Hard drive performance issues




I'm currently in the market for a new hard drive between 2.1 and 4.3Gb (or $250 to $350). I've been shopping around and found that the Quantum Bigfoot is cheaper than most. However, I also noticed it is bigger and slower than most. I intend to run both Linux and Win 95. I mainly use it for uni (Linux), word processing and games.
Is the Quantum Bigfoot a good buy? Will I notice any difference between the Quantum Bigfoot at <14ms, WD Caviar at <12ms and Quantum Fireball at <10ms? How about rotational speed (RPM)? Will I notice any difference between 5200 and 4500?
- Huy


There was a time when the one reliable measure of hard disk performance was the average seek time. This is the 14ms, 12ms and 10ms you quoted for each drive. Like nearly every other piece of computer equipment, the many and varied techniques used to increase performance complicate things.
Data access time is how long it takes for the read/write head to get to the desired location. Two measures (among others) associated with this are seek time and rotational latency.
Seek time is the amount of time taken to move the read/write head between tracks (the average time is given).
Having moved to the correct track, the read/write head still has to wait for the correct sector to rotate underneath it. This is the rotational latency, and it is related to how fast the hard drive spins. When hard drives were all a standard 3600rpm (with an average rotational latency of 8.3ms) every drive had the same latency. Faster RPMs decrease this latency (a 4500rpm drive will have a latency of 6.7ms). Techniques such as cylinder skewing and head skewing increase the likelihood of the next needed sector being directly positioned under the read/write head when switching between tracks, and have also reduced this latency.
Measures of data access time used to provide an accurate picture of performance. However most hard drives now have read and write caches allowing faster data transfer to be achieved, and this makes data access time statistics very misleading. More appropriate are measures such as data transfer rate -- how quickly data can be moved between the drive and CPU once the read/write head is positioned correctly --and data throughput rate, which combines data access time and data transfer rate.
There are also a number of benchmarks available for hard drives. The better ones are those that give an indication of drive performance under real world conditions such as BAPCo Benchmark and QBench. The trick is getting manufacturers and retailers to quote these figures.
All this aside, the question remains, how will the speed of your hard drive affect you? Well, as usual it depends. If you don't work with large files or large databases, and if you don't open and close programs very frequently, and if you don't switch between programs all the time, and if you have sufficient RAM (32Mb recommended), then the speed of your hard drive will not affect your productivity or the performance of the majority of games. Also, it will help you to afford that extra meg or two of space, which you will probably fill up before you realise it.
- Roy Chambers


Category: Hardware
Issue: Oct 1997
Pages: 158

These Web pages are produced by Australian PC World © 1997 IDG Communications