External usb 3.0 drives

gwalec

Distinguished
Apr 17, 2010
26
0
18,530
Hello
In the market for a external usb 3.0 hard drive. Been doing some research and found the LaCie Minimus 2TB drive. Looking for storage and perforamnce to use as a back up. I was also looking at the WD My passport ultraportable. Like that its powered from a usb cable. However, found that LaCie uses burshed aluminium casing instead of plastic for better heat dissipation which i think is necessary for 7200 rpm drives???? Also like the added storage LAcie drive offers. Dont really care it its ultraportable or not as it will be a desktop drive. I was wondering what you guys think? Any other options i might be overlooking? recomendations? I have a gigabyte x58A ud3r mobo and was wondering if anyone has had any or read about any issues with usb 3.0 on that mobo? Any help would be greatly appreciated. Cheers!
 
Your drive's performance will be limited by USB 3.0, so there is no advantage in a 7200 RPM model. IMHO a 5400 RPM model would be preferable. That said, perhaps the aluminium case will handle the additional heat, as you say.

I don't like USB powered hard drives because they can exceed the port's current rating during spinup, but in your case you may be OK. Gigabyte's current motherboards have "3x USB Power Boost":

http://community.wdc.com/t5/My-Passport-for-PC/Gigabyte-motherboard-3x-USB-Power-Boost/m-p/155004#M4946
 
Eh? USB 3.0 has a bandwidth of over 600MByte/sec, so it won't be a bottleneck for a faster hard drive. A 7200 RPM drive with the same platter density as a 5x00 RPM drive will be able to transfer data significantly faster even if it's connected via USB 3.0.

If the question was about USB 2.0 then I'd agree that the speed of the drive wouldn't make any difference, since real-world USB 2.0 transfer rates are limited to less than 40MByte/sec.
 
I can't speak for that review, but I do notice that it's over a year old and for all I know the performance they saw could have been due to bottlenecks in the drive rather than in the USB connection.

What I can tell you is that the USB 3.0 standard uses a 5Gbit/sec raw transmission rate and is said to be capable of real-world transfer rates of over 400MByte/sec.

I regularly see transfer rates of 110MByte/sec (that's a raw rate of well over 900Mbit/sec when you consider protocol overhead) with my external drives. My personal experience is that faster drives perform faster on USB 3.0 than slower drives do, and that the performance is comparable (with 10%) to the performance of the same drive connected to an internal SATA port. The USB connection itself is not the bottleneck.
 
I only have USB 2.0 enclosures, so I'm relying on reports in the various forums. Most people seem to be achieving in the vicinity of 60 MB/s.

That said, I did find a marketing claim by Seagate to the effect that its BlackArmor PS110 USB 3.0 portable EHD is capable of a sustained transfer rate of 100MB/s.

http://www.seagate.com/ww/v/index.jsp?locale=en-US&name=BlackArmor_PS_110_USB_3.0&vgnextoid=68c0ddea732b5210VgnVCM1000001a48090aRCRD

The package includes a 500GB 7200RPM 2.5inch drive.

Just for comparison, here is a benchmark which shows that a ST9500420AS drive has a max sustained data transfer rate of 100MB/s:
http://www.legitreviews.com/images/reviews/967/72004_hdtune_read.jpg

This would tend to confirm that the USB 3.0 interface is not necessarily a bottleneck, as you say.

That said, have you noticed whether there is a limit to the burst rate, and whether or not this is dependent on the drive? That should tell you the real world maximum for your particular configuration.

 
The stats I see regularly are based on Windows internal performance counters which have an update interval of a second - so short bursts to and from the hard drive cache don't really show up on it.

This benchmark of an Intel SSD indicates that USB 3.0 is certainly capable of transfer rates up to around 200MByte/sec. That's faster by a fair margin than a 7200rpm hard drive - so there's no doubt in my mind that it would run faster over a USB 3.0 connection than a 5400rpm drive would.
 
Thanks for the corrections.

ISTM that the actual performance depends very much on the USB host controller and the USB-SATA bridge. Maybe the early hardware just wasn't up to it?

One thing that still surprises me about those results is the eSATA performance.

http://cdn5.tweaktown.com/content/3/0/x3004_04.png.pagespeed.ic.szKmUMb8ad.png

A score of 120MB/s is in the 1.5Gbps range. I would have thought that eSATA was essentially a direct connection, even when the data passes through a bridge.
 
The physical media protocol used by SATA uses 8/10 encoding, so 1.5Gbit/sec on the wire means 150MByte/sec of actual data throughput (less protocol overhead, although with SATA that's pretty small for large block transfers).

Also, very few disks spin fast enough or have the data recorded densely enough to saturate a 1.5Gbit/sec SATA connection - so when a drive tests at 120MByte/sec the limiting factor is almost certainly NOT the SATA connection.
 
I was referring to the review of the 3Gbps SSD which was benchmarked at over 210MB/s under USB 3.0. Its eSATA performance was about half that.

I would have thought that the setup should have been able to exceed 150 MB/s, assuming that a 3Gbps connection had been successfully negotiated. If a bridge chip can acquire the data, packetise it, and translate the source and destination protocols, at 210MB/s, then surely it would be no big ask for any bridge IC to pass on the data, untouched, straight through to the host, at even faster speeds. AISI, it should be operating just like a switch. I know these data rates aren't achieved in practice, so I find it quite puzzling.
 
There's very little physical difference between SATA and eSATA aside from the different signalling levels required to extend the allowable cable length. My guess would be that the low performance was some sort of issue with the external enclosure they used for the test.