Taking on Heidi's Challenge (Future of Computing)

[Glynn R Harris]

I'm newly registered in here, enticed to the forum by Heidi Monson's comments in the newsletter-- she asked what we thought would be the future of computing. While my crystal ball is rebooting (three minute wait), I thought I'd start a thread for fun...
Firstly, let me say, if you believe the rhetoric of the microsoft-moonies, go right ahead and put all YOUR data on the web. Hard drives may go away, but not because there is not a need for local (private) storage. Check the Napster discussions and you'll find a few who almost feel *violated* by Metallica's sting-- because they not only searched drives users had Napster pointed to, but Everything in the user's possession. I don't happen to use Napster or MP3 stuff and I believe in copyright enforcement, but I also know that, no matter whom you trust to keep your private affairs, no one can really protect you from prying invasive eyes but YOU. So, local, "off-net" storage MUST exist.
But hard drives? I am more certain than ever that the day will come when we see solid-state memory cartridges take the place of spinning platters of magnetic-plated media. Look, for instance, at the work of a company ( try http://www.ovonyx.com ) that Intel has recently made an association with. This is what we want: truly NON-VOLATILE (like core memory!) storage, no batteries, no 3-D laser array exotica. Just a chip. Now, imagine, some of these are incorporated on your motherboard. Imagine further a slot port on the front of your PC with two lockable sockets for cartridges, the cart size maybe similar to a Zip. You can lock them so that no one takes them out while you are at lunch. Or, you can unlock them: pull out cartridge A when you want to change your OS or configuration (keep several around if you want). In cartridge B might be a digital video movie you recorded off the 'net. Same media, for hard drive, for music (no more CDs), for backup (no more CD-R or tape), for video (no more DVD), for games and everything else we record to.
Who knows how far off this is? Only density concerns hold us back NOW. But we are getting better at that every day. A Zip-sized cart might well hold 30-50 tested little slices of chip, each tested and verified for its operation, embedded together and connected via an overlay. I predict (laying on a Gypsy accent) we could see Gigabyte catridges for approx $15-25 within this decade. Commercially distributed movies and music, if they broke away from the optical-CD as a format, might move the economics further; cartridges of "pre-recorded" variety (write lines broken) might be manufactured for $2-4 apiece. This works for PC software too...
Now, I could tell you of the changeover we will soon make to a standard 16:9 screen ratio for PC monitors, or how, with fewer motorized drives, we'll transit back to desktops/laptops over towers, or how multiply-independent processors on a unified chip or two will shrink "Beowulf" array technology so we ALL have super-computers, or, at least, REAL Multitasking for once. (A shrunk wolf... Beopoodle?)
Or, I could give you the cynical lowdown on how the PC Industry really wants you to pay as you use, in a repeat of TimeSharing days, so that every email you send, every time you calculate on your spreadsheet, your credit card is charged...
But maybe I'll save that for a future post. In the meantime, take on Heidi's challenge: What IS the real future of computing?

[hd581]

Why aspire for a whopping 2 GB or so of solid state memory? Even only 2 GB would be incredibly expensive.

They should manufacture a 128 MB chip of non-volatile RAM. This would be ideal for laptops and servers. If the power goes out, the computer just shuts off in the middle of what its doing. No save to RAM or save to Disk. When power is restored, it comes back on like nothing ever happened.

This is something that could be marketed reasonably and is a good way for Solid State devices to get their foot in the consumer door.

[Glynn R Harris]

