Just to avoid confusing from my post I'll clarify. I've never responded to the SSD vs RAM discussion. Both can be useful depending on your current system config.
I've responded 1) to sir astral's claim that one OCZ can beat 3 or 4 Raptors in Raid0. I've posted numbers that that is nowhere near true, but I'll chalk that up to being overenthousiastic. No harm done. SSDs sure are fast and I like them too. And 2) I've responded to Makerz (or something) that the whole "we need more RAM" argument is the fault of Windows and we should still be able to suffice with 512 MB.
End of twifightDG's quote
I made the claim that it would beat 3 to 5 HDD's. This is not a mistake or a false claim based on enthusiasm. You cannot base who is the baddest drive in town on "Synthetic Benchmarks". Synthetic Benchmarks are helpful in determining which drive is over all faster in a perfect world yes, but it will not give you a realistic measurement against day to day operations, and I will explain why.
Synthetic Benchmarks often push the drive to its maximum speed by utilizing sequential/linear reading and writing. But most of the time a drive does not read in linear & sequential patterns. Most of the time the drive is doing linear and random R/W operations. This is the primary reason for "Defragmenting" HDD's. Getting the data on the drive to be as linear as possible so that performance is improved! But even with that, the chances that the next file you have to read/write being the exact next sector behind the current one is very small. But with at least some defragmentation you can at least read the current file it is working on at maximum speed thus yielding a performance gain. In all truth the more random activity that your drive experiences the less performance it actually has. I have seen HDD drives of all kinds drop to just 10% of their peak performance when the going gets tough. SSD's do not experience this sort of performance decrease. Lets do a little math here to show some numbers. And lets do this with a very nice Velociraptor. Let assume a 3ms latency. Now as we know often times most files are not very large. Typically a file will be just under a megabyte.
At 0.1ms latency the SSD will have that file available 30 times faster than the HDD which is 3.0ms. Now imagine that you have 1000 files to read at 100kb size each for a total of 100megs. An SSD will be able to read all those file in 0.1 seconds where it will take the HDD a whole 3 seconds to read them all. And this will be if the HDD does not suffer additional peformance hits during read. The Average sustained read speed of the Velociraptor is at least 100MB/s in linear operation. The OCZ Vertex is 230~250MB/s sustained read. Obviously the SSD is already avg over 2 times faster in thoroughput in a 1 on 1 senario. Now if we combine 3 Raptors in RAID0 that should be 300 MB/s for them. Now think about this. How expensive are Raptor drives? 3 will easily run you over the cost of a 120gig SSD. Right now newegg charges $159 for a 150gig Raptor until 7/12 in which it will go up to $179. If you buy 3 it will be $477. You can get the 120gig SSD for $385 just about $100 cheaper than 3 Raptors.
Now, in a perfect world you would get 300mb/s because your data files would be in needed read order all the time, which is simply not going to happen. In reality RAID0 drives do not give you full performance in a 1+1+1 senario. You will lose about %20 per drive and will most likely get a 250mb/s average linear read. During random operations the raptor could easily go down to 50mb/s or even lower, while the SSD keeps plugging along at 200mb/s or better. This is where you will see performance being to equal the work of 5 HDD's.
The reviews and benchmarks are in. I have see the results where a single SSD has out performed 8 yes 8 SAS drives all by itself. That is how bad Random IO can hamper a spinning disk's performance.
http://it.anandtech.com/IT/showdoc.aspx?i=3532&p=6
Check out this article. I want you to pay special attention to the section where Random IO test are performed.
A single SSD stomped 8xSAS drives and beat 16xSATA drives in RAID configurations designed to work best with OLTP databases. Now of course people will not be running OLTP databases for gaming and general computing which is why I said you would experience 3 to 5 HDD's in performance for a single SSD. Additionally I would like to mention that the OCZ vertex drives are even faster than the SSD's used in the benchmark in the link up there. Also in that review on the very same page a single SSD still beat 4xHDD's in a sequential read.
If you can read that review and still tell me that my claim of 3 to 5 times performance is due to over enthusiasm then there is no help for you. This test simulated real working senarios. The synthetic benchmarks can only show what a drive can do in perfect conditions. They never show you what you will actually get when the rubber hits the road. Only a test that simulates real world disk activity will show you what you can expect to get from your disk. You can post synthetic numbers all day long, but they will still be just synthetic numbers. What your drive is actually going to do for you is something else.
Ladies and Gentlemen, if you have the funding with the desire to spend it, SSD over a RAM upgrade. The experience is truly something else.