Do you guys think that in the next decade we will have computers that are so powerful that software won't be able to consume all the resources(excluding calculating pi and other math intensive computations), honestly what is going to require a 64Ghz processor 50 Gigs of RAM 10Tb HD and 10Gid video card? Numbers assume Moore's Law and everything keeps up with processor speed more or less.
Well, most modern/popular software (like Windows, and Games, and stuff) are like goldfish: they expand to fit their tank. As computers get faster, software will keep pace with it.
Quote from: Nate on August 26, 2006, 11:50:50 PM
Do you guys think that in the next decade we will have computers that are so powerful that software won't be able to consume all the resources(excluding calculating pi and other math intensive computations), honestly what is going to require a 64Ghz processor 50 Gigs of RAM 10Tb HD and 10Gid video card? Numbers assume Moore's Law and everything keeps up with processor speed more or less.
Umm... by that time... Starcraft 2?
Duke Nukem: Forever
Quote from: iago on August 27, 2006, 12:18:15 AM
Well, most modern/popular software (like Windows, and Games, and stuff) are like goldfish: they expand to fit their tank. As computers get faster, software will keep pace with it.
I always thought it was the other way around with software exceeding what the hardware can do and the hardware trying to keep pace. One example of this would be with the quake 2 source code and such. It has a bunch of assembly code but the recommendation today is that if one wants to improve upon the source code they should remove the assembly code and use the C code. Why? Simply because the hardware has improved and going the extra mile to squeeze out an extra clock cycle is no longer needed for the game to perform reasonably.
Quote from: zorm on August 27, 2006, 01:20:47 AM
Quote from: iago on August 27, 2006, 12:18:15 AM
Well, most modern/popular software (like Windows, and Games, and stuff) are like goldfish: they expand to fit their tank. As computers get faster, software will keep pace with it.
I always thought it was the other way around with software exceeding what the hardware can do and the hardware trying to keep pace. One example of this would be with the quake 2 source code and such. It has a bunch of assembly code but the recommendation today is that if one wants to improve upon the source code they should remove the assembly code and use the C code. Why? Simply because the hardware has improved and going the extra mile to squeeze out an extra clock cycle is no longer needed for the game to perform reasonably.
Modern software, in my experience, usually pushes the limits of the computer, but doesn't exceed them a whole lot.
Look at it this way: if computers suddenly got 10x faster tomorrow, I'm fairly confident that software would quickly (maybe over a few months or a year) expand to fit the void.
If software suddenly got 10x slower (harder to run, whatever), I doubt computers would be able to catch up, at least, not for awhile .
That's what I meant by expanding to fit the available space.
Quote from: Nate on August 26, 2006, 11:50:50 PM
Do you guys think that in the next decade we will have computers that are so powerful that software won't be able to consume all the resources(excluding calculating pi and other math intensive computations), honestly what is going to require a 64Ghz processor 50 Gigs of RAM 10Tb HD and 10Gid video card? Numbers assume Moore's Law and everything keeps up with processor speed more or less.
With things like quantum computing on the horizon, I think the numbers will look different than that. There have already been quantum computers made (non-programmable) that have solved the traveling salesmen problem on a level of complexity that with a "normal" computer would seem impossible in a timely manner.
And yes, I do think that there will be plenty of software that will leave hardware capacities room to grow.
Quote from: Sidoh on August 27, 2006, 04:17:49 AM
With things like quantum computing on the horizon, I think the numbers will look different than that. There have already been quantum computers made (non-programmable) that have solved the traveling salesmen problem on a level of complexity that with a "normal" computer would seem impossible in a timely manner.
And yes, I do think that there will be plenty of software that will leave hardware capacities room to grow.
Have they actually solved an NP-Complete problem in polynomial time using quantum computers? I haven't heard about that, but that'd be a massive breakthrough. If you solve one NP-Complete problem, you've solved them all.
I'm not entirely sure that quantum computers will ever make it to the desktop, though. But I guess that remains to be seen.
Quote from: iago on August 27, 2006, 10:40:31 AM
Have they actually solved an NP-Complete problem in polynomial time using quantum computers? I haven't heard about that, but that'd be a massive breakthrough. If you solve one NP-Complete problem, you've solved them all.
I'm not entirely sure that quantum computers will ever make it to the desktop, though. But I guess that remains to be seen.
From what I have heard, yes.
That isn't necessarily true yet, though. They solve an NP problem by exploiting physical phenomena. Quantum computers aren't programmable -- yet. You create one to solve
a problem.
I have heard of proposed methods to create programmable quantum computers, but I am not sure that it was a practical method of doing so.
I highly suspect that programmable quantum computing will be seen in a desktop environment sometime in the future.
Quote from: Sidoh on August 27, 2006, 02:16:47 PM
That isn't necessarily true yet, though. They solve an NP problem by exploiting physical phenomena. Quantum computers aren't programmable -- yet. You create one to solve a problem.
However, a problem is considered NP-Complete if it is equivalent (ie, reducable) to every other NP-complete problem. Thus, An NP-Complete problem must be reducable to any single other proven NP-Complete problem (which is the same property). Thus, if a computer can solve a single one, it can solve any type of NP-Complete problem. So we just need to find a way to make it take proper inputs for the Travelling Salesman problem, and bam! we can solve any NP-Complete problem.
NP-Hard problems are also problems that cannot be solved in polynomial time, but NP-Hard problems have not been proven reducable to an NP-Complete problem.
I knew I took that horrible algorithms course for a reason...
Quote from: Sidoh on August 27, 2006, 02:16:47 PM
I have heard of proposed methods to create programmable quantum computers, but I am not sure that it was a practical method of doing so.
I highly suspect that programmable quantum computing will be seen in a desktop environment sometime in the future.
it really depends on the manufacturing cost, which I have no idea what it is.