There's something I've been wanting someone to do, but I don't have the extra hardware lying around. And it could be useful (as well as satisfy a curiosity).
Lately a lot of push for computers these days is to build with the fastest "secondary memory" available if you can afford it, and then dump whatever you can onto it. The reasoning for this is because it decreases load times. And while the performance benchmarks show that indeed, things like SSDs and partitioning off RAM show that their speeds are loads and loads better than the traditional HDD, what does it really do for the system? My half-hearted research lead me to hobbyists doing this and posting results, but I haven't seen any professionals doing this.
For instance, a lot of people are pushing for bigger SSDs because they want to install their games on them. However, I tried this out with one of the bigger games I have in my collection: Skyrim (with about 10GB of mods stacked on top of it). Granted, it was a symlink to the hard drive following a symlink back to the SSD (which I don't think would matter), but I didn't notice any appreciable decrease in load times. And this is a Samsung 830 SSD which should be at least three-four times faster than my WD Caviar Black 2TB.
Obviously this means that there's something going on in the background. Not that I want you guys to figure out what that is, but more to the point: how much performance do you really get by going to faster secondary memory? I believe a bulk of loading, assuming the storage has nothing else to do, isn't really transferring files from itself to RAM, but initializing everything for the program itself, which also suggests the CPU is involved in "load times" as well.
What I want to see is a comparison of real world applications loading (from the time I say "go run/load this" to "I can do something with it") from an HDD, an SSD, and a RAM drive, from a variety of processors, and see what happens as a result.