Test Setup
Testing on this card was performed in a number of ways. Unlike most of our normal reviews a considerable amount of time was spent using this card under real world usage This time was originally going to be considerable less, but instead it turned into an almost four month period of time. So the break down of testing will include an extended section of real world experience in it.
Hardware
System 1 – Sandy Bridge
| Processor | Intel i7 2700k |
| Motherboard | Asus Maximus IV Extreme |
| Ram | 8gb G.Skill RipjawsX 2133mhz |
| Graphics Card | Gigabyte GTX 670 OC |
| CPU Cooler | Noctuai NH-D14 |
| Power Supply | Antec High Current Pro 750w |
| OS | Windows 7 Ultimate 64bit sp1 |
System 2 – Ivy Bridge
| Processor | Intel i7 3770k |
| Motherboard | Biostar Z77 |
| Ram | 8gb Dominator Platinum 2400mhz |
| Graphics Card | Gigabyte GTX 670 OC |
| CPU Cooler | Noctuai NH-D14 |
| Power Supply | Antec High Current Pro 750w |
| OS | Windows 7 Ultimate 64bit sp1 |
Hard Drives
- Intel 330 series sata6 60GB SSD x5
- Western Digitial VelociRaptor sata6 300GB 10k rpm HDD x2
Test Software
| ATTO Disk Benchmark | Queue4 (stock setting), Queue 10(heavy use of on drive cache) |
| Crystal Disk64 | Small drive/ssd test, heavily geared towards small files |
| AS SSD | Small drive/ssd test, heavily geared towards small files |
Notes about testing
These controllers offer a large amount customization. With a large amount of time and knowledge about baseline applications you can optimize your arrays for just about any test. Being as we don’t have years of time to tweak only a few variations will be ran. That being said this controller is primarily aimed at spindled disks. which shows quite a bit in the testing. This is not abnormal, in part due to the diminishing returns shown by writing to a large array. While this may seem extremely bad when it comes to performance of some of the single SSD’s we have done on this site. I assure its not out of place.
The test that really speaks the most for this card is ATTO Disk Benchmark. Its the closest to heavy demand network usage as you will see in a server environment Desktop applications outside of professional imaging programs generally are not written to take advantage of of this type of controller. Causing it to reflect poorly in the SSD benchmarks. This is mostly due to the intermediate level of caching that takes place in the card. As well as the single threaded nature of these benchmarks that don’t make heavy use of on drive or controller based caching.
