We use a Sandy Bridge-based system to test hard disks and SSDs. Our test system consists of an Intel Core i5 2500K processor, an Intel DH67BL motherboard and 4GB DDR3-1333 memory. The SSDs are connected as secondary drives and all benchmarks are run in Windows 7.
First we use the Atto Disk Benchmark to measure the SSD transfer rates with data blocks of 4kB and of 1MB. Atto is set at the standard value of "queue depth 4," which means that four tasks are simultaneously sent to the SSD each time. These values resemble most what happens in real life.
The AS SSD benchmark roughly measures the same as the Atto Disk Benchmark, with one important difference. AS SSD uses incompressible data, while Atto's test data can be compressed very well. SandForce-based SSDs therefore don't benefit from the built-in compression tricks within AS SSD. As such, the results are "worst case" for these SSDs.
The next two benchmarks are the most important, in our opinion. PCMark Vantage simulates the hard disk access of real programmes, and indicates what the performance of the drive is in different scenarios. The overall score gives a value for general use and the partial scores indicate the speed for different types of use. PCMark7 does the same, but with new Windows 7-based scenarios. PCMark7 also displays the scenarios in real-time, and performance gains that you wouldn't notice in real-life won't be reflected in PCMark7's scores either. While the differences between SSDs are very small in PCMark7, the newer benchmark is much more realistic for this reason.
Lastly, we added a benchmark that has little practical value, but is fun for those that are really interested in the subject. We measured the maximum 4k read and write IOPS with an Iometer and a queue depth of 32 and incompressible data. These are the values that manufacturers often like to brag about.
We put the 120/128 GB and the 240/256 GB SSDs in separate charts in order to avoid clutter and make comparing easier.