Dream Machine


Grench

Forum Addict!
Joined
Oct 3, 2008
Messages
6,629
OK, so, this thread is meant to occupy mine and maybe other minds while we sit in a holding pattern waiting for our Pandora to arrive. I.e. something to do instead of getting upset with the current pace.

The idea here isn't to grind over the possible, done before, looks like product X that competitor Y is producing. Lets have some real novel and out there ideas.

I have this idea of a modular pocket-able dynamic self clustering system. It would run off of a server style node-cluster Linux system.

There are three main pieces to it. All of them should be 1/2" thin or less.

1. A controls module. Preferably just about like what we have for the Pandora's base today. Each of the 4 edges would have a hinge mount with integral data connections. It would have it's own motherboard, ARM CPU, RAM, battery, networking, SDHC card slot, etc.

2. A screen module. Preferably something on the range of a 1024x800 equivalent 4.8" diagonal Pixel Qi display. Each of the 4 edges would have a hinge mount with integral data connections - preferably optical based. It would have it's own motherboard, ARM CPU, RAM, battery, networking, SDHC card slot, etc.

3. Hinge modules. Connect the hinge (with integral data connection) to the top, bottom or either side of one module and connect another module. No limit to the numbers of modules that can be combined. The hinge should be a bi-hinge so that the unit can fold 180* but strong enough to hold it locked - maybe even put an adjustable friction lock on it to hold it at angles. At the point of connection, the two modules join together to create an N node Linux cluster with N CPUs, N times the RAM, N video chips, etc...



You want a bi-fold portrait e-book reader? Put in a long side hinge and attach another module. Instant two screen e-reader clamshell.

Want a long skinny tablet? Put the hinge on the short sides.

Want a Pandora form factor? Put the hinge on the long side between a controls module and a screen module.

Put the hinges in the right places and you can make a 5 display plus keyboard 6 node cluster computer with One Keyboard, 6 CPUs, 6 SDHC slots that folds up to be a 3" thick super computing brick.

With multiple wireless devices built in the data connection can go into a highly parallel mode.

With the modular assembly nature of it - all the motherboards can be the same regardless of it being a display module or a controls module or other. Additional module possibilities could be done. Need more CPU? A module with two motherboards and no control and no screen.

OK, so, that dream is likely years away. What is yours?
 
sebt3 said:
You're looking at http://www.buglabs.net/products ...

No - those are clunky and restrictive by comparison. They're also single-purpose modules. My idea is for multiple general purpose computer modules that link together to form a cluster that can be a device of any desired form factor.

The closest I've seen to the bi-hinge idea is the HP28s calculator. It has a dual-hinge design so that the lid flips out to be the left keyboard then can continue to flip behind to take up less desk space.
http://en.wikipedia.org/wiki/HP-28_series

Picture that with the hinge and sides being separate parts. Each module/node can then join to up to 4 other modules/nodes creating a cluster. Since each module has it's own motherboard, CPU, SDHC, RAM, networking, etc... It winds up being a truly multiplicative instead of additive creation.
http://en.wikipedia.org/wiki/Cluster_computing

So, having a base module and a screen is a dual CPU system with twice the RAM, etc...
 
Last edited by a moderator:
christoph said:
here is another one to start with: http://www.liquidware.com/

Again - that is taking individual function modules and stacking them together onto a single central controller/CPU. My main idea is to stack fully functional computers together into a computer cluster.
 
Last edited by a moderator:
Grench said:
OK, so, this thread is meant to occupy mine and maybe other minds while we sit in a holding pattern waiting for our Pandora to arrive. I.e. something to do instead of getting upset with the current pace.

The idea here isn't to grind over the possible, done before, looks like product X that competitor Y is producing. Lets have some real novel and out there ideas.

I have this idea of a modular pocket-able dynamic self clustering system. It would run off of a server style node-cluster Linux system.

[...]

With the modular assembly nature of it - all the motherboards can be the same regardless of it being a display module or a controls module or other. Additional module possibilities could be done. Need more CPU? A module with two motherboards and no control and no screen.

OK, so, that dream is likely years away. What is yours?


This isn't quite the same as the BUGbase linked above, but it does share a similar idea. I really like your idea of duplicate Mobo/CPU/GPU/RAM plugging in for more power or multiscreen etc.. but having two 600MHz CPU's is not actually 200% better then one, there is always loss, and the loss is bigger with each CPU added. Same with SLI graphics, having 2 vid cards is not 2x video performance. Though having 2 vid cards each powering one monitor by itself CAN be quite the speed boost.

Either way, done properly your idea has a lot of merit and cool possibilities.

My dream is a far more efficient solar generator, one that can become part of the lid of any device, and hold (and store, with better power cells) enough energy in an hour or two of light to power the unit all day and through the night. Not just the generator though, but vastly improving the efficiency of what we can do with X amount of energy, hopefully using less and less energy to power more and more complicated technology. Similar to the way modern ARM processors can get higher speeds at lower voltage then old Pentium II's could dream of.

