SATA III - performance

[Carol Haynes]

Just installed two new hard disks in my machine (they have been sitting waiting for a while so they were purchased before the price hike):

The two drives are:

WD10002FAEX - 1Tb Caviar Black SATA III
WD20EARX - 2Tb Carviar Green SATA III

I am a bit confused at the performance of these two drives:

I have both SATA II and SATA III interfaces on the motherboard but I had only one socket left on the SATA III interface so I connected the 2Tb green drive on there and the 1Tb black drive to a SATA II socket.

I expected the SATA III connected drive to be faster - even though the drive is not quite so fast because the SATA II interface should limit the speed of the Black drive.

Oddly the Black drive is still faster than the green drive even though it is connected to a SATA II interface.

Any ideas? Somewhat strange result as far as I can see.

[Ath]

Mechanical disks aren't yet up to par to fill the bandwidth of a SATA II interface, so neither that of a SATA III interface, only SSD's can come close to that kind of data volume, and then only the fastest. The WD Black model has 7200 rpm specced, but the Green model I couldn't find definitive rpm specs for, so most likely it's either variable between 4200 and 5600, or down to a constant 5400 rpm, resulting in a lower overall data transfer speed.

[Carol Haynes]

Ah - I hadn't seen the 'IntelliSpeed' aspect of the Green drives. Whatever that actually means (WD seem remarkably quiet on the issue).

[f0dder]

RPM isn't the only thing that affects drive speed - a 5400rpm with higher data density could outperform a 7400rpm drive with lower data density. Now, there's probably more platters in the 2TB drive carol just ordered, but if both drives had the same number of platters, the 1TB drive would have a higher data density than the 2TB one

As for SATA II vs III, it indeed doesn't really matter for mechanical drives - SATA-II has a theoretical peak of 300MB/s. It matters with some of the really big & spendy SSDs, or if you're connecting a a RAID rack to an e-Sata port

[Carol Haynes]

Thanks - apart from marketing hype it does make you wonder what the point of SATA III is really when it comes to mechanical drives.

[Ath]

Well, they're at least tested and compatible with that type of connection

[4wd]

Thanks - apart from marketing hype it does make you wonder what the point of SATA III is really when it comes to mechanical drives.

You can be happy knowing that the 100kB text file you're loading from the HDD internal buffer is happening at 6Gb/s instead of 3Gb/s...that's assuming it's in the buffer of course. 

[Carol Haynes]

Actually yesterday I copied over 500Gb of data from one drive to another (both SATA III drives on the SATA III interface). Was rather unimpressed with an average transfer rate of around 35Mb/s (IIRC)

[4wd]

Actually yesterday I copied over 500Gb of data from one drive to another (both SATA III drives on the SATA III interface). Was rather unimpressed with an average transfer rate of around 35Mb/s (IIRC)

Wow, that's less than half the speed I get copying large files between a WD Green 750GB SATA2, (on a SATA2 i/face), and Samsung 1TB SATA2, (on a SATA3 i/face).  Going the other way is about the same as you get mainly because the 750GB is severely fragmented.

[Carol Haynes]

It was an estimate of the average - to start with transfer was running at about 80Mb/s but over such a long copy process it slowed considerably (or at least seemed to). The final rate was under 20Mb/s.

I suppose it could just be the way Windows 7 reports transfer rates - I would have to do more controlled and timed experiments to work out the actual rates.

[f0dder]

Was that for a zillion small files, or a few large, Carol? And were any of the drives fragmented?

Even near the last sectors, you shouldn't have dropped all the way down to 20MB/s if you're doing large sequential non-fragmented transfers.

[Carol Haynes]

I was moving a large Steam installation to a new drive - so a mixture of large and small files. Not fragmented though at all.