tuartboy said:
first, my point was to say that the raptor is 2 year old technology and that there are much better price/performance/size ratios out there. sure, seek times are <5, but apparently the 16 mg buffer does in fact make up much of that performance loss.
Don't forget that lower rpm drives also generate less heat and often have a longer life due to less wear and tear on the drive heads.
Also, I don't know what you do with your mac, but I use the heck out of mine. I started with a 1 gig HD in 1995 (huge at the time) and thought i would never fill it. Then I got a 13 gig. Then a 40. Then an 80. Then a 200. Now I have over 500 Gigs and I am going to need more here pretty soon in my desktop to store next semester's work. I'll admit, I do video and lots and lots of 1 gig + photoshop work, but I don't think I am that abonormal, maybe just faster at filling it.
Article is attached as requested.
Couple of points:
What makes the fast-boot-drive scenario work is that the System/Scratch action is taking place on a different set of heads than the Data action. Having a 500 Gb boot drives implies that you are putting data on the boot drive, which negates this advantage no matter how fast the drive. Simply, the heads of a single drive cannot be in two places at once. And especially on a large 3.5" drive they are going to be spending a lot of time transiting from one location to another between reads and writes.
A 16 Mb cache does exactly nothing for performance on RANDOM reads and writes. The first time the drive is asked to access a sector, it has to transit to that sector and read it. In that process, the cache is loaded with that info. The second time the drive is asked to read the
same, unchanged sector, it can read it from the cache instead, thus saving time. So the effectiveness of a large cache depends entirely on the nature of the access to the information. If the program and OS go back to the well many times for the same data it is effective. If the program and OS are constantly changing data and reading in a random fashion, the cache's effect is minimized and the rotational latency and head transit times will predominate.
One reason the Raptor is faster is that it is a physically smaller platter, meaning the data is closer together and the heads have to travel less than a larger drive. It is instructive to look at the differences in drive performance between the outer tracks and the inner tracks. On the same drive, data throughput will be far worse on the inner tracks. Benchmarks of course use freshly formatted drives on their outer tracks.
IF you elect to use a large hard drive for a boot/scratch drive, I highly recommend partitioning it so that you force the system and scratch files to live in the fastest sectors of the drive (the fastest is the first partition created). By partitioning, you are also restricting the physical area the heads have to seek in.
However, partitioning Boot/Scratch and Data on the same drive does not help the performance when you are alternating between data and scratch files, because although they are on separate partitions, it is the same head that is doing all of the work.
You can test this for yourself. Take a large file or folder, and Duplicate it, time how long it takes. Now take the same data and Copy it to a second drive. A Copy between two internal drives is almost twice as fast as a Duplicate or a Copy between two partitions on the same drive.
The assertation that a 10,000 RPM drive is subject to more head wear is simply incorrect. Because it is a smaller platter, the head actuators actually have less work to do. And if you're thinking that the heads "wear out" from reading the platters at higher speed; the heads never touch the platters. If they did, it would be "game over" for the drive, whether it was 7200 RPM or 10,000 RPM. Consider that 10,000 and 15,000 RPM SCSI drives are routinely used in servers and RAID arrays that run 24/7...
