Phase-change memory:

Scientists from IBM, Macronix and Qimonda said they developed a material that made "phase-change" memory 500 to 1,000 times faster than the commonly-used "flash" memory, while using half as much power.

"You can do a lot of things with this phase-change memory that you can't do with flash," IBM senior manager of nanoscale science Spike Narayan told AFP.

"You can replace disks, do instant-on computers, or carry your own fancy computer application in your hand. It would complement smaller technology if manufacturers wanted to conjure things up."

The day's not far off that you'll be able to carry an unimaginable amount of knowledge around in your pocket.

Shadowrun Credsticks.

I've occaisionally toyed with the possibilities if my computer had 1000 times more memory (1000 gigabytes), which operated at a reasonable speed. One fun idea was to generate 3D scenes from every conceivable viewpoint, and then you can navigate your viewpoint through the scene with no calculation and no delay at all. Of course, you have to generate the views from every viewpoint, which might be time consuming...

Oddly enough, I just read an article on cell phones of the future that made them sound like Shadowrun commlinks.

My humongous, fast RAM fantasy involves a huge database running persistent Prolog instead of SQL.

The Singularity approaches.

Well, I can imagine a lot of things... :)

With electronics, things really do get smaller, faster, AND cheaper over time. Thus you don't see people hoarding electronic gadgets. Of course, that field is a looong way from its basic physical limitations. This is unlike rocketry where the physics and chemistry are well known, and the trick is in cost-effective engineering management and clever business plans. But I do enjoy the challenge.

So, if the faster memory is coming, will that go hand in hand with forgetting things faster too? A good Blue Screen of Death will appear exoseconds sooner?

Make that nanoseconds.....stupid keyboard....

:)

Has anybody else noticed that most databases are entirely too small to need all the machinery associated with SQL and disk access (B trees, multiple index strategies, and what not)? For most of the stuff I see done, it would make more sense to just load the whole DB in machine memory for really fast searches and access, probably with write-through to disk to deal with persistence in case of power failure. I keep a contact list on my computer called my "farley file" (named after the same thing in Heinlein's Double Star--did he invent that name or did he get it from some where?) It's not sorted, it's not carefully comma separated and indexed, and it's fast as all get-out. I just open the file and search for a name, or company, or whatever info I happened to enter along with the contact.

I can't remember the name off hand, but there is at least one DB business out there based on precisely that idea — loading the entire DB in to memory for speed.

Considering that I started out with a 40Meg 5-1/4 full height (as my first mass storage; my first computer, with 2K of ram never did a reliable save to tape) the currently available 1gig flash sticks (which are only 20 bucks if you get them with googletools loaded) have some residual unimaginability.

Leading to such obvious jokes as: "Do you have an unimaginable amount of knowledge in your pocket, or are you just happy to see me?"

I had a couple of test machines for a point of sale I worked on that had the database loaded into memory using the old ramdrive command. Also if you google around you can find pci ramdrives that have 4-8-16 slot sdram expansions with a controller that implements a mountable partition for you automatically. They simply use an on board battery to keep the memory persistant between power cycles. Certainly not for serious business usage but nifty to bring up from time to time to show off ones uberness.

*rolls 20d6 for geek proficiency check*