Articles: Storage

Bookmark and Share

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 ]

Performance in WinBench 99

The moment we look at the linear reading graph we see that the new HDD shows excellent results thanks to its higher data density. Note that the line on the diagram doesn’t have any clear-cut steps with long horizontal segments.

The thing is that Seagate used the adaptive formatting technology in its new drive. This technology implies that for each surface-head pair the best algorithm for the most optimal zonal density distribution is selected, which offers the best combination of performance and reliability.

If the parameters of the head and platter are high enough, more sectors can be placed on a single track and the longitudinal density increases. If the head or the platter leaves more to be desired, there is still nothing to worry about: there will be more tracks arranged on this surface.

By the way, this technology gives way to the notion of “N-gigabyte” platters, i.e. the fact that the capacity of each selected platter doesn’t necessarily have to equal one half or one third of the HDD capacity. Because it is quite possible that there not very good heads o both sides of the platter. However, this is nothing you should worry about because you can still reach the desired total drive capacity at the expense of other platter(s). Or in the worst case you might have to make the HDD one grade smaller.

The major advantage of this technology is the fact that you do not have to check the heads parameters similarity before the heads block assembly and this way we can completely eliminate the defective drives from the production increasing the yields to 100%.

So, how could we find out what’s inside the new HDD? How many sectors per track does the new drive have? And what is the difference between the data density on different HDD surfaces?

Since these new HDDs use adaptive formatting, each of them is actually unique. Our drive has the following zone map:

The diagram above shows the real density zone map for each of the read/write heads. We can see very well that this map is unique for all four heads. It is also remarkable that each surface has different number of tracks.

By the way, this is one of the reasons for somewhat unstable results of random access time measurements. Since most tests use not too many requests for random access time measurements (usually 256-2500 requests), the requests get distributed between the platters unevenly.

As for the data recording density, we managed to reach the SPT (SectorPerTrack) of 1120 sectors for two heads. If it hadn’t been for the low spindle rotation speed, Momentus 5400.3 could outperform even Barracuda 7200.7 in the linear read speed.

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 ]


Comments currently: 22
Discussion started: 02/27/06 06:05:16 AM
Latest comment: 06/13/08 01:26:14 AM

View comments

Add your Comment