For the benchmarks, the test rig is my standard HDD test center. There is nothing over the top here.
- Asus A8N-SLI Deluxe
- AMD64 X2 4200+
- 2×512 Corsair XMS Pro Series with the flashy lights
- Gforce 6800GT
- Silverstone Zeus 500 watt PSU
- 2 – 74GB Raptors in a RAID0 Configuration.
- Windows XP Pro with Service Pack 2
- Water Cooling.
As with our previous hard drive reviews, the benchmarking will come from various software applications: HD Tach 3 RW, SiSoftware Sandra 2007, FC-Test (thanks X-bit), PCMark 05, Everest Ultimate Edition and a new addition, Iometer. I have opted to omit Winbench 99 because the results were very similar to the NEWER software mentioned. All our benchmarking software is available online so you can try yourself.
Up until this test, the fastest burst speed recording of a single drive belonged to the WD500KS pumping out a whopping 192.37MB/s. It got smoked! As for average read speed, however, the drive performed at the low end of the scale shooting out only 51.2MB/s. The Deskstar also takes the cheese hat in the Random Access times for a single drive (not 10,000 RPM).
Sandra’s benchmarks proved more of the same. The Deskstar drive simply outperforms the offerings from other manufacturers in buffered write speed, but only quietly does so in random write and is middle of the road in sequential write. Read results are very much similar.
The real meat and potatoes here are in the FC tests. FC-Test was designed by Xbit-Labs to facilitate real world testing of hard drives. It performs 3 major categories of testing in 5 different classes. Read, Write, and Copy tests are performed in Install (414 Files, 575MB Total Size), ISO (3 Files, 1600MB Total Size), MP3 (271 Files, 990MB Total Size), Programs (8504 Files, 1380MB Total Size), and Windows (9006 Files, 1060MB Total Size). The application captures the time it takes to do these 3 tasks in each subset, then calculates the speed in which it was completed. This test has proven to be repeatable and expressive of a drives performance, with the exception of the Install test. This time around I finally decided to omit the install test as it proved very buggy producing results from all drives that were unfathomable.