Wondering what you kind of performance can expect from the cheapest Athlon 64 processor from AMD? As usual, AMD does not disappoint with its bargain (relatively speaking) high-end chip. Read this review to see how it performs against the 3200+, which is clocked the same but has twice the L2 cache.
I, like many people, always enjoy a tale of a formidable underdog. In the world of computer processors, Advancd Micro Devices, Inc. plays the role of this scrappy protagonist. For many years, AMD played catch-up with main rival Intel by reverse engineering Intel’s x86 processors. AMD’s aim was to offer a competitive product at a competitive price. Eventually, the formidable underdog achieved that goal, and the thorn in Intel’s side has been doing it ever since. However, the tables have turned recently. Intel seems to be taking a page or two or maybe even a couple chapters out of AMD’s playbook.
The two companies’ strategies have diverged in the past, but not nearly as much as they did when the teams at each company decided their respective 64-bit computing strategy. Intel, with its massive R&D budget, decided to go the completely new native 64-bit architecture route a la Itanium; whereas, AMD, with nowhere near its key rival’s deep pockets or financial security, opted to play it a little safer and offer an architecture that basically added an extension to the ubiquitous x86 architecture, called x86-64 (not creative but practical). These were big risks for both companies, but each seems to have attained at least some measurable success in their strategies. However, it probably would not be a stretch to say that Itanium has not been as successful as Intel had hoped, but I do not want to get into that since this review is about one of AMD’s Athlon 64 processors – the Athlon 64 3000+.
AMD’s 64-bit strategy seemed to be quite a bit more broadly scoped than Intel’s. With its Opteron, Athlon 64 FX, and Athlon 64 product lines, AMD has been striving for the proliferation of its AMD64 computing platform in server, high-end workstation, and mainstream and enthusiast consumer markets. The release of the 3000+ was AMD’s first big step in trying to get some momentum in the lower market segments, because when it was released, it was priced at about half the price of the only other mainstream desktop AMD64 chip at the time – the Athlon 64 3200+. At well over $400 retail, the 3200+ just wasn’t really a viable chip to create heavy demand, especially in the mainstream desktop PC market. What was AMD to do? Well, they decided that a version of the 3200+ with half the cache (512KB) would be easier to get good yields on and that it would perform at a performance rating of 3000+. See, that’s where the plan really pays off for AMD. They can effectively take a 3200+ that doesn’t pass its quality tests on its full 1 megabyte of L2 cache and make it a 3000+, as long as at least 512KB of the cache is good. I guess it works out well for everyone – less wasted silicon for AMD and a cheaper 64-bit ready processor for the masses!
Specifications & Features
The table below illustrates some of the key architectural features of AMD’s current Athlon 64 lineup. Note that the only difference between the 3000+ and the 3200+ is that the L2 cache on the 3000+ is half that of the 3200+.
|Processor||Operating Frequency||L1 Cache||L2 Cache||Integrated DDR Memory Controller||Socket||Approximate Retail Price|
|Athlon 64 3000+||2.0 GHz||64KB + 64KB||512KB||72-bit (64-bit interface + 8-bit ECC)||Socket 754||$220|
|Athlon 64 3200+||2.0 GHz||64KB + 64KB||1024KB||72-bit (64-bit interface + 8-bit ECC)||Socket 754||$280|
|Athlon 64 3400+||2.2 GHz||64KB + 64KB||1024KB||72-bit (64-bit interface + 8-bit ECC)||Socket 754||$418|
One of the key performance-enhancing benefits of the entire Athlon 64 family is the integrated DDR memory controller. This is AMD’s engineers’ attempt to reduce some of the bottleneck in getting data to and from the system memory. A testament to their success in improving memory performance can be seen in a couple of my recent reviews (here’s one) where I compared Athlon 64 systems to an Athlon XP system featuring a motherboard with the nForce2 Ultra 400 chipset. Every Athlon 64 system that I have compared to the dual-channel nForce2 system peforms markedly better in synthetic memory benchmarks, even though they are all just single-channel chipsets.
Since Monarch Computer Systems provided the processor for this review, I decided to compare it to the Athlon 64 3200+ in the Hornet Pro small form factor system that Monarch also sent us. I have listed the test system specs below:
Test System Configuration:
- AMD Athlon 64 3000+ or Athlon 64 3200+
- Monarch Hornet Pro featuring the Gigabyte GA-K8VT800M (VIA K8T800) motherboard (review coming soon!)
- Corsair TWINX1024-3200LLPRO (review)
- Reference GeForce FX 5950 Ultra
- Hitachi Deskstar 7K250 120GB Serial ATA 7200RPM Hard Drive w/8MB Buffer
- Pioneer DVD-ROM
- Sony floppy drive
PCMark04 (version 1.0.0)
The 512 kilobytes less of L2 cache does not make a huge difference in this benchmark; however, it does make the most impact on the Memory Score, which seems reasonable.
SiSoftware Sandra (version 2004.10.9.89)
Somehow the 3000+ performed better in a couple of these tests, but the differences are negligible (less than one percent).
3DMark03 (version 3.2.0)
Apparently the amount of L2 cache does not make significant difference in 3DMark03 either. This seems to be the norm, and it’s not much of a surprise, since both these chips utilize the same architectural advantages.
Gun Metal Benchmark 1
More of the same with both AquaMark3 and Gun Metal.
Unreal Tournament 2003 Demo – Antalus
I ran the UT2K3 benchmark with the benchmarking utility from BensCustomCases.com and a custom script file that sets up the maps with 12 bots and maximum detail.
Unreal Tournament 2003 Demo – Asbestos
Unreal Tournament does not get much more exciting than the rest of the benchmarks. Once again, we see that the 3000+ can hold its own against the 3200+, even though it only has half the L2 cache.
When the Athlon 64 3000+ was released, it was a huge bargain at nearly half the price of the 3200+ but still nearly delivering a performance parity. It seemed like a wonderful gift from AMD’s Dresden fab. If you look at AMD’s price list, you will see that the 3200+ is not even near twice the price of the 3000+ now. The price-to-performance ratio is more in line now, but that is really because the 3400+ has taken the top spot in the desktop Athlon 64 line, so now it’s nearly twice as expensive as the cheapest desktop A64, which, of course, is still the 3000+. Does this make the 3000+ seem like any less of a bargain? Well, yes and no. Of course, it does not seem like the bargain it was a couple months ago because the price of the 3200+ has fallen sharply. However, the Athlon 64 3000+ is still a great processor for a little over $200. Although I have not shown it because I don’t have any Intel processors at my disposal, AMD’s Athlon 64 3000+ competes very well with any other processor in the same price range.
Most AMD loyalists may still be having a hard time thinking about spending that much money on an AMD processor because AMD has deservedly gained a reputation of offering the public one great bargain top-performer after another. Once people get used to such a trend, it’s hard to make them think otherwise. Another key issue that could be slowing the adoption of these x86-64 processors is the lack of supporting software. AMD was quite a bit ahead of the pack when it released its AMD64 platform, but people who buy an Athlon 64 processor (or some other AMD64-supporting CPU) could look at it as buying a little bit of obsolescence insurance.