Kenwood UCR004010 CD-ROM
When the computer CD-ROM was originally developed and conceived, committees took care to outline specifications that would keep the new format completely compatible with "Red Book" audio CD standards. CD-ROM drives were thus designed to spin at a constant linear velocity of about 176 KB/sec (150 KB/sec after standard overhead is imposed). CD-ROM drives were rather slow beasts, although the hard drives available at the time were not exactly screamers either .
In 1991, NEC Technologies hit on the rather simple idea of a "multispin" CD-ROM. When reading computer data, why not spin the disk at double the rate, and spin back down to normal speed when playing back CD audio? Thus was born NEC's Intersect 83m, the first double-speed CD-ROM to hit the market. I still have fond memories of purchasing the drive with an accompanying Trantor ISA SCSI adapter for the paltry sum of $700.
Other companies eventually jumped on the bandwagon and released double-speed units. Some were SCSI-based, but others were rather inelegant proprietary interfaces. Then, of course, came the NEC Multispin 3X, a standard speed which was quickly ignored and bypassed as companies raced towards a quad-speed plateau, pioneered by Texel (now Plextor). Pushing to each next level while maintaining rapid spin-up and spin-down potential tested clever engineering; soon, however, it became obvious that CD audio's part in the multimedia landscape would be relatively minor. MIDI/event-driven synthesizers and PC-based DACs took over sound functions, freeing the CD-ROM to serve mostly as a data retrieval system. Thus the floodgates opened: 6x, 8x, 10x & 12x drives quickly followed.
A point was reached where it was difficult to push the mechanisms much further. The workaround devised was to keep disks spinning at a constant angular velocity. Formerly, CD drives maintained constant linear velocity operation by spinning the disk faster as the inner (and thus shorter) tracks were read from. If the CD could be spun, say, 4000 to 6000 rpm while reading from the innermost tracks to deliver 8x or 12x performance, surely the same spin rate could be employed in the longer, outermost tracks to deliver a performance boost when reading from those areas. Thus, "16x Max" drives were born, where the CD-ROM more or less maintained the same rotational speed, giving outermost tracks a speed boost. Of course, luck would have it that CDs have always been recorded from the inside out .
Today, we've reached yet another plateau. CAV CD-drives are reaching rotation speeds greater than 8500rpm, allowing an outer track read rate 40 times faster than the original 1x CD Drives. Inner track speed, however, tops out at a less impressive 17x. Heat, noise, & spin-up/spin-down times are all becoming problematic issues when further CD speed increases are pondered. Can it get any faster?
Enter Zen Research, Inc. Zen hit upon an innovative solution: Why not read tracks in parallel? Using a combination of optics and signal processing, Zen's "TrueX" multi-beam technology reads seven CD-ROM tracks in parallel, a great step beyond the conventional single-beam laser used in today's optical mediums. In Zen's mechanism, a single laser is diffracted and split into seven beams, each illuminating seven independent tracks all read in parallel through a multi-beam detector. The parallel data is then ordered into a conventional single bitstream through signal processing.
The first company to incorporate TrueX technology into a product is a relative unknown in the CD-ROM industry: Kenwood Corporation. They've delivered their UCR004010 CD-ROM drive, packaged and marketed by Hi-Val, Inc. This tray-loaded ATA CD-ROM also features a rather hefty 2 megabyte data buffer (all that parallel data has to be stored somewhere before being serialized, I guess) and a quoted random access time of 90 milliseconds or less. The drive features a standard 4-pin CD Audio out in addition to a two-pin SPDIF digital out, for connection to newer sound cards such as the SoundBlaster Live. This unit features "40x40" speed according to Hi-Val, meaning that the drive will sustain 40x (6MB/sec) throughput no matter what part of the disc the unit is reading. The drive is thus a CLV design.
It looks like Zen Research wants to emphasize the marketing aspect of "true 40x" speed rather than delivering a CAV equivalent that may break the "90x max" barrier. Such a beast certainly would live up to the phrase "hard drive speed from a CD-ROM." Though 6 MB/sec is nothing to sneeze at, it still falls a bit short of today's top ATA drives, which can muster a sequential transfer rate of 13+ MB/sec on their outermost tracks.
Hi-Val's manual is quite easy to following, outlining ATA's general operation first before handholding the user through a step-by-step installation. The drive is backed by a one-year warranty.
Installation was pretty painless. Interestingly enough, the Kenwood drive comes pre-jumpered to operate as a master rather than a slave drive. Most ATA CD-ROMs come out of the box preset to "slave." It's a nice touch, perhaps an acknowledgment that the unit should be used on it's own channel, though that's probably impractical for the majority of all-ATA systems.
As soon as I completed benchmarks on the drive, things started to get a bit flaky. Sometimes when powering up the system would pause for a while, attempting to auto-detect the Kenwood on the secondary ATA channel before ultimately failing. Other times, the system would boot up with the CD-ROM fully detected yet would seem to read from the drive very slowly. The same problems sprung up from a second sample immediately. It was only from the third sample that I finally managed to take complete readings (which didn't differ from the figures of the first drive, incidentally) while still having a fully-functional unit leftover to play with. I'm not sure whether I just ran into a bad batch or whether this is a signal of reliability problems to come. Time will tell.
