Over the course of years we've generally witnessed an increase in what many consider to be the most important factor when it comes to hard disk performance: spindle speed. There've been some notable exceptions. The biggest came from Quantum in the form of the Bigfoot series. At the time of the Bigfoot's early 1996 launch, ATA drives had reached spindle speeds of 5400 RPM. The Bigfoot, however, took a bit of a step back, reverting to the 3600 RPM spindle speed that's accompanied us throughout the history of the PC hard disk. Quantum argued that a relatively little known parameter, sequential transfer rate, represented an important chunk of net hard disk performance; in this respect, with its increased platter diameters, the Bigfoot didn't lag behind. In addition, Quantum posited that the larger form factor (which burdened much of the blame for the units slow seek times) made the drives much easier to integrate into systems of the time that had surpluses of 5.25" drive bays.
The argument was a persuasive one... the Bigfoot's economical cost and easy integration resulted in large purchases by such industry giants as Compaq and HP. But, as much as Quantum downplayed the importance of access times, it became clear that the Bigfoot family's performance was simply not up to snuff when compared with any other contemporary drive. The final Bigfoot, four generations later, was the "TS" variant. It had ameliorated the situation somewhat by ratcheting spindle speed up to 4000 RPM and shaving average seek times to 10.5 milliseconds. After the TS, the Bigfoot quietly faded away.
For a little while, low-end duties at Quantum were carried by 5400 RPM units such as the Fireball EX and Fireball CR. Eventually, however, such "performance series" 5400 RPM drives were phased out in favor of the "lct," or "low cost technology" line. These drives retained the 5400 RPM spindle speeds featured by their "predecessors," but shaved costs in other areas, such as buffer size (remaining at 512k as competitors advanced to 2 MB) and platter counts. These changes reflected poorly on the drives according to WinBench 99, which showed the lct (and its successor, the lct10) trailing behind offerings from the competition. However, newer methodologies, including the deployment of Intel's IOMeter benchmark, have in fact shown that the lct series is a peppy performer. The relatively low access time displayed by the lct and lct10 power the drives to the top of the 5400 RPM category.
Unfortunately, things are going to change. Quantum has sensed an opportunity with the next iteration of the lct family. These days, increasing attention is being given to the "environmental factors" of drives... ie, noise and heat. Though hardware enthusiasts such as ourselves appreciate such operation, these elements are growing increasingly important in the nascent yet poised-to-explode home entertainment market. Hard disks are going to find themselves in such unsuspecting situations as providing principal storage for the Personal Video Recorder market ("pause, rewind, create instant replays, etc, all on live video feeds!"). These applications demand the utmost in quiet and cool operation. Nobody wants to hear a hard disk grind away while watching TV.
Quantum has thus decided to "optimize" the spindle speed of the next-generation Fireball lct15. Instead of retaining the 5400 RPM spindle speed that's become the mainstay of today's entry-level drives, the company has decided to lower spindle speed (again, a la Bigfoot) to 4400 RPMs. Speaking strictly from a performance standpoint, this increases rotational latency (a component of access time) from 5.6 milliseconds to 6.8ms. Further, Quantum has "optimized" seek times, specifying the lct15 at 12 milliseconds. The family features 15 gigabyte platters and is available in one or two disk configurations for a flagship capacity of 30 gigs. A 512k buffer rounds out the package.
A bit underwhelmed? Frankly, so are we. But let's examine the rationale that went into Quantum's decision. 15 gig platters allow 30 gigs of capacity to be yielded by only two platters. Combined with a spindle speed of only 4400 RPM, the result is a minimization of necessary spindle motor power. In addition to reducing the heat that needs to be dissipated, such lower power levels result in quieter idle noise levels. Similarly, the slower actuator requires less power it its own motor, resulting again in cooler and quieter operation. Finally, these slower, low-power parts cost less than their swifter competition... hopefully this'll yield some rock-bottom prices on the lct15.
"4400 RPM" simply doesn't have a good ring to it and Quantum knows it. They've taken care to refer to the drive as a "sub 7200 RPM" model, disclosing the actual spindle speed in a relative handful of cases. They're fond of the term "optimize." I.E., the drive features "optimized" rather than "reduced" spindle speeds and seek times. They've prepared a lengthy whitepaper arguing that the speed traded off in favor of quieter and cooler operation is well worth it for the typical user. Summarized, Quantum claims that the supposedly significant effect that sequential transfer rates have on overall disk performance combined with buffer hits to cover instances of random access allow users to feel only minimal degradation in performance when dealing with typical tasks such as e-mail, web-browsing, and word processing. In other words, if you don't perform disk intensive tasks, the lct15 won't feel much slower. Uh... yeah. Well... the proof is in the pudding, they say. No, not their pudding... ours!
While Quantum has incorporated benchmarks into their white paper, they're not discerning enough to quantify noticeable differences. Consider the case of Winstone, for example. Measuring disk performance using Winstone yields minimal differences. Before our launch in March 1998, we experimented with Winstone, attempting to correlate it to the differences exhibited by WinBench. Pitting the slowest drive in our possession (then the Fujitsu MPB3064AT) against the fastest (the Seagate Cheetah 4LP) resulted in differences of less than 5%. Few would argue that the Cheetah resulted in a vastly more responsive system. We believe differences in performance are better quantified with WinBench and especially IOMeter. So, without further ado, let's get started.
WB99/Win2k Low-Level Measurements