Comments on: Laptop Hard Drive Decision - which one is best for my needs?

Question: Laptop Hard Drive Decision - which one is best for my needs?

nokry56 — Mon, 04 Aug 2008 15:50:01 -0800

Which HD would be a better (read: faster) drive for my soon to buy MBP?

After Apple releases their new Mac Book Pros I will be purchasing a high-end 15" one. Will be purchasing 3rd party memory (4GB) and am thinking about buying a 3rd party hard drive to replace the 200GB that comes default. I'm torn between this more expensive 320GB 7200RPM drive, this slightly less expensive 320GB 7200RPM drive, or this 500GB 5400RPM drive.

I was reading online that the Seagate 500GB drive would actually be faster than a 7200RPM drive because of its platter size or something? Could anyone explain how this could be? Would the 500GB drive be faster than the other two drives I linked to?

I will be using the MBP for media storage, audio editing, and gaming, so want the largest drive possible that won't be a bottleneck for performance.

As a side question, how would I get the HD installed without voiding my warranty?

By: sonic meat machine

sonic meat machine — Mon, 04 Aug 2008 15:55:46 -0800

The less expensive 320GB will be about the same speed as the 500GB; the more expensive one will be faster.

By: kbanas

kbanas — Mon, 04 Aug 2008 16:31:43 -0800

I don't know. I can't speak to how platter size may or may not play a role in things, but I can't imagine a scenario where I would prefer a 5,400 RPM drive to a 7,200 RPM. I can imagine scenarios where, to access a particular file located on a particular part of the disk, they might clock in similarly, but I have to imagine that the law of averages takes hold, and over the long term the 7,200 RPM drive will provide you overall faster data access... It might also eat battery a little faster, but feh to that.

When I bought my new MBP, I bought the smaller 160GB disk because it was 7,200 RPM, versus the 200GB option, which was 5,400 RPM. If I could've, I would have gone for a 10,000 RPM model at half that storage capacity! I work doing IT support with laptops all the time, and I have convinced myself that I find 7,200 RPM drives to make much snappier, more responsive systems.

I could be full of shit, but I like to think that I'm not.

By: Class Goat

Class Goat — Mon, 04 Aug 2008 16:37:49 -0800

You might be surprised to learn that the platter rotation speed is not the most important specification in terms of subjective snappiness of performance. The average drive seek time is overwhelmingly the most important spec.

By: traco

traco — Mon, 04 Aug 2008 16:39:14 -0800

I don't know if this is still true or not, but my understanding is that all other things being equal, more RPMs means more energy consumption. Something to consider for a laptop.

By: nokry56

nokry56 — Mon, 04 Aug 2008 16:40:28 -0800

@sonic meat machine
Could you please explain why that is?

Thanks for all answer so far. Keep 'em coming...

By: traco

traco — Mon, 04 Aug 2008 16:52:24 -0800

After contemplating, I believe the 500GB might be faster because the data is more densely stored on the platter.. thus the platters don't have to spin as far to get the same amount of data.

By: nokry56

nokry56 — Mon, 04 Aug 2008 17:03:59 -0800

@class goat

the 500GB does have a very high seek time. i'll take that into consideration, thanks.

By: BryanPayne

BryanPayne — Mon, 04 Aug 2008 17:10:18 -0800

The 500GB drive is a 12.5mm and does not fit in current model 15" MBP's. Only the 17"

7200RPM drives will take a small hit on your battery. Most people claim around a 10% reduction.

By: majick

majick — Mon, 04 Aug 2008 17:17:51 -0800

The WD is probably the fastest of the bunch by a shade, but it's a Western Digital -- I can't in good conscience suggest anyone buy one. Owning a WD is never "better" for any meaning of the term than owning something else.

The Hitachi would probably beat them both in a flat out sequential read, but disk access patterns aren't like that any more. Unless your work consists of cloning hard drives day in and day out, sequential transfer rate is of secondary performance benefit at best. Average seek time is king.

By: nokry56

nokry56 — Mon, 04 Aug 2008 17:21:13 -0800

@BryanPayne
Thanks for noticing. Meant to link to this

By: BryanPayne

BryanPayne — Mon, 04 Aug 2008 17:33:09 -0800

That drive is sold as a kit, but yes, should work fine. However, I've personally never been a fan of Samsung drives. I've always thought they should stick to making televisions.

