Brian Hickman said:
For anyone interested in performance and battery drain difference between different RPM drives, read this:
http://www.tomshardware.com/mobile/20031031/index.html
Hickman
I hate to burst anyone's bubble but there is a huge problem with these tests - the driver buffer is 4x larger on the 5400 & 7200 RPM drives. This obfuscates the results somewhat, so we really don't know if the speed gain (claimed of up to only 15%) was from a faster spindle alone or from the drive cache. Unfortunately there wasn't a proper control done; all three drives are different sizes, manufacturers, two different interfaces, different number of read heads, and areal density.
A proper experiment would involve the same drive mechanics (all XGB, from the same manufacturer with the same areal density and on the same interface, with the same cache size, same number of read heads) on all three drives with spindle speed being the only difference. You can actually see the problems with the test if you look at the various results.
Digital Video:
The 7200 barely beats the 5400 drive - granted this is probably only partially a test of the HD, but for 33% more RPM, I would expect more than a paltry 2.4% speed increase. By comparison the 7200 RPM drive gave an 18.3% increase over the 4200 RPM drive, but has a 71.4% faster spindle speed!
Winbench 99:
This test is actually more interesting because it gives us a comparison factor for what a larger cache can do. Granted the desktop comparison drive is old, so a newer model would likely perform better, but the 3.5" 7200 vs the 2.5" 7200 (2MB cache vs 8MB cache) provides a 15.4% speed gain. By comparison the note book hard drives give 26.2% (5400 v 7200), 86.5% (4200 v 7200), and 47.6% (5400 v 4200). A note of caution here is that synthetic benchmarks like Winbench
rarely translate directly into real world performance. Seeing as how the cache difference alone accounts for the general 15% speed up claimed by the article, I hope you start to see why it's failing apart.
System Book WinXP:
Again we find less than predicted gains. 4200 v 7200 gives us only a 21.7% increase - almost all likely to be attributable to disk cache. The 5400 v 7200 gives us only a 4.3% gain, and 5400 v 4200 gives us 16.7% - again within reach of disk cache alone.
Sysmark 2002:
Overall results only show an 8.6% (7200 v 4200), 3.2% (5400 v 4200) or 5.2% (7200 v 5400) gain. This test probably shows more than anything that for most people the difference won't be spectacular while your doing your ordinary work. If your work entails the HD more than these tests did you would likely see better results.
Battery Impact:
Here we see another flaw that is exposed due to different sources for each drive. The Hitachi 7K60 is rated at similar power usage when active (using about .1W more during read/write and .2W more during seek). However hard disks are often in idle/standby mode MUCh more often than seek/read/write. Here the Hitachi drive shows it's superiority by using .12W less in idle, and .11W less in standby. [Note: The Hitachi drive mentions an "Active Idle" mode which I presume is an intermediate between activity and idle, the Seagate doesn't mention this statistic. The Hitachi also mentions a Sleep mode that uses only .1W and would thus be superior to the Seagate for extended non-usage of the HD]. The 40GN by comparison does have the lowest power usage of them all, using nearly half a watt less power in active modes, and .2W less in idle than the Hitachi drive. [NOTE: Hitachi actually made the Travelstar drives for IBM, thus this is likely why they are similar in these modes].
Finally, the drive caches on the Seagate and Hitachi drives almost certainly allow these drives to operate in their lower power modes more frequently, thus giving them better power usage characteristics than you would otherwise expect. I bet if these drives carried 2MB caches instead of 8MB, you'd see much higher power usage (and thus more significant loss of running time).
Bottomline:
This review does provide useful information, however the conclusion is flawed.
1) The drives tested are just too different from each other. A completely impartial test would have tried to fix as many characteristics between the various drives as possible. In this case far too many things that could affect the test significantly were different between the drives.
2) There was no mention of the obfuscating effect of cache size. This is shown primarily in the Winbench 99 synthetic benchmark where an older 3.5" 7200RPM drive with 2MB of cache is trounced by the 2.5" 7200RPM drive with 8MB of cache. Unfortunately this comparison is also imperfect, but at least here something is fixed between the compared drives.
3) Often the tests didn't show a huge margin of difference between the drives (the 5400 v 7200 often showed <10% difference) and the huge margins often could be just as attributable to the large cache of the 5400 & 7200 models.
I don't dispute that drives with faster spindle speeds can't produce better results - they can. But this review if nothing else actually points out that there are other avenues to speed gains than simply spindle speed. The data could also extrapolate that cache size has a very definite affect on performance, and if you look at the results with this eye, you can see that going from 2MB to 8MB often showed a large improvement, while 5400 RPM to 7200RPM showed a small one. Sometimes with tests like this what your looking for is what you see rather than what is actually there. The 5400 & 7200 RPM drives perform better because they are built with newer technology, not just because they spin faster!
So for those of you looking to get new hard drives for your laptops, I would recommend looking at what size disk cache they offer first, then what spindle speed they offer. The larger disk cache enables the drive to read-ahead more data when requested, and to store data for writes when the drive head has to seek. Both of these allow for better performance [less waiting on disk writes) & battery life [less reading directly from the drive means more time in lower power usage states]. If you find you really do need that extra 5-10% above what a larger cache offers, then go for the faster RPM drives. But don't expect a speed boost equivalent to the difference in RPM, I'm afraid that you simply won't get one in most cases.