iClaim - Dec/Jan '05

Computing Performance Test Configurations
The what and the why of the computing performance tests.
By Thomas Boedeker and Joseph L. Kashi

Dec/Jan '05 Issue

In “Testing Computing Performance: Part 1,” December/January 2005 Law Office Computing, we compared a mixture of low-, medium- and high-end computer systems. The following are the hardware components we used in our various test configurations and our method for choosing particular components. To the extent possible, we used test configurations that were identical except for a single component, which we varied from system to system. This helped us understand the extent to which a change in the rated performance of the single varied component actually influenced overall system performance.

Central Processing Units. We used 32-bit Athlon-based systems because these CPUs are generally considered to be at least as fast as, if not a little faster than, comparable Intel Pentiums at most routine application programs except when serious multi-tasking was required.

System Boards. Except for the low-end benchmark system, all tests were made on computers using Asus A7N8X Deluxe system boards — these use the nVidia nForce2 chipset and are considered to be the Cadillac of Athlon-based system boards. The newest versions of the A7N8X-E board are optimized for fast, reliable performance. The A7N8X-E boards have two separate local area network connections, including a fast Gigabit Ethernet LAN, a Serial ATA RAID hard disk controller, dual memory channels for better Double Data Rate memory performance, IEEE 1394 FireWire and six universal serial bus 2.0 connections. Selling for under $100, this probably is the best system board on the market in its category and very reasonably priced for its performance and features.

Memory. We used DDR memory because it’s required for Athlons and also because DDR memory is the most common and least expensive memory used today. Many Intel-based systems and the highest-end 64-bit AMD Athlon FX-53 (newest Socket 939 versions) computers also use DDR memory. Except for the lowest-end system, all test configurations used two matched DDR memory modules and the Asus system board’s dual DDR memory channel feature for best possible performance.

Hard Disks. All systems, except the low-end benchmark, used a 10,000 rpm Western Digital Raptor Serial ATA hard disk, one of the best and fastest hard disks available on the general market. The 36.7GB Raptor usually sells over the Internet for the reasonable price of about $120, while the 73GB version costs about twice as much. The 2800+ system used a high-speed RAID 0 disk array composed of two identical 36.7GB Raptor drives combined by the Asus system board’s Serial ATA disk controller into a single 73GB striped disk array — the fastest conventional hard disk arrangement readily affordable.

Operating Systems. All computers used either Windows 2000 Pro Service Pack 4 or Windows XP Pro Service Pack 1a, and the NT file system native to Windows 2000 and XP.

Low-end and high-end reference systems. We used a system built from old parts we had lying around, including a generic MSI KT4V system board with a Via chipset, an AMD Athlon 1800+ CPU, DDR 266 SDRAM and a 5400 rpm IBM UltraWide SCSI hard disk, whose performance is representative of most two- to three-year-old IDE or ATA-33 hard disks.

The highest-end system was equipped with an AMD 3200+ CPU, an Asus A7N8X-E system board, a gigabyte of DDR 433 memory and a 10,000 rpm WD Raptor hard disk.

CD or DVD drive. There was no noticeable difference between a generic 4x DVD-RW device and a name-brand NEC 4x DVD-RW except the generic DVD driver actually burned CD discs about 50 percent faster. There is about a $40 price difference between generic CD-RW and DVD-RW drives, which can read and write both DVD and CD disks. Unless price is an overwhelming consideration, we recommend you get a more versatile DVD-RW drive. The DVD-RW drive probably will create CD disks somewhat faster than a basic CD-RW drive, primarily because its newer electronics can better handle a faster data stream.

Video display adapters. We used three different, standard nVidia chipsets in both name-brand and generic video cards. We found no difference between name-brand video cards and a good generic brand such as Chaintech.

Power Supplies. These are more critical than you might imagine. An inadequate or faulty power supply can cause booting and reliability problems that are superficially indistinguishable from a bad system board, CPU or memory. We used high-end 450 watt to 550 watt power supplies in all configurations. Be sure your system includes extra cooling fans in the system case itself. Faster CPUs, memory and video cards dissipate a lot of heat and are very heat sensitive.

THOMAS BOEDEKER is the city manager of Soldotna, Alaska. He is the former borough attorney for the Kenai Peninsula. For a break from work and computers, Boedeker plays golf and reads mysteries and science fiction. Boedeker has been an active attorney and litigator for more than two decades.

JOSEPH L. KASHI practices law in Soldotna, Alaska. He received his bachelor's and master's degrees from the Massachusetts Institute of Technology and his law degree from Georgetown University. He is admitted to practice before the Alaska Supreme Court, the U.S. Supreme Court and the U.S. Court of Appeals for the 9th Circuit.