Originally Published: Thursday, 17 August 2000 Author: Brian Richardson
Published to: enhance_articles_hardware/Hardware Reviews Page: 1/1 - [Printable]

Duron vs. Athlon

Given a few hours of free time and access to new AMD hardware, Brian had the chance to test drive the new Duron & Athlon processors. Of course, you're dying to know how they compare.

   Page 1 of 1  

Well, I was in the right place at the right time. Given a few hours of free time and access to some wonderful new AMD hardware, I had the chance to test drive the new AMD Duron and AMD Athlon processors. More importantly, I examined the performance differences between these two CPUs. Of course, you're dying to know how they performed.


AMD's latest revisions of the successful K7 "Athlon" processor are on the market. Both the AMD Duron and AMD Athlon appear in the "Socket A" format, abandoning the "Slot A" architecture. Both are 0.18 micron chips with integrated Level Two cache running at processor speed. The Duron is the "economy" processor, using only 64KB of Level Two cache. The current Athlon in Socket A form-factor, code-named "Thunderbird", uses 256KB Level Two cache. From a marketing standpoint, the Duron is designed to compete with Intel's Celeron and the Athlon goes head-to-head with Intel's Pentium III.

At the moment, motherboard support for the Socket A processor family is limited. The only chipset currently supporting Socket A is the VIA KT133 ... AMD, SiS and others will release chipsets soon. My tests were run on the MSI 6330. In my opinion, this is a very capable motherboard ... stable, well documented, full of features. But this isn't a motherboard review, so let's get to the benchmarks.

Duron vs. Athlon

The following system configuration was used for benchmarking:

  • MSI 6330 Motherboard (Socket A, VIA KT133 Chipset)
  • 64 MB PC100 SDRAM (CAS 2)
  • ATI Rage Pro 2X AGP Video Card (8 MB)
  • Intel EtherExpress PRO/100 Network Card
  • Seagate 8.4 GB Hard Disk Drive (IDE/ATAPI, 5400 RPM)
  • Acer 50X CD-ROM Drive (IDE/ATAPI)

Both processors were running at 700 MHz. I used BYTEmark v2; and UNIX Bench 4.0.1, compiled using Mandrake's default gcc compiler. It's important to note these scores vary a bit each time the benchmark is run, so don't worry about +/- 0.001 score variances ... use these numbers wisely. The first score listed is for the Duron, the second score for the Athlon (Duron vs. Athlon).

BYTEmark* LINUX/UNIX v2 (10/95)

  • Memory Index - 3.838 vs. 3.817
  • Integer Index - 3.462 vs. 3.481
  • Floating Point Index - 7.725 vs. 7.985

UNIX Bench 4.0.1 (index scores)

  • Arithmetic Test - 91.7 vs. 91.7
  • Dhrystone 2 - 140.4 vs. 140.7
  • Excel Throughput - 174.9 vs. 298.3
  • File Copy (1024/2000) - 213.4 vs. 236.8
  • File Copy (256/500) - 315.2 vs. 269.2
  • File Copy (4096/8000) - 28.4 vs. 28.4
  • Pipe Throughput - 372.8 vs. 371.9
  • Process Creation - 490.8 vs. 524.8
  • Shell Scripts - 173.8 vs. 185.7
  • System Call Overhead - 258 vs. 289.7
  • Final Score - 179.8 vs. 192.9


I guess it's easy to tell which benchmarks rely on cache size. Despite the inexplicable file copy test where the Duron came out ahead (I ran it twice and got similar numbers both times), the AMD Athlon is the winner. But a close inspection of the benchmark numbers shows that the Athlon is not that much faster than it's bargain counterpart. Heavy 3D applications will definitely benefit more from the larger L2 cache, but most any other program will show little difference in performance. This makes the Duron an ideal choice for the "average" LINUX platform, including file serving duties.

I'm sure most readers are drooling over the 900+ MHz AMD Athlons, but can't afford their entry price. No problem, I have a solution. Buy a Socket A motherboard and an AMD Duron 700 MHz ... then upgrade to the AMD Athlon 900-100 MHz CPUs when the faster models drive prices down (sometime around November). You can sell that Duron to a friend and pick up some Christmas cash, or use the money to buy extra PC133 memory.

Either way, I don't think you'll go wrong buying any of these Socket A processors.

Brian Richardson waits in eager anticipation for his DSL installation. After eight months of badgering the local telephone monopoly, high-speed bandwidth is days from his grasp. Readers can expect Brian's productivity to be inversely proportional to number of UT games he plays per day.

   Page 1 of 1