Anand disagrees with you: http://www.anandtech.com/show/4712/t...-with-fw0009/5 There's SF-22xx walking all over the FW009 M4 in his "light workload" bench which just happens to be 26% 4k IO.
Sure, if we only consider the 4k bench the M4 appears quicker, but look at a bigger picture.
It's even worse if we consider the heavy workload: http://www.anandtech.com/show/4712/t...-with-fw0009/4
Reliability? Oh, they stink. Buy the M4 anyway, but know the V3 is 'faster.'
Take a bit of time to learn how to use it. When you know how, you can create a custom test that can make either the M4 or the Vertex 3 look faster based on what type of data they read and write and how many threads they read and write at the same time.
Anand's test, whether he meant to or not, favors the Vertex 3 in both reads and writes.
As-ssd, Crystaldiskmark, Anvil and Atto all give the read advantage to the M4 and all give the write advantage to the Vertex 3. Could I make a benchmark that contradicts either for my own purpose? Obviously.
That's why I prefer to leave the bench tests to the programs that have been out for years and have no vested interest in one brand of ssd over another.