The floppy boot disk was first used to load up Fdisk.exe and partition the KA into four sections (using FAT32), as detailed in the previous article. After a reboot, Format.com was then used to format each partition. After another reboot, this floppy boot disk was then used to access the SD-M1202 in MS-DOS in order to launch Windows 98's setup program.
The original version of Windows 98 was first installed to completion. Once installation was complete, the Windows 98SE Upgrade was immediately applied. In order to get the >1.44 MB driver updates over to the testbed's Archive partition, Prolific Technology's SMART-Linq USB Data Bridge Cable was installed. This simply involved plugging one end of the cable into a USB port on the testbed, while the other end was plugged into a USB port on my personal system. After the cable was plugged in, Windows recognized and prompted for the driver disk. A reboot later and I had access to every file on my personal system, with file transfers right at 8Mbps.
Once the latest available driver updates were transferred to the Archives partition, they were applied to all relevant hardware per the previous article. In addition, the Adaptec 2940U2W was flashed to the latest bios revision.
Finally, we installed the benchmarks and required applications. After all benchmarks and applications were installed and initialized, each partition was defragmented with Win98SE's native Disk Defragmenter.
Because many of the CD benchmarks out there test the same processes, we sought to avoid redundant benchmarking. Hence, we picked what we felt were the best tests from each benchmark we considered, giving us both accurate and comprehensive results.
The following benchmarks were chosen:
- CD Winbench 99 Version 1.1: Comprehensive set from Ziff-Davis. We use all of their low-level tests (access time, transfer rate and cpu utilization), as well as their application-level test: CD-ROM Winmark 99
- CD Tach 98 Version 2.00: One of the unique measurements of CD Tach 98 is interface burst speed. While CDSpeed99 also measures interface burst speed, it only does so for SCSI drives, while CD Tach 98 accurately measures both SCSI and IDE drives. The other CDTach98 tests were redundant to CDWB99's low-level tests, so we skipped over them.
- CDSpeed99 Version 0.66: This benchmark also measures many of the same processes as CDWB99 and CDT98. We chose its ability to measure digital audio extraction (DAE) in a transfer rate graph across an entire audio CD. CDSpeed99 also tests DAE quality by extracting from the beginning, middle, and end of an audio CD and comparing sectors between the original and extracted audio. Extraction quality is rated from 0 (worst) to 10 (best). Finally, accurate streaming is tested. Basically, with a (theoretically) flawless audio CD, a drive that supports accurate streaming will always give perfect DAE.
- CDDAE99: This program extracts digital audio from audio CDs (often referred to as "audio ripping"), writes in .WAV format to the destination of your choice, and verifies the quality of the extracted audio. We use this program to audibly verify (through our Grado SR60 headphones) the quality of DAE for each tested drive by listening to the .WAV files extracted from our reference audio CD as they play from the Test partition of the hard disk.
- DVDSpeed99: Used in testing DVD-ROM drives, DVDSpeed99 measures the DVD transfer rate, seek times and CPU utilization.
The following applications were chosen to aid in benchmarking:
- Adaptec Easy CD Creator 4.02: We use this program (updated to v4.02 since we reviewed it), along with a stopwatch, to measure the time it takes to burn audio and data CDs to CDR and CDRW discs when evaluating CDR/W drives. Because different CD-recording programs vary in their performance, emphasis is placed on cross-drive comparisons, rather than absolute read/write times.
- InterVideo WinDVD, Version 1.2.99d: This program serves a dual purpose. First, DVDSpeed99 can only benchmark unlocked DVDs. Most DVDs use a form of copy protection which prevents the disc from being read except from DVD players. So prior to testing, the test DVD is started and played briefly through WinDVD to unlock it, allowing us to benchmark with DVDSpeed99. The second purpose is to visually examine DVD playback for smoothness and clarity when testing DVD-ROM drives. This process will be explained in detail in the following article.
Our final optical testbed article, where we will detail the methodologies used in each of the above-mentioned tests!
- Disc copy: We use a stopwatch to time how long it takes to copy a CD-ROM from the tested drive to the Test partition on the testbed's hard drive. The tested CD-ROM is High Heat Baseball 2000, 619 MB in size with 3063 files. Average file size is about 205kb.
- File copy: We wanted to contrast how optical drives read single, large files with how they read smaller files in multiple folders. The largest single file we could find was on the CDTach 98 disc. The file CDTACH2.DAT is 635 MB in size. We use a stopwatch to time how long it takes to copy from the tested drive to the Test partition on the testbed.