Looking to Really Kick Up Photoshop Performance? SSD

Question

You've saved up a bit of cash since the holidays, right?
Good sale seen today:  OCZ Vector (their latest model, and one of the fastest on the planet) 256GB for well under $1 a gigabyte:
http://www.newegg.com/Product/Product.aspx?Item=20-227-916
http://www.amazon.com/OCZ-Technology-2-5-Inch-Warranty--VTR1-25SAT3-256G/dp/B00A9YQPCY
$229.99 and free shipping.
Limit 5. 
Most modern motherboards can easily support a RAID array using the onboard Intel ICH chipset, and the latest even sport SATA III. 
For $460, two of these drives on SATA III would make a 512GB RAID 0 array that easily breaks the 1 gigabyte/second mark and could hold your OS, applications, scratch, and some data.  You could make whatever spinning HDDs you currently have secondary storage, where you keep data you're not actively working on.
From personal experience I can say this: 
Running a system from a RAID array of SSDs as compared to an HDD is like moving ahead 5 to 10 years in one step. 
Warning:  Use a system so-equipped and you'll never be able to stand using a system running from spinning HDDs again.
-Noel

Answer 1

I personally have been running four 480GB SSDs in RAID 0 since last April.  Yes, everything's pointed to drive C:.  It not only works, it screams; I think of it as "living the SSD dream".  This system is lightning fast and NOTHING I've been able to think of bogs it down.
There are two main reasons why people generally advise putting scratch files, caches, etc. on separate SPINNING hard drives:
1.  In order to avoid thrashing, where multiple processes read/write simultaneously and cause the heads to seek all over the place.  Trouble is, seeking operations, while quick, are VERY slow by comparison to reading and writing data, and a disk that's seeking isn't reading and writing much.
2.  In order to use multiple SATA links and drive controllers simultaneously, in order to increase throughput.
RAID 0 (striping) actually accomplishes item 2 very nicely, by spreading the I/O requests from all reads/writes across multiple links.  The more drives you have in a RAID array, the better the performance, generally speaking.  There are some practical limits, but they're so high as to make other parts of the computer the bottleneck.
And SSDs avoid thrashing entirely, as there's no physical seeking and no latency.  To give you an idea, a single spinning hard drive can sustain around 100 megabytes/second throughput for a sequential read or write operation - that's one big file, one big read or write operation.  But if you start doing real world reads and writes - which are generally small - the things slow down horrendously.  Most folks looking to optimize data storage systems use 4K byte read/write operations as the most meaningful benchmark.
Here are benchmark results showing how much data my system reads/writes with a Hitachi internal drive (used for backup storage).  This is the Passmark Advanced Disk "Workstation" benchmark, intended to simulate a mix of real world operations on a Windows workstation.
The key thing to note in the diagram above is that the actual throughput is around 2 megabytes per second!
Now, for comparison, here's the same benchmark run on my SSD array.  Note that it's almost 60 times faster.  That's a difference you feel.
In the past year or two SSDs have really arrived as mainstream storage, eliminating the problems with "wearing out" flash cells by implementing complex wear-leveling controllers, and by just offering much larger capacities.  Even though I use my system quite heavily 24/7, if I were to write 2 terabytes a day (which is about 20x what I actually do write) my SSDs will only begin to reach their design lifetime limits in 10 years. 
By then we'll have crystolic fusion drives. 
-Noel