Originally Published: Monday, 15 May 2000 Author: Brian Richardson
Published to: enhance_articles_hardware/Hardware Reviews Page: 1/1 - [Printable]

The Great Camino Benchmark: Part One

Ever since Intel decided to use RAMBUS technology on their newest chipset lines, a great debate has raged over the actual performance of the technology. Intel's latest PentiumII/III chipsets - "Camino" (i820) and "Carmel" (i840) - utilize what has been advertised as the next great leap in memory technology. But most early independent reviews & benchmarks make RAMBUS seem like an expensive failure.

   Page 1 of 1  

The Setup

Ever since Intel decided to use RAMBUS technology on their newest chipset lines, a great debate has raged over the actual performance of the technology. Intel's latest PentiumII/III chipsets - "Camino" (i820) and "Carmel" (i840) - utilize what has been advertised as the next great leap in memory technology. But most early independent reviews & benchmarks make RAMBUS seem like an expensive failure.

My personal opinion is that RAMBUS, in it's current form, isn't ready for prime-time. Memory companies cannot get the hang of making the complicated modules, chipsets don't properly handle the memory, and current processor/software technology doesn't have the ability to properly utilize 1.6 GB/sec of bandwidth. Despite all of this, Intel's i820 chipset is on the market.

So that brings the next question ... what about LINUX? LINUX isn't the first choice of the magazine reviewer for evaluating a new platform. Why? It usually boils down to three points:

  1. Most folks use Windows
  2. LINUX may not support all features of the hardware
  3. Benchmarking for LINUX isn't as consistent as benchmarking Windows

So the benefits/problems with new technology don't always get represented for LINUX users in a standard review. Word-of-mouth and discussion within the LINUX community generally reveals the issues the average Windows user gets in a traditional review & benchmark.

The Platforms

That last point is the one I'll try to address in the next few articles. I was lucky enough to receive a SuperMicro PIII-SCA motherboard last week. This motherboard is based on the Intel i820 chipset, and has the unique advantage of utilizing both RAMBUS RIMM slots & SDRAM DIMM slots. What does this mean? It means I can setup a single system, and properly test the performance difference between RAMBUS & good'ol SDRAM by merely swapping the memory modules. Plus, as an added bonus, I can compare this to an Intel 440BX based motherboard (the AMI S781). This should reveal how LINUX handles RAMBUS & SDRAM, along with a comparison of the performance between the i820 & 440BX.

The Problem

So how does one benchmark a LINUX system? I read some of the HOW-TOs and lurked around on the web looking for "the one and only LINUX benchmark suite" ... it doesn't seem to exist. What I did find were a myriad of tests & programs for benchmarking various aspects of LINUX. Using some trial and error (mostly error), I intend to evaluate these systems with what I feel are the "viable" benchmarks.

Of course, I forsee several problems when I reveal my results:

  1. "My distribution's better than your distribution!" ... I'll be flamed into an alternate dimension by the folks who thought I should have used distro X instead of distro Y. The distribution(s) I used for this test were selected for various reasons, which I will elaborate on at the right time.
  2. "Why did you use that program? It's worthless!" ... Again, the flames of angry readers will be fanned under my chair when I reveal the programs I used to make my determinations. Well, this happens in the Windows world, too ... I'm ready for it.
  3. "You ran Windows benchmarks! You bastard-whore-traitor!" ... Please attempt to keep your profanity to a minimum when you find out I actually ran benchmarks under Windows 98SE. I have my reasons ... they will be revealed in due time.

The Plan

Okay, here's the idea ... in my next installment, I'll walk you through my experiences installing LINUX on this new SuperMicro motherboard. I will also delve into the methodology I'll use for benchmarking.

Part three (the final installment) will show the benchmarking results, and the conclusions of the benchmarking experiment. I won't limit my findings to the i820 vs. BX battle ... opinions about LINUX benchmarking will also be revealed. It's guaranteed to be fun for the whole family ...

As Brian Richardson braces for the impact of opinionated LINUX e-mail messages against his fragile little skeleton, he toils at the task of exterminating software. In his spare time, Brian is developing a style of karate for computer geeks that utilizes their strongest muscle ... the right index finger. He is thinking of calling it "click-fu".





   Page 1 of 1