kernelthread.com

PowerBook Drives: Theoretical Comparison

Disk Drive Options

Apple updated the PowerBook line in September 2003, with one change being an optional ($125 extra) 5400 rpm hard disk drive for the 17" and 15" PowerBooks (the stock drive is 4200 rpm). Moreover, even 7200 rpm drives are available for purchase from parties other than Apple. This page is a theoretical comparison of a few common drives for the PB17 (including the ones used by Apple). Please realize that I do not have access to any of these drives: I have simply gone through the technical specifications of these disks, and am presenting an annotated summary on this page.

All drives mentioned have the same buffer size (8 MB). They do get heavier with increasing rotational speed, however, the difference between the lightest and the heaviest is only 16 grams (0.0353 lbs), which is approximately 0.5% of the PowerBook 17's weight (6.9 lbs).

As I mentioned earlier, this is only a theoretical comparison. The manufacturer quoted performance numbers have approximately same ratios as the rotational speeds. Thus, the 5400 drive has a 28% faster maximum data transfer rate (over the 4200), while the 7200 is 48% faster than the 4200 with respect to the same parameter.

Understandably, power dissipation increases with higher rotational speeds. This can be correlated to battery life as follows (say, 5400 rpm 5K80 vs. 4200 rpm 2080AT).

The PB17 has a 58 watt-hour battery. Assume that with a stock 4200 rpm 2080 AT, you get 3.5 hours of battery life. Further assume that you are reading/writing all the time (rather contrived worst case). Let us use the average of seek, read and write power dissipations as the effective power dissipation. This would be (2.2 + 2.0 + 2.1) / 3, or 2.1 Watts. The same parameter for the 5K80 would be (2.6 + 2.5 + 2.5) / 3, or 2.53 Watts. Now, assuming exactly same usage scenario, we have:

( 58 Watt-Hour ) / ( X Watts + 2.1 Watts) = 3.5 Hours ... (1) ( 58 Watt-Hour ) / ( X Watts + 2.5 Watts) = Y Hours ... (2) From these two equations, Y = 3.42 Hours approximately

Thus, a worst case reduction in battery life is about (3.5 - 3.42) = .08 Hours, which is slightly less than 5 minutes. It can even be argued that the faster throughput / lower latency might reduce power consumption in other components, thereby offsetting the disk's increased dissipation. In any case, the battery life would be reduced by a few (4 - 5) minutes, if at all.

There also would be the issue of greater localized temperature increase in case of the higher RPM drive (due to higher wattage). The temperature increase would depend on the efficiency of the cooling mechanism. Taking into account the theoretical worst case again (disk reading/writing all the time with .4 Watts extra dissipation throughout), you might see a temperature increase of 1 to 2 degrees (very crude approximation).

Again, as expected, the faster RPM drives are louder. The 7200 RPM drive has up to 25% higher decibel value than the 4200 one, while the 5400 RPM drive has up to 13% higher decibel value. While this may sound like a lot, consider that the absolute difference is 6-7 decibels (7200 vs. 4200). For comparison, a loud radio in an average house, or a noisy restaurant, or roadside traffic ... are at around 70 decibels. A soft whisper might be 20 decibels. A pin dropping is 10 decibels. I know that doesn't help much. However, consider the fact that except in carefully controlled laboratory experiments, a one-decibel change cannot be perceived. Then again, many of us are used to working in a quiet enough environment that may be similar to the "controlled lab" scenario.

Shock/vibration tolerance values are similar across the drives. In any case, they should hopefully never need to be put to test :-)

Conclusion

Apart from performance/cost trade-offs, other issues that often keep coming up (battery life reduction, noise, heat etc.) are of little or no consequence. If cost is no bar, then it does seem to make sense to get Apple's $125 upgrade. If cost is an issue, then Apple's upgrade might not be the most effective way of getting disk performance.

However, the noise factor might still be debatable. The fact that different people have different tolerance for noise makes this subjective.