Mac Musings

Drive Matters

Daniel Knight - 2006.06.14

As country music fans are hearing on the radio, size matters. In a clever twist on the usual meaning, Joe Nichols sings about the size of one's heart, "big ol' kisses that go on and on and never end", and filling her world with laughter. Size matters.

In the world of computers, size matters, too, especially when it comes to hard drives. How many gigabytes do you want or need?

But speed matters, too, and the quickest way to get a handle on speed is by asking how fast the drive spins. After all, a 7200 rpm drive has to move data faster than a 5400 rpm drive - or does it?

It's Not That Simple

If you've been following the story over on Macworld about different benchmark performance between the 2.0 GHz white MacBook and the 2.0 GHz black one, you might be scratching your head. Except for the hard drive, the two 'Books were configured identically, yet some Speedmark results were quite different.

The MacBooks, like most computers made today, can ship with drives from any of a number of vendors, and Apple has made it easy to swap the hard drive in the MacBook. By slipping a 7200 rpm drive into the black MacBook, it came very close to matching the 5400 rpm drive in the white MacBook.

What's up with that?

My first thought was that perhaps one of the drives had a bigger cache than the other one. I've seen a big difference between 7200 rpm drives with 2 MB and 8 MB caches with my eMacs.

But as I read further, I saw that both drives had 8 MB caches, yet the 5400 rpm Seagate was edging out the 7200 rpm Fujitsu in several tests.

Digging Deeper

In the world of notebook computers, 4200 rpm drives used to be the norm, and the first 5400 rpm drives provided a huge speed boost (they spin 28.5% faster). Jumping to 7200 rpm spun drives 1/3 faster than that, but that's not the whole story.

Another way to make drives more responsive is to add a buffer. In the ancient days of personal computing, drives simply read their data and dumped it over a bus to the computer as fast as they read it. Old timers may recall setting the interleave for drives used with the Mac Plus, SE, and II - and wonder why we no longer do that.

Setting interleave slowed down data transfer to a speed these old computers could handle, and then someone came up with the clever idea of building a data buffer (or cache) into the hard drive. Now the drive could read the data at full speed, store it in cache memory, and feed it to the computer at the top speed it could receive data.

And that's why one of my favorite Mac SE speedups is putting in an hard drive with a buffer to replace the ancient bufferless drives Apple used back in 1987.

And the bigger the buffer, the more data in the cache, the more efficient the process.

On top of this, there are different schemes for storing data in the buffer. A drive may look ahead a few sectors, or it may buffer the directory, or it may buffer writes, all of which can make things more efficient.

Physics

That's drive electronics, and every brand has their own way of doing things. So you can see why a Seagate, Hitachi, or Fujitsu mechanism might offer different performance despite having identical spin rates and buffer sizes.

But there's a third factor beyond RPM and cache size - data density. Looking at the manufacturer's specs for 5400 and 7200 rpm 2.5" Seagate Momentus and Hitachi Travelstar drives, you'll see that the transfer rate of the 7200 rpm drives is less than 1/3 faster than the 5400 rpm drives - just 9% faster for the Seagate, and a more impressive 28% faster for the Hitachi.

The big breakthrough over the past year or so is perpendicular recording, which can easily store 50% more data per track. Just how much data is there per track? That's hard to know, in part because the manufacturers don't promote the number and also because the amount of data stored per track varies between the inner and outer tracks of a drive platter.

Still, if Drive A spins at 5400 rpm and stores 40% more data per track then 7200 rpm Drive B, it will be able to read more date per second. That might help explain how a 5400 rpm drive can sometimes outperform a 7200 rpm one.

Conclusions

Tech Report has an extensive article comparing these four drives. It looks at peak transfer rate, how well each drive handles multiple data streams, and typical read and write speeds. And just as Macworld discovered with the MacBook and both brands of drives, Tech Report says, "At times, those performance differences were surprising, as we certainly didn't expect to see Seagate's Momentus 5400.2 beating Hitachi's Travelstar 7K100 in so many tests."

Various drive parameters can be tweaked, optimizing a drive for use by a single user or in a server, even optimizing for the way an operating system works and how many data streams it may work with at any given time.

In the end, there's more to picking a hard drive than capacity, rotation speed, and buffer size. Seagate, Toshiba, Hitachi, and Fujitsu each have their own buffering algorithms and optimization schemes, so you'll want to look for benchmark results using the hardware and operating system you'll be using. (OS X on Intel Macs may produce very different results than OS X on a G4 or G5 system.)

Price is the final factor (after all, Low End Mac is primarily about value). If ultimate performance isn't crucial and the 5400 rpm drive is a lot less costly, it may well be a better choice for you. But in the end you have to come up with your own value equation - how important is capacity vs. performance vs. price?