Depending upon the controller you stick them on of course, it's not just sequential speeds that get a significant boost from R0... ...though i accept that "appreciable difference to real world performance" could mean "to you" &/or the boost in responsiveness is less noticeable for reads in adding a 2nd SSD than it is going from a HDD to a single SSD (though write speeds improvements continue to be noticeable providing you're reading from a fast drive/writing from memory). So, for myself (in Win7 throughout), the difference between 1x 120GB V Turbo & 2x 120GB V Turbos was noticeable - adding in a lsi 9260-8i (improving the controller from the ich9r) made another noticeable improvement - & moving from 2x 120GB V Turbos to 4x 50/60GB V2s again made things noticeably faster & more responsive. As to trim, whilst for some drives it's kind of important (though, as examples, the C300s can be comparatively slow to recover speeds & certainly there was the recommendation for the indilinx drives (the V Turbo being one of) that either there was regular S3 idle time for Garbage Collection to run or a manual cleaning process), OCZ's pre-release testing apparently showed the SF based drives are as good with & without trim enabled on a single drive d.t. their version of GC... (it's not actually GC, but SF won't release details - & this 'should' apply to all SF based SSDs) Then, what you've completely ignored is that building a R0 array allows someone to upgrade in stages - rather than having to have many hundreds of pounds on day one to buy some huge SSD, they can buy smaller drives when the money's there... ...okay, this doesn't work 'if' someone bought a small C300 with that foolish HighPoint card that bittech recommends to go with it (since, yet again, it can't do raid & would bandwidth limit both of the SSDs since it's only 1x pcie), but more generally... [Edit] Having been doing other things, it's just crossed my mind that, not forgetting things like the z-drive & revodrive, *at least* some of the later JMicron & subsequent indilinx drives have used dual internal controllers in R0 to get the performance they did - 'if', as you claimed, R0 provided "almost zero appreciable difference to real world performance" then these would have been no better than any other drives which isn't the case. Mmmmm... thinking about it & taking it all a stage further then, as an example, since the ONFI 2.1 spec for nand allows between 166 & 200MB/s, doesn't that mean that the Marvell controller in a C300 must be accessing data from different NAND at the same time (basically an internal version of R0) in order to reach advertised speeds??? - which would then indicate that most/all(?) previous nand based SSDs would have needed to do likewise... Now, assuming i've not gotten the wrong end of the stick about this latter point somehow (& accepting that adding a DDR cache has helped things), this would mean that this internal version of R0 is the only reason why we have the SSD speeds that we actually do atm...