Good, HD... As I referred to a bit of this "ovonyx" memory on the motherboard, I was thinking of exactly that.
I have an old calculator, looks like a Selectric and about as heavy, but it actually has core memory in it. I found this out one day when moving. I plugged it in at the new house and turned it on... the results of my last calculation some three months before were onscreen! Pretty wild. I can certainly see a sever or laptop application for that kind of thing. I'm sure Intel and Ovonyx will make that SIMM you suggest the first product they ship...
I expect, however, that events on the horizon NOW will soon make 128M SIMMs and RDRAMs parts that you'll sell on eBay for sentimental and collector's value (no doubt encased in a lucite cube, how cute!). I'm not talking this year, or even next. My horizon, at a guess, is seven or eight years... and not based on a product, per se, but a PROBLEM...
While the Ovonyx memory is my idea of hard-drive's "final tour", I'd have to see real parts and their read/write speeds before I'd hop on the bandwagon to replace system RAM with the stuff. But SOMETHING needs to happen in that arena.
RAM has been getting pretty fast, lately, but has just BARELY been keeping up with the processor speeds they feed. We'll need even faster memory to go much beyond the GHz barrier.
While Intel struggles to supply the stuff and Samsung et al are trying to make us all accept high-priced RDRAM, they're all talking of CPU architectures where the pipelines are wider and the parallelism is deeper. These CPUs under development now are going to drink instructions like frat-boys at Spring Break.
Most likely, we will need Terabyte arrays. Not only will they have enough space for a super-CPU, the memory will (in my opinion) be BUILT-IN to the CPU assembly. No other bus, no cache structure can do what you can do WHEN you are millimeters from the CPU itself.
And high cost? Yep, if built right now (yes we could). But isn't it "Moore's Law", that explains why my Pentium 200 MMX is not as valuable as it was when I bought it, oh gosh, has it only been four years? (It seems like a century, especially since I've been reading lately about everyone else's Athlons and overclocked Celerons and P-IIIs).
Expensive is a relative thing. Memory made as part of an assembly, considered as part of a CPU, might make some sense... my first motherboard had lots of weird DIP chips on it that were the cache. Then I bought a MB with a socket for a cache-memory card. My next MB had neither; the cache was one chip soldered in. Now most of the newer CPUs have internal cache; what I'm saying is that main memory is going to make that same migration.
Opinions?

[GregM]

As I write I am still downloading the PDF presentation on the "ovonic" memory. The blurb page made great claims, but lets wait to see results. It looks like a compond heterojunction of Antimony Telluride and Germanium Telluride. That is not simple silicon process and may not be as easy to manufacture. I am sure hard drives will eventually give way to solid state, but they won't die easily, as they keep developing technology to make them cheaper, denser, and faster.

I totally agree with your preferrence of private storage, maybe more strongly than you. I want control. Is someone going to give me a terabyte of space for free, or am I going to pay monthly for it? When communication lines are down, is my PC going to nothing but a night light? At my work we have we have a terabyte server and people were encouraged to store there personal data on it, then they told us we were restricted to 200MB. My mail files were bigger than that (I have saving some e-mail for over 6 years). I said why bother, my desktop came with the minimum 6GB drive and even with 100Mb network, my local data is faster. Also, being an NT network, it does go down every so often.

What about other predicted future trends? Will uncle Bills next push, for voice interface be the big thing. What a wonderful way to type a private memo. HAL is a long way off and I see the "silent" interface having advantages that will keep in use for some time.

"I believe he is talking about Television. That ceased to be an entertainment media about 2040." Data's approximate words in a Star Trek episode. Yes, computers and TV will merge, somewhat akwardly at first, I believe, but probably before 2040. The first steps have been taken by the PC community making demands on the design of HDTV. Mergers like AOL with Time Warner will set us up for it, but not for free. Instead of free HBO you will probably end up paying for the equivalent of websites as part of a package deal or "enhanced" digital television. Right now I am waiting for the local phone company to install a category 5 line and digital TV and 1 Mb/s internet service. The caller ID will appear on the TV screen and you will be able to see web type data on shows at the same time on the same screen.

But these are minor changes compared to what will develope as bandwidth. As my crystal ball boots up I will brag that about 1989, when the LJIII was being shared in the office, I predicted in a decade, most of us would have color laser printers at home. I wasn't way off, not forseeing the inkjet eclipsing the laser in economy.

Here is an open ended prediction. There is a remote modelling technology which is being developed for industry whereby a CAD design at one location can be sent to a plant anywhere in the world with a special machine which can generate a solid model. Couple that with IBM's technology for placing single atoms at will with an atomic force needle and you will have a system for designing products that can be assembled at remote locations, perhaps eventually in the home. See a product on the web, buy it, and a machine slowly materializes it next to your PC/TV. No waiting for Federal express. Of course the products will have to be special designed for the process using limited materials and interacting parts, but if it can profit, well look how home computers have advanced since the Altair. Eventually nanomachines may be able to build about anything. Levi may be suing for pirated jeans shared over the net. My timeframe is only within the lifetime of some people old enough to read this.

