Benchmarking methodology
Tests were conducted on an Athlon system running at 700MHz, using a Gigabyte GA-7IX motherboard with 192MB of RAM. A Deskstar 34GXP with a 20.5GB capacity was attached as the primary master, while a Quantum Fireball KX with a 27.3GB capacity was attached as the primary slave. All tests which required writing to a hard drive used the 34GXP. When testing a CD-ROM/RW drive, that drive was hooked up as the secondary master; there was no secondary slave. Each drive was updated to the latest firmware revision available at the time the testing commenced.

I wish I could say I used Windows 2000 for the benchmarks, as in my opinion the Win9X operating systems rate somewhere between filth and slime. However, much searching failed to find a way to disable the CD-ROM read-ahead cache in Windows 2000; consequently, all tests were run in Windows 98 SE. As you may have guessed, the CD-ROM read-ahead cache was disabled for all tests.

With the exception of the CD Speed 99 tests, each test was run at least twice, and the results averaged. The CD Speed 99 results shown are a representative test run chosen from at least two test runs. If multiple test runs varied significantly, it will be noted when that test result is presented.

File copy
The file copy test involved copying approximately 626MB of data from CD to hard drive. For each drive, the test was performed with three different types of media; a pressed CD, a CD-R and a CD-RW. The CD-R and CD-RW discs were recorded using the Plextor, and were a disc to disc copy of the pressed CD.

The results are informative, to say the least. Although the Samsung is rated at 52X to the new Plextor's 40X, the Samsung wins the pressed CD file copy test by only eight seconds. The 12X Plextor, meanwhile, lags far behind, taking over a minute longer to complete the file copy. Of the three drives, the new Plextor and the Samsung were markedly inconsistent between trials, as multiple trials were separated by up to forty seconds on this test. The 12X Plextor varied by only a second or two between trials.

Both Plextor drives were more consistent on the CD-R test, as all trials on those drives were within a second of each other on this test. They also consistently beat the Samsung, in spite of its 52X rating. The Samsung varied by over thirty seconds between trials, but even at its best was slower than both Plextor drives.

The RW tests show the Plextor drives extending their lead over the Samsung. Interestingly, the only drive not yet affected by inconsistent results (the 12X Plextor) shows them here, with approximately 23 seconds separating multiple trials. The Samsung really takes a beating here, but just as notable is the spectacular performance of the new Plextor drive, which bests its predecessor by over a minute.

The inconsistencies witnessed here imply that real-world performance of optical storage can vary greatly, even when the drive and media remain constant. Inconsistencies in the pressed CD could be explained by varying levels of success in dealing with minor scratches on the disc, but the CD-R and CD-RW tests discount this explanation. Both of the latter discs had no visible scratches, yet all drives exhibited inconsistent performance with at least one of these types of media.