So one day, a Pandora-ish device that never runs out of juice, never needs charging, and can power other things plugged into it. Though, at that point of technology, the need for physical plugs for anything including charging should be removed.

Another idea I had is kind of like speech-to-text, but I want speech-to-code. A program (or machine) that you can describe what you want and it will write the code (or at least the programming outline) for you. There already is pretty good AI out there that can come REAL close to having a natural conversation with random people, I would think converting the results of a conversation into a program shouldn't be TOO far fetched.

Dunno, random ideas from a 4:20 session, take them how you will.

EDIT: took me so long to type, I got ninja'd! 3 times!

- trix
 
Last edited by a moderator:
Grench said:
OK, so, that dream is likely years away. What is yours?
This could actually be made any day. It would be insanely expensive, as every bit more or less would be/cost as much as the Pandora. It would also be kinda limited in the software department since everyone would either code for one screen, with our without a keyboard unit attached, or what ever config they decided was the right one. Say, 3 screens and one keyboard unit, for the guy who is trying to code a car game, 5, (cross pattern),9 screens (3x3rows) for someone making a shooter, 11 for someone making a flight-sim... (3-5-3), and the dude with the flight-sim thinks he needs a bit more buttons, and the added CPU power can never be wrong, 3 keyboard units.

The key to open source is a uniform development environment. Either, like on PC, you make it so, through a software layer between you and the hardware, or, like on consoles, of old, you code directly for the hardware. A system like yours would have to be coded on a software layer losing a LOT of the recourses in the process. (Consoles of today are some sort of middle ground as i understand it. Developers do have access to some hardware directly, if they spend the time, and take the effort, but other resources are only accessible through a software layer on the unit. This is mainly to let the developer of the hardware change chips for cheaper versions while keeping compatibility... Or so i think at least. It means we end up with less optimized code, that runs slower. Consoles and dedicated gaming computers used to rock, nowdays they suck.)

Your system isn't bad, i just don't see any natural use for it. It's a mind-boggling concept, and similar to what they use when building "autonomous robots". They are made up of 1 unique part, instead of 2-3 but still.
B!
 
Last edited by a moderator:
trix said:
My dream is a far more efficient solar generator, one that can become part of the lid of any device, and hold (and store, with better power cells) enough energy in an hour or two of light to power the unit all day and through the night.
I've said this before and I'll say it again: designing something that makes you intentionally leave electronics in the sun is a bad idea. At only about 40% efficiency, almost 60% of the energy from the sun is being converted directly into heat which then gets transferred to the electronics inside it, potentially doing a significant amount of damage over time. If portable solar charging is your interest, it is far safer and more practical to make it a separate devices which plugs into the electronics in question. Not only does it keep the electronics from the heat source, but it becomes something modular that you can use for multiple devices.
Imagine, if you will, the battery cover having a solar panel inside it. Pop the cover off, flip it over, run a 2 foot wire from the panel to your device, let it charge.
 
Last edited by a moderator:
My idea: Speech to hardware. With speech to code, there would be no issues related to lacking a single target for applications. Imagine you needed a device that no one would ever mass-produce, and which would cost hundreds of thousands of dollars to develop. The computer could just design it instead based on your description of it, and then print out as many copies as you needed.
 
WizardStan said:
trix said:
My dream is a far more efficient solar generator, one that can become part of the lid of any device, and hold (and store, with better power cells) enough energy in an hour or two of light to power the unit all day and through the night.
I've said this before and I'll say it again: designing something that makes you intentionally leave electronics in the sun is a bad idea. At only about 40% efficiency, almost 60% of the energy from the sun is being converted directly into heat which then gets transferred to the electronics inside it, potentially doing a significant amount of damage over time. If portable solar charging is your interest, it is far safer and more practical to make it a separate devices which plugs into the electronics in question. Not only does it keep the electronics from the heat source, but it becomes something modular that you can use for multiple devices.
Imagine, if you will, the battery cover having a solar panel inside it. Pop the cover off, flip it over, run a 2 foot wire from the panel to your device, let it charge.

Note the parts (that you quoted) that say "far more efficient" and "enough in an hour or two of light" that were put there specifically to address your concern.
 
Last edited by a moderator:
typs lik dis said:
My idea: Speech to hardware. With speech to code, there would be no issues related to lacking a single target for applications. Imagine you needed a device that no one would ever mass-produce, and which would cost hundreds of thousands of dollars to develop. The computer could just design it instead based on your description of it, and then print out as many copies as you needed.

Your idea is already in progress. A few of the professionals designing for The Venus Project are working on improving speech recognition and understanding by computers, the end result to be used (in their society) towards goals similar to what you describe. That is, the end user making hardware by description. Also, in their society (if you are interested in that sort of thing) the computer would automatically network your request with other people with similar interests so that if the computer cannot do it automatically, it can put enough professionals (and materials) together to get it done.

I was going to link the source of your specific idea, but it's repeated (over and over) in every text or movie describing what life would be like in this society.