When my dad went to school, a teacher told him Buck Rogers was silly - rockets couldn't fly in outerspace. Now the Air Force is building a Laser to shoot down incoming missles, which could come from third world countries. His moral: anything they can imagine in science fiction will eventually come to be. You should see what's coming up next in the crystal ball.

[Brastius]

Here's a prediction: data storage that takes up no space, at least in our dimension. Imagine a storage system that utilizes a worm hole connected to a pinhole universe (yes, I like this idea as seen on the Big Bang discussion) that is used Soley as a place to keep data. Data retrieval would be accomplished by switching the point of entry into that universe. I have no idea how long from now this might be feasible, possibly in a thousand years, give or take a decade.
Then there's the possibility that eventually the inhabitants of this Universe might object to being used for calculating the answer to life, the Universe and Everything.

-Brastius

"Arthur hoped and prayed that there wasn't an Afterlife. Then, seeing a paradox here, he just Hoped there wasn't an afterlife."
-Hitch-hikers Guide to the Galaxy

[hd581]

Glynn: I agree that there will always be local storage no matter what. But, there might be a chance that both shared and local storage will be utilized in the future.

I'd have to see real parts and their read/write speeds before I'd hop on the bandwagon to replace system RAM with the stuff. But SOMETHING needs to happen in that arena.

Anything even approaching RAM speeds is going to be faster than a hard drive. PC100 RAM access is around 7-8 ns. Hard drives access data at 8,000,000 ns (8 ms).

I agree that some sort of progress needs to made here in terms of speed. I read on another post that there is an upper limit to how fast they can spin a platter (sonic boom prevents it from going past a certain speed) and they haven't improved on the 8 ms access speed for almnost 10 years.

Imagine how drastically an affordable and practical Solid State hard drive would change computer design.

• If it has read/write speeds comparable to RAM, it will essentially replace it; there will be no further need for RAM at all.
• Programs will be installed and then run once and never need to be closed again. (virtually unlimited RAM). This means that everything you use will always be open. This also means you won't need overlays or swapping or anything that is memory limited.
• The processor will need to be redesigned to take advantage of the fact that everything will be running at all times (though perhaps not active). Perhaps this will spur more popularity for dual processors.

Food for thought...

[This message has been edited by hd581 (edited 06-01-2000).]

[GregM]

Touché, Brastius. I won't try to one up that. But I was fairly serious. I like your quote, though. I found the Hitchhiker's Guide to the Galaxy to give as good and comprehensable theory of life, the universe, and everything, as anything else. If the smartest man in the world says "we probably can't comprehend super string theory yet" I will have to stay with my big screw up theory (supported by the above work) or the simpler concept of non-existance. Sorry for crossing threads, but you started it.

On a slightly more serious note, I think hd581 is on the right track. I think this Von Nueman architecture is getting a bit long in the tooth. Intel's idea of all functions being sucked into the black hole of the microprocessor has serious limitations and makes more business sense than computing sense, IMHO.

I have thought a model more like the brain would be beneficial. The brain has effectively a much slower clock speed and signal propagation is a tiny fraction of electrical signal speeds, yet attempts to emulate its functions in silicon have mostly been pathetic. Granted, the software has had millions of year more development time, but the archetecture has had almost a billion years of development. Extreme parallelism, diffuse and specialized processing are some of the brains characteristics that differ greatly from the desktop processor. Sure, a Pentium can crunch numbers a million times faster, but we haven't been doing that long enough for it to be incorporated into the brain. How about a brain with an FPU?

[Glynn R Harris]

I'm all for a brain with an FPU. Gawd I would never have been such a ditz in Algebra and Calculus at school.
If anyone's working on it, I would also ask for radio transmitter/reciever combo implants. Helpful in relationships, I would think.
My sister is blind (the result of diabetic retinopathy) and now wears two prostheses, fake eyes that are painted to look like the real thing. I've always felt like little video chips could be incorporated into a glass eye... the problem I guess is how do you send the signal and make the brain interpret it correctly. The wiring install alone would probably be a feat of major proportion.
My dad, however, now wears a hearing aid that has a microprocessor in it. My mom has a pacemaker that is supposed to last ten years on its battery, and it can be "talked to" at the doctor's office, gives reports of what all it has been doing, can be adjusted accordingly by the doctor.
I guess I can look forward to a future where, as my parts wear out, they can be replaced... the Picard/Locutus image floats in my mind though... WARNING: the Surgeon General has determined that you will be Assimilated!

