Vertex 3 specs and benchmarks

[Wendy]

What's crazy is in most of the real-world benches, (i.e. Storage Bench) the V3 120's right at the top....

I don't think it's crazy at all.
For to many years 4K random performance has always been said to be the most important. While it most certainly is for servers, it isn't for a desktop, as the usage pattern is totally different.

For a desktop sequential performance is more important when you get into the real world. I noticed this way back when the first Indilinx SSDs came out. It couldn't compete with the Intel G1/G2 with 4K random performance, yet when you run real world tests the Indilinx drive was faster. It has taken a long time to prove this.

Anand uses the Anandtech storage bench, and although I don't know how the measurements are taken, the tests run "real apps" with "real data"/

Over at MyCE.com I have access to a bus analyzer and some proprietary software and trace filters. I to run scripted real applications and data sets in the "MyCE Reality Suite", and for a desktop usage pattern, the SSDs with better sequential performance always do better in these tests.

I can also see how many small file random transactions take place while running these tests, and from the data gathered so far, any SSD with > 5MB/s 4K random performance will be able to cope with these tests without any problem.

Of course, an SSD with very high 4K random performance will take less time to process these random transactions than an SSD with low 4K performance, but the transaction times are so small that it makes near to no difference until you start loading the system with medium to high multitasking, and I believe that is the reason why the Vertex 3, and other SSDs with high sequential performance do so well in a real world desktop PC.

To sum up.
Times have changed, and I believe that relying totally on 4K random performance to determine the overall speed of an SSD in a desktop, is flawed.