By: Good Brain

Good Brain — Mon, 04 Aug 2008 18:34:34 -0800

Most single-user disk access is pretty random. Average seek time is the measure of how fast random accesses occur. The biggest contribuing variable to seek times is rotational latency (the amount of time until the required data is avaliable under the disk head. Spindle speed has a direct impact on rotational latency. Hence. 7,200 rpm drive will be better for most uses than a 5,400 rpm drive, the downsides being power consumption, noise and heat.

For sequential acesss on larger files, like a video clip, transfer rate, rather than seek time, is the limiting factor. Transfer rate is influenced both by data density, and rotational speed. For a given disk form factor and rotational speed, a higher capacity drive will tend to have a higher transfer rate.

By: flabdablet

flabdablet — Mon, 04 Aug 2008 19:22:26 -0800

On the other hand, I've always found Samsung drives to be quick, quiet and reliable. Clearly, your mileage may vary.

By: sophist

sophist — Mon, 04 Aug 2008 23:41:24 -0800

Just FYI, this is the series of HDs that ships with current MBPs. The seek read time is 12ms, just like all the other drives listed, even though it is a 5400 rpm drive.

By: flabdablet

flabdablet — Tue, 05 Aug 2008 20:07:17 -0800

Average rotational latency for a 5400RPM drive will be 0.5 revolutions / 5400 revolutions/minute * 60000 ms/minute = 5.6ms. For a 7200RPM drive, it would be 0.5 / 7200 * 60000 = 4.2ms. These latencies must be added to the average seek time (something of the order of 10ms) to get the average random access time; so a 33% increase in spin rate translates to something like a 10% improvement in average random access time for two otherwise identical drives.

But as Class Goat hints, most drives are not otherwise identical. You really need to look at their average seek times, which have everything to do with head positioning and nothing to do with rotational speed.

Seagate quotes 11ms read-seek and 13ms write-seek for the ST9320421AS; the others you linked to are both 12ms read-seek (the extra time on write seeks is to give the head positioner a little more time to stop ringing so you don't get wobbly data blocks). So the Seagate wins on seek as well as on latency: its total random access time will be about 11 + 4.2 = 15ms, compared to 12 + 5.6 = 18ms for the 500GB Samsung. This 13% performance improvement will cost you about 5% more power than the Samsung uses. Looks like a clear winner.

But wait... these average seek times are built on the assumption that the drive is full and that data could be anywhere. If you were to make a 320GB partition on the 500GB Samsung drive, limiting its effective capacity to the same as that of the Seagate, you're now only using 64% of the available disk surface and the average seek time will drop. It won't drop all the way to 7.6ms, because drives run at roughly constant areal data density, not constant per-track data density) but I wouldn't be surprised to see it achieve about 9ms and bring the average access time down to 15ms, same as the Seagate.

But of course you could do the same partitioning trick with the Seagate and drag extra performance out of that as well. And this is probably what you should do, if you're in pursuit of Ultimate Snappy Drive Response. Get the Seagate, make a relatively small (maybe 40GB?) main partition at the start of the drive (constant areal density means blocks on the outer tracks will have faster transfer rates) and install your OS, your software and most of your user files on that. It will fly. Then put your huge media files (which tend not to need much hard drive speed, just loads of space) in a second partition that fills the rest of the drive.

Incidentally, since most drive power consumption occurs during seeks, this seek-minimizing partitioning strategy will also minimize power consumption. I would not be surprised to find that a carefully partitioned ST9320421AS used less power than a stock drive with a single large partition.

By: nokry56

nokry56 — Tue, 05 Aug 2008 21:31:20 -0800

Thanks so much flabdablet & all.

Flabdablet, your answer finally cleared my head on how to interpret all of the hard drive specs.

I'm currently thinking I'll probably go with the 500GB Samsung with a 60GB OS X partition (adobe apps, ableton, & WoW), a 80 GB BootCamp partition (games), and 360 GB data partition. Or I might go with the Seagate and decrease my data partition by 1/2

Any thoughts on how much of a speed difference I'd see with the Samsung vs. the Seagate?

By: flabdablet

flabdablet — Tue, 05 Aug 2008 23:01:15 -0800

Based on these benchmarks which, although they don't test the specific 7200RPM Seagate drive you're considering, do yield quite impressive numbers for its predecessors, I'd expect the Seagate to be quite noticeably quicker.

Samsung is very strong in media research and development. They're probably better than anybody else at actually packing bits onto the surface of disks, which is why their 5400RPM drive can achieve data transfer rates that other manufacturers need to go to 7200RPM drives. But Seagate has been in the disk drive business for rather longer, and I think is a bit better at making drives whose ancillary mechanicals and electronics best balance the performance available from their media.