[Meat Puppet]

See post on nano tech pros and cons for my answer ..

[qball]

My two cents:

Solid state memory will exist and be commonplace in my life time (the measure of the rest of my lifetime is left as exercise for the reader), regardless of the cost or technical issues that exist today. Local and remote storage will continue to coexist. I personally like the idea of remote storage as I may access it anywhere.

As far as remote storage security, that reminds me of a PGP (pretty good protection) conversation. My argument: Why are you using the web/network to send/receive highly sensitive information? Don't store anything sensitive remotely and you should be fine (though this seems to work opposite of human nature). It is far easier to be discreet and have nothing to hide, than to hide your indescretions.

Lastly, it does seem odd that we are still recording stuff on rust and sticky tape, though don't forget your optical drives!

[tonym]

Hello All,

I'll be happy with a computer industry/infrastructure that get their hardware *hit together!

We're stuck with legacy **** from the late 1970's. The future of computing won't happen until we kill and bury the AT? form factor.

Computers that use the present form-factor were designed to be laid on desktops in horizontal cases (the XT!). This design was carried over when tower (upright) cases were popularized and bastardized until we have the nearly crapped-out computer hardware formats of today! They are wiring and thermal nightmares!

My future computer would have some simple things included:

1. locking/positive contact connectors on ALL connection points. This includes HDDs/floppies/ZIPs/etc. These would also include polarized, foolproof connectors. did you ever have to re-dissemble a system becuase one lousy IDE connector slipped off during the previous dissembly??
2. A redesign of the I/O card interface area. These cards should be accessible from the front *or* the rear of the system, based on the choice of the *CUSTOMER*. The PCI philosophy can be "compartmentalized" (reference CompactPCI) such that it is a *subsystem*. This would make it easier to adapt to the front of the case, if that was desired. Also, it would remove the boneheaded upside-down cooling that we're stuck with. Legacy thinking that tries to defy the laws of physics!
3. PSU's and peripherals that plug into the system. Why the hell do you need to unscrew the system every time you want to upgrade almost anything?? I can understand the mobo, but I have to take my system apart to replace a CD player?? Or to upgrade my PSU to 300W from 250W. Just plain dumb!
4. A standardized cable bundle for signal egress from the system. I've got many systems with hundreds of cables and it gets to be a mess! The signals going to/coming from the PC are the same! Why can't there be a single, unified wiring harness that groups all these cables together for ease of system hookup?
5. True front-to-back unobstructed cooling. The aesthetics of the PC rule it's technical performance. Function follows form! This is dumb. The paramount interest should be to make the PC as reliable as possible...this means better cooling. Right now, the convoluted cooling path in a PC makes it almost impossible to assure consistent cooling in a given case. This is dumb! in order to get a pretty face plate or front cover, the "wholesomeness" of the system is sacrificed! We need cooling vents along the entire front and the backside to facilitate proper cooling and ventilation.

You folks can argue about solid-state memory and other Buck Rogers whiz-bang stuff, but we've got to get past the legacy obstacles before computing can really improve and take us into the future!


Tony


[This message has been edited by tonym (edited 06-13-2000).]

[This message has been edited by tonym (edited 06-13-2000).]

[Glynn R Harris]

Tonym;
I seem to have read somewhere of a move on to get hard drives and floppies and cdroms on a sort of "serial" cable, maybe even hybrid optical+power similar to Firewire, on the supposition that if you narrowed down the godawful mess of ribbon cables in the PC, thermal problems in the PC would be eased. At any rate, maybe it would mean that a mobo might have one or two connectors and you'd simply chain drives up; no Master/Slave and all that nonsense, and maybe you'd hotswap without hassle.

Actually, tonym, every "legacy" problem seems (at least to me) eventually to be solved by ANOTHER "legacy", which makes its way to problem pretty quickly. This ISN'T a problem: it's called Evolution, and it's a herky-jerky affair.

