Sequential throughput should not be what you're looking for. Try comparing 4K IOps instead; closer to reality.
Also, keep in mind all write scores are fake; when writing zeroes. In the same way i can make a floppydrive do 10GB/s; simply write zeroes to it, compress it all to one byte per second; done! The real write speeds are about 100MB/s when writing incompressible data such as movies/music and other binary files.
Perhaps you would want to wait until beginning of 2011 when new generation of SSDs presents itself.
No i am referring to the new 25nm NAND coupled with new controllers (presumably Intel 6Gbps controller) coming in Februari. Perhaps other SSD vendors also come with innovations.
I'm not a particular fan of HSDL; if you want another interface than SATA then native PCI-express would be the preferred choice.
Regarding RAID performance; SSDs already use 'RAID' internally; RAID0 or striping or interleaving. Whatever you want to call it, it has the same performance characteristics as a RAID0-array.
Consider the Intel X25-M 80GB/160GB; it has 10 channels. Virtually all SSDs have multiple channels. Those multiple channels are always used when doing sequential I/O (large files read/write) but sometimes remain unused when doing random I/O with low queue depth. Advanced SSDs use AHCI/NCQ to allow the OS to send multiple requests at once. This can increase random read performance by a factor up to the number of channels - in this case factor 10. That means that random reads would behave just like a 10-disk RAID0.
You can observe this when doing CrystalDiskMark on Intel X25-M with both IDE mode and AHCI mode. You will see that in IDE mode, the 4K-32 read score is about the same as the normal 4K read score; but when using AHCI, the 4K-32 read score is almost 10 times as high as the 4K read score. This is where the 10 parallel channels come into action.
So to answer your question whether RAID does scale performance: yes it does; SSDs already use this internally. You doing the same with software RAID just means more stripes. This will scale random read, random write, sequential read and sequential write. RAID0 does not only scale sequential I/O; that is a myth. Though the quality of software RAID under Windows can be very poor at times; Intel onboard RAID should be decent.
Why you say that lardacus? From all I can see, even with the Raid 5 penalty for IOPs, it is still far greater than the RevoDrive.
Also with the RevoDrive, I know my mobo can boot from it as I am not too sure with the RevoDrive. Its a Striker Extreme II I think (socker 775). Not opened the case for a long time and its currently in transit to my new home for a few more weeks.
With my intended upgrade in the near future to the Asus Rampage III Gene, my SLI config will be a tight squeeze and unless I get a bridge or spacer of sorts, it would be hard to fit the RevoDrive in with the VGA both taking up dual slots (or I water cool the whole system and free up the space occupied by the heat sinks).
But yea, why are you so convinced about the RevoDrive?
Well, I'll give it to him, he's not the only one who's convinced with the Revo, I have been using SSDs for sometime and now after a year after it's release, I do feel that the PCIe based storage has the future in it's hands.
A year ago the prices were maddening, but with the down scaling of the OCZ Fatality to a Revo, it became more affordable.
Me for one, would certainly like to see my OS and programs as good as printed on a PCIe Slot instead of a HDD anymore.
I'd love to leave the HDD for my Movies, Data and Music.
And yes, these drives are bootable so they certainly do take the cake.