Related Articles

Benchmarks

So how does a USB-HDD compare to a HDD that is hooked up directly to a IDE controller? We take a quick peak with a small set of benchmarks.

Benchmarks

Athlon 3400+
Soltek K8AN2E-GR NForce 3 250Gb
2 x 256MB OCZ DDR400
Maxtor DiamondMax Plus 9 80GB ATA133 HD
Windows XP SP1
NForce Unified Drivers 4.27 (with Nvidia's IDE drivers)
ATI RADEON X800 XT

HD Tach 3

Some interesting data garnered from HD Tach here. For the most part, the limiting factor for the USB enclosure is not the hard drive itself but the transfer speed of USB2. At the inner most track, transfer speed finally has an effect and drops slightly in both cases. Remember the theoretical bandwidth of USB2 is 480Mbps or 60MB/s. The SilverRiver was the only USB2 device and tops out at roughly 32.6 MB/s for the entire read range. The Maxtor on the IDE connection started off at 60MB/s and it decreased slowly as the read head moved towards the inner track.

CPU usage on the USB bus is signficantly higher than that of the IDE at roughly 10% compared to 2%. According to HD Tach, random access time did not vary much on either the IDE or USB2 bus.

PCMark 04

PCMark highlights why synthetic benchmarks are a nice supplementary tool to real world usage but probably not as the only source of information - PCMark has XP Startup times as being faster on the USB2 drive which will not likely be the case. File copying noticeably suffers, giving up roughly 40% to against the same drive hooked up directly into the IDE controller.

Sisoft Sandra 04 Disk Test

SiSoft Sandra reports simliar performace to what we saw in HD Tach especially on the Sequntial Read front. Again, we note that the SilverRiver does not exceed the 30-odd MB/s ceiling that we observed on HD Tach. Performance of sequential writes on the IDE front seems a lot higher compared to read speeds. We do not see a huge variance between reading and writing on the USB bus meaning that the upstream and downstream bandwidth are not asyncronous.

Conclusions

As far as I am concerned, the only way to physically move large amounts of data is with an external hard disk as nothing beats the convenience or speed. Thermaltake has put together a nice package in the SilverRiver allowing the end user to mount a hard disk or an optical drive if they choose. The construction of the casing is very sturdy. The connectors inside seemed a tad finicky on our unit as we had a couple problems that were resolved with the reseating of the internal IDE cable. Once it was reseated, the unit was problem free.

Benchmarks do not tell the entire story and this is especially true in the case of the SilverRiver. Reading and writing to an external hard drive do not feel slow so some of the benchmark numbers may be a bit misleading. Throughput of roughly 30MB/s did not happen back in the ATA33 days and the SilverRiver pared with the Maxtor never felt like it was laboring unlike some older hard drives that I have used.

For anyone looking for storage that cannot be satiated from a USB flash drive, the SilverRiver is a good a place to start as any when looking for an external enclosure.