- trix
 
Last edited by a moderator:
trix said:
Note the parts (that you quoted) that say "far more efficient" and "enough in an hour or two of light" that were put there specifically to address your concern.
"sufficiently more efficient" to prevent overheating is still a long, long way off, and will probably never be achieved. I could do the math if you really wanted me to (I don't want to, it's long and boring) but you're probably only looking about about 20% of the suns energy being converted to heat on a mattblack plastic device like the Pandora right now. 80% is lost due to reflection, about. So you would have to basically double the current efficiency of the most expensive existing solar panels to just keep the same amount of heat that would go into it right now without a solar panel.
It's a noble goal, but if this interests you, don't set your sights so high. You will invent something that doesn't work as intended and no one will have gained anything. Design a disconnected device, like I said, and work on improving the efficiency of that.
 
Last edited by a moderator:
WizardStan said:
"sufficiently more efficient" to prevent overheating is still a long, long way off, and will probably never be achieved. I could do the math if you really wanted me to (I don't want to, it's long and boring) but you're probably only looking about about 20% of the suns energy being converted to heat on a mattblack plastic device like the Pandora right now. 80% is lost due to reflection, about. So you would have to basically double the current efficiency of the most expensive existing solar panels to just keep the same amount of heat that would go into it right now without a solar panel.
It's a noble goal, but if this interests you, don't set your sights so high. You will invent something that doesn't work as intended and no one will have gained anything. Design a disconnected device, like I said, and work on improving the efficiency of that.

Even the reverse of that could be nowhere near right: an 18% reflective object is Middle Gray, Matte black has got to be under 5% reflective, and I wouldn't be surprised if it was under 1%
 
Last edited by a moderator:
trix said:
typs lik dis said:
My idea: Speech to hardware. With speech to code, there would be no issues related to lacking a single target for applications. Imagine you needed a device that no one would ever mass-produce, and which would cost hundreds of thousands of dollars to develop. The computer could just design it instead based on your description of it, and then print out as many copies as you needed.

Your idea is already in progress. A few of the professionals designing for The Venus Project are working on improving speech recognition and understanding by computers, the end result to be used (in their society) towards goals similar to what you describe. That is, the end user making hardware by description. Also, in their society (if you are interested in that sort of thing) the computer would automatically network your request with other people with similar interests so that if the computer cannot do it automatically, it can put enough professionals (and materials) together to get it done.

I was going to link the source of your specific idea, but it's repeated (over and over) in every text or movie describing what life would be like in this society.

- trix

Yeah, but in my idea, the computer can design everything. Also, some of their buildings look really ugly. They aren't shooting very high, IMO.
 
Last edited by a moderator:
davidgro said:
Even the reverse of that could be nowhere near right: an 18% reflective object is Middle Gray, Matte black has got to be under 5% reflective, and I wouldn't be surprised if it was under 1%

Adding to that, it does strengthen WizardStan's point that leaving a Pandora (or most anything else electronic) out in the sun is a Very Bad Idea. If it had solar panels, they couldn't help much on that point unless they actually had a big airspace between them and the case (which would mean they provide shade) or if were efficient at the whole spectrum, instead of a very narrow frequency range (which I think is in infrared for regular solar cells) which of course researchers are trying, and we'll know if they make it to market.
 
Last edited by a moderator:
Then again anyone who has used a solar calculator knows you don't have to e standing in the sun to use it. I think the day will come, though not for some time, when the power requirements for handheld computing will e small enough and solar tech will be efficient enough, that just a well lit room may provide enough power to actually run a device completely off the power generated at that moment by the solar cell.

Also, maybe someday capacitors will be the main power storage for vehicles which will solve the whole car charging dilemma.
 
davidgro said:
Even the reverse of that could be nowhere near right: an 18% reflective object is Middle Gray, Matte black has got to be under 5% reflective, and I wouldn't be surprised if it was under 1%
There's more to light than what you can see.
 
Last edited by a moderator:
typs lik dis said:
Yeah, but in my idea, the computer can design everything. Also, some of their buildings look really ugly. They aren't shooting very high, IMO.

As far as the Venus Project, it's about the society and the resource based economy, not just the looks. It's not supposed to be perfect, just better.

About the idea, that's what they are after as well, a computer that can design everything based upon a plain english description. Of course, technology is not even close to there yet so first it will only be able to assist or guess.
 
Last edited by a moderator:
WizardStan said:
davidgro said:
Even the reverse of that could be nowhere near right: an 18% reflective object is Middle Gray, Matte black has got to be under 5% reflective, and I wouldn't be surprised if it was under 1%
There's more to light than what you can see.

That's true, but sunlight does peak in the visible at the ground. (which is of course why that's the range we evolved to see) so it's a close enough approximation. I think it's likely that matte black plastic also absorbs IR, and perhaps near UV, but I don't know that. When I get my Pandora I'll check with a TV remote and my UV keychain LED using a digital camera. Anyway the point is that anything not bright white (brighter than most white things are) can get very hot when out in the sun.
 
Last edited by a moderator:
Back
Top