for themselves, and strictly from a price point of view, the AMD 3500+ beats the P4 640 but not by the margin one might expect. None of this takes into account multi tasking or 64 bit applications and benchmarks, something we'll cover more in depth with AMD's release of the Venice line and Microsoft's upcoming release of the 64 bit Edition of WinXP Professional.

All testing aside, I actually liked the 640, while maybe not quite the gamer the 3500+ is, the third generation Prescott is showing some of the potential we thought we'd see more than a year ago. First and foremost its nice to finally see 64 bit enabled on a Intel CPU.I also think that we have not yet seen the best that the Prescott core has to offer, but then again, with dual core on the horizon, we may never actually see it.

All in all both processors performed very well, as I said earlier, I've been using both extensively over the past two weeks for my many hours of CS beatings at Nate's hands and really have noticed no differenced in how frequently I die between the two (There goes that excuse). Although i'll be the first to admit to being an AMD "fan boy", I have to say I was happy with the performance the P4 640 gave me.

The Legit Bottom Line

If you are an Intel fan and already have a 5XX series processor, pass on the 6XX series. If you are looking to upgrade to an LGA 775 system and are on a budget, the P4 640 is a great choice. 

While I'm sure the Intel vs AMD debate will rage on, I thopught both CPUs performed very well. In the end, this debate might be decided by supporting components (memory support, chipsets) more than the competitors themselves.