ASCII= Postscript, nope, Truetype
Eight.three= Windows "Joliet"
ISA= EISA, nope, MCA, nope, VLB, nope, PCI
PC= XT, nope, AT, nope, ATX
SA450, RLL= ESDI, nope, IDE, nope, ATA
MDA= CGA, nope, EGA, nope, PGA, nope, VGA
RAM Chips= SIMM, nope, DIMM, nope, RDRAM
8086= 80286, 386, 486, Pentium, P-Pro, P-II, P-III, Itanium?
5-1/4"= 5-1/4HH, 3-1/2"
45W-PS= 100W, 150w, 200w, 300w
ROM= flashROM
300 baud= 1200, 2400, 9600, 14400, 28800, 33.6K, 56K, ADSL
and all along the way, we have had about a thousand evolutionary cul-de-sacs, PCs from the early Sanyo MBC-555, Corona, Wang, Xerox, TI Professional---- all trying to do the PC one better by making a slight improvement, but one that just didn't "take". I personally have sound cards with built-in Proprietary CD interfaces. Yuk. Remember HST modems? 720K disks? 1.2M ("quad-density") 5-1/4" disks? I've had about 4 iterations of tape backup--- thank God for CD-ROM.

I, too, think the PC should undergo a total revamping. But I have seen a lot of changes (and had to purchase most of them) in the last 15 years. I'll probably always be upgrading MOSTLY out of necessity, not a desire to push the limits. This is not YET a "mature" technology... and we fool ourselves into thinking it is, just as wireless telephone customers thought with their first analog carphones.

One of the things I was wondering, when I first wrote on this thread, is: where do you see something falling into place, where is it all going to simmer down?

I think solid-state storage is the end-all of disk technologies, for instance. I cannot see motorized spinning platters holding our data from now on. But of course I visit Fujitsu.co.jp and see that they have made another leap in magnetic storage density, likely to show up in drive products within a year.

I've gotten the feeling that more and more processors in the user's system is how we are going to move, once we get the .13 micron 10GHz processors and have exhausted our "easy" speed gains. Are we going to move towards backplanes with 64-Processor-Array subassemblies and such? If so, the PC revolution hasn't really even begun.

I started converting some of my important documents to HTML (the future) and then learnt of a not-finalized vector-based web technology standard, from the Adobe website. I don't yet know what it all means. I'm looking.

I have a brochure here from a company that made a really neat PC that hit the market about two-three years ago... Nexar. It was a cool PC, because the processor part could be upgraded very easily... a subassembly card that snapped in. The company said "now your PC will never be obsolete".

The CPU and memory were accessible from a door on the Right of the case, and the IO cards were in the door on the Left-- because the mainboard was made to have the CPU mounted UPSIDE DOWN/BACKWARDS, or however you want to call it. The drives were on trays you just slid in or out. The Nexar had air ducts from front to back. Etc, etc. Good try.

It sounded to me, tonym, when you were describing your perfect machine, that a Nexar would have fit the bill nicely. Does anyone know whatever happened to Nexar? (A software company now appears to have appropriated their name on the Web...)

Sometimes being "ahead of your time" like Nexar is like being a premature baby; damned fragile and incapable of living without an extensive support system until it matures a bit more.

Take care,
Glynn.

[This message has been edited by Glynn R Harris (edited 06-21-2000).]

[Whir]

What about Quantum computers? Last I checked they actually had a working model with five electrons. Or was it atoms? Poor memory. Seems to me, like some of you have said, that if we truley want to get out of this legacy/microprocessor ****, we need to stop thinking in terms that apply to the standards. To heck with magnetic storage, or opitcal storage, or whatever kind of storage those Ovonix carts would be. We're talking electrons being used as bits. Electrons are funny things, being able to be in two places/states at the same time... I suppose we'd have to learn how to master making them do what we want them to do before we could force them into obediance for storing our data. But think of the size of the computer you suddenly have when one electron equals one bit. 10GB storage devices would still be microscopic. Seems a little far fetched for sure, but no farther than using pinhole universes.

We definitely need to get away from the current standards of computing technology though. Solid state electronics will only be able to go so far. There has to be a barrier (not saying we'll find it in the next ten years, but hey!)

Just my twisted thoughts.

-Whir