We also decided to retest the Plextor UltraPlex PX-32Tsi, a drive with an outstanding reputation for performance, compatilibity, and reliability. ZD's WinBench 98 CD-ROM tests were run five times in Windows 95 with Supplemental cache settings set to "Large" and read-ahead optimization set to "Quad Speed or Higher."
Ziff-Davis WinBench 98*
|Benchmark||Kenwood UCR004010||Plextor PX-32TSi|
|CD-ROM WinMark 98 (KB/sec)
CD-ROM/Transfer Rate (KB/sec)|
CD-ROM/Access Time (ms)
Transfer Rate (KB/sec)
ZD's overall weighted-average test, the CD-ROM WinMark 98, had the Kenwood edging out the Plextor by 3.5% despite reporting the Kenwood's transfer rate to be greater than 6 MB/sec on both the inner and outer tracks. Audible seeks occur fairly often in the WinMark 98 tests; the benchmark may very well favor drives with lower access times. Despite the Hi-Val claimed access time of "less than 90 milliseconds," WinBench reported access times greater than 100 milliseconds. Perhaps Hi-Val is referring to seek time rather than access time. One of the principal drawbacks of the Kenwood drive's relatively low rotation speed (the drive tops at "12x speeds" and is a CLV design, which implies that the disc may be spinning much more slowly than even that on the outside tracks) is increased latency: Once the read head is in place, the drive must wait for the data in question to pass over it. Random access is no longer considered to be a dominant factor in CD performance; one look at classifications (there are "32x" and "40x" drives, not "90ms" and "80ms" units) shows access times fading into obscurity in favor of sequential transfer rates. Still, the ZD test performs quite a few seeks which is probably the reason for the narrow margin between the Kenwood and the Plextor.
Let's turn now to TestaCD Lab's CD-Tach 98, a more comprehensive CD-benchmarking package. Again, five trials were run in each test. For all tests except 2k read and burst read tests, Supplemental cache settings were kept at "Small" and read-ahead optimization was set to "Quad Speed or Higher." 2k reads and burst tests were conducted with read-ahead optimization set to "None."
CD Tach 98*
|Benchmark||Kenwood UCR004010||Plextor PX-32TSi|
|16k Read Results (KB/sec)|
| 0 MB
| 80 MB
| 161 MB
| 242 MB
| 322 MB
| 402 MB
| 483 MB
| 564 MB
| 644 MB
|2k Read Results (KB/sec)|
| 0 MB
| 80 MB
| 161 MB
| 242 MB
| 322 MB
| 402 MB
| 483 MB
| 564 MB
| 644 MB
|Access Time (ms)|
| Full Stroke
|Burst Speed (KB/sec)|
|Drive Rating (x)
While WinBench 98 samples read speeds at only two locations, the inner and outer tracks, CD-Tach defaults to readings from nine different locations. It's here that Zen Research's accomplishment is truly heralded: All readings from the Kenwood drive with standard 16k reads were in excess of 6 MB/sec. The Plextor unit, on the other hand, shows a more traditional slow start, increasing as the head moves towards the outer tracks. Interestingly, the outermost read for the Plextor drops off a bit, diverging from results reported by WinBench98. Some CD-ROMs apparently slow rotation speed as reads occur at the outer edge of the disc when their read mechanisms are unable to keep up. The Plextor falls into this category.
CD-Tach has a "Drive Rating" system that weights the results from 16k reads at various positions. The "net" speed of the UltraPlex was reported at "21.3x." The Kenwood, on the other hand, weighed in at an astonishing "43.0x." No comparison here! The CPU utilization reported by CD-Tach is quite interesting. Though the UltraPlex and Kenwood both maintained parity at standard read speeds of 4x, 8x, and 12x, CPU utilization at "maximum read speed" varies considerably. Did the SCSI-based Plextor cream the ATA Kenwood? Not at all! The UltraPlex, when transfering at an auto-detected rate of "30x" used up 99% of the CPU's time. The Kenwood drive, on the other hand, at an auto-detected rate of "43x," required only 14%. Hmm
So, WinBench 98's weighted test, which includes a lot of seeking, rates the drives at essentially the same speed. CD-Tach 98's weighted test, on the other hand, includes no seeking and thus rates the Kenwood drive at twice the speed as the Plextor. Where do typical CD-ROM uses, such as application installs, fall? I didn't actually go ahead and install applications with each drive- I wanted to avoid dirtying up our test-bed machine's registry. I decided that I'd perform some simple file copies from the WinBench 98 CD.
The test consisted of copying BIGFILE.DAT from the CD-ROM to the boot drive, a Maxtor DiamondMax 2880. It looks like the file is just a custom-allocation of six-hundred million sequential bytes on the CD. The copy was performed with Supplemental caching set to "Large" and read-ahead optimization set to "Quad-Speed or Higher."
File Copy Test*
|File||Kenwood UCR004010||Plextor PX-32TSi|
Here the Kenwood drive finished the copy in 30% less time than the UltraPlex. This is a bit faster than what one would expect from a "40x max" drive but still not quite the difference that I hoped for. Other informal installation tests (on my own personal machine rather than the Storage Review testbed) using some large games (Ultima Online, Final Fantasy VII) showed approximately the same difference.
An increasingly popular use of CD-ROM drives these days is digital audio extraction, the copying of tracks from audio CDs to popular audio file formats (such as *.WAVs), done entirely in the digital domain. This allows for near-perfect transfers, save only the effects of digital jitter (a nigh-infamous villain in the high-end audio world). The UltraPlex enjoys a stellar reputation of being able to perform DAE at top speeds; the Kenwood drive, featuring the Zen mechanism, has the potential to perform DAE operations with aplomb. How does it do? The CD-Tach 98 CD includes CDDATEST, which tests the DAE speed along with some error correcting ability of various drives.
Digital Audio Extraction*
|Specification||Kenwood UCR004010||Plextor PX-32TSi|
|Drive Capable of Extraction
|Maximum Extraction Speed (KB/sec)
|Drive Reports Alignment Errors
|Auto Alignment Correction
Despite the Kenwood's excellent sequential transfer abilities, the Plextor drive managed a higher average extraction rate. Both drives are capable of reporting alignment errors. Here's an explanation from CD-Tach 98:
"Alignment errors occur when your computer can not keep up with the data stream from the CD-ROM drive. This error results in a pop or click in the captured WAV file. If your drive reports these errors audio capture programs can produce perfect files at near the maximum speed of your drive. If these errors are not reported the program has to assume they exist and read redundant information to avoid them. This drastically slows the audio capture process."
The Plextor, however goes a step further and supports "Auto Alignment Correction," while the Kenwood does not. According to CD-Tach, "this takes all of the error processing burden off your computer," presumably lowering CPU utilization during the process.
According to the Hi-Val box, the Kenwood drive is fully compatible with CD-R media. Even so, reading CD-Rs with the unit I had was a chancy proposition at best. Simply getting a display of a CD-R's root directory sometimes took as long as 30 seconds. Copying from some CD-R's often resulted in errors during the process. There were a couple discs which the drive had no trouble reading, yet they were the exception and not the rule. I noticed no correlation between readibility and brand of CDs or model of CD-R drives used to burn the discs.
As CD rotation speeds ratcheted ever higher, drives became louder and louder. The tray version of the UltraPlex, though a fine drive, can be quite loud. The Kenwood, presumably, would be much quieter since it never hits a speed like that of the UltraPlex, even at its peak. Indeed, the spin of the CD is quieter, though its still audible. Seek noise, on the other hand is quite loud. The drive's read mechanism could clearly be heard racheting back and forth during the random access tests in benchmarks. Another benefit of low rotation speeds is faster spin-up/spin-down times, or, perhaps the elimination of spindown entirely. The Kenwood CD-ROM never seemed to spin down, always being ready to read at a moment's notice. This is evidenced by the rather warm CDs that come out of the drive after they've been in there for a little while; just a note of interest, nothing to be worried about.
All in all, the Kenwood UCR004010 CD-ROM drive is an interesting foray into higher CD performance. As claimed, the drive does maintain a consistent transfer rate from the inner to the outer track. If it were ever possible to get a "40x40" drive such as this as the lowest common denominator of a video standard, the quality would be staggering enough to make us drool . For typical uses such as application and game installations, the drive certainly is fast, but not near the speed increase that many would expect. DAE performance, though still shy of the Plextor, is still among the top when compared to available offerings. There's several drawbacks, though. First and foremost is the inability to read all CD-R discs in a reliable, fast, and consistent basis. Secondly, I had try three times before I came up with a drive that I could play with for an extended period of time- the first two, again, flaked out shortly after installation. Finally, the drive is a rather hefty $199. Plextor's UltraPlex, considered by many to be quite overpriced, can be had complete with a PCI SCSI controller for about the same price. Though it may not be quite as fast in a few areas, its definitely a venerable, well-balanced unit. Keeping the CD-ROM on the SCSI bus also allows a less cluttered setup on the ATA interface. So, it's a lukewarm recommendation that I give to the Kenwood drive. It's the fastest out there, yes, but it has its quirks.
That said, this seems to be just the beginning for Zen Research and multi-track "TrueX" technology. Remember, the Kenwood drive features a relatively low rotation speed. Imagine what would happen if TrueX was married to the high-speed 32x CAV drives on the market today! According to Marie Bahl of Zen, Kenwood plans to release a SCSI drive incorporating TrueX in time for the holidays. Hopefully other manufacturers of CD-ROMS will pick up the technology to offer ever-better performance at more competitive prices. We're also at the dawn of DVD acceptance. Imagine the possibilities of the media if multi-track reading is picked up early in the adoption process. My dream is of a SCSI based Plextor DVD-ROM drive using a Zen mechanism. Mmmm..
Note: All reported test results are the average of five trials.