If this is true...

[Vexxx_2007]

Moore's Law: Transistor die sizes are cut in half every 24 months. Therefore, both the number of transistors on a chip and the speed of each transistor double every 18 (or 12 or 24) months. (Gordon Moore)

Quote from sysprog.net (http://www.sysprog.net/quotlaws.html)

If that is true, Why isn't my processor fast as all hell? And furthermore, if it's true, shouldn't old processors not deteriorate in price and overall usefulness?

Kinda makes me wonder.

:think:

[xmastree]

If that is true, Why isn't my processor fast as all hell?It is
And furthermore, if it's true, shouldn't old processors not deteriorate in price and overall usefulness?
They do, so long as software gets more demanding.

[d_stilgar]

People are speculating how long it will go on. Some say we will reach a molecular limit in the next 50 years. I think we will be engineering on an atomic level in that time, but we will then re-increase the total chip size to fit all of our two and three atom transistors. I think we'll also see massive amounts of cores.

The other possibility is that we somehow break the 4 and 5 ghz limits that we've seen for the past 15 years or so. Sure people overclock like mad and do crazy things to cool the processor, but there is still a fundamental, physical limit that is keeping us from going faster.

xmastree is right though, software gets more demanding, and your processor is as fast as it's ever been.

[Omega]

Why isn't my processor fast as all hell?

It is. Oh dear god, it is. I've got a quad core CPU on this computer and it's beastly. The thing is that you don't realize it because as technology advances, so does software, and the software we utilize on our computers is made to take advantage of the speed. If you try and run Word '07 on a Pentium 1 166MHz... you'll be there a while. Try it on even my laptop's P4 2.4GHz and it loads almost instantly.


And furthermore, if it's true, shouldn't old processors not deteriorate in price and overall usefulness?

They only deteriorate in price due to better technology and they're only any "less" useful because current software is made to run on current hardware. There is some current software that uses so few resources that you could run it on old P1's and P2's, but for the most part, companies like to utilize the power they have.

[Drum Thumper]

Load up a copy of the original DOOM, or if you don't have that, get a copy of the old shareware game Scorched Earth and load it up on your shiny AMD 3000+ based mobo. Then realize that those games were designed to run on much much slower processors (the 286 chipset for Scorched Earth, a minimum of a 486DX based system for DOOM, iirc).

[Omega]

I play DOOM2 on my laptop when I have nothing better to do. =D

[Xpirate]

DrumThumper, I have done the same thing. I ran old games just to see how blistering fast the newer hardware is. I wrote an old drag race game that goes to fast to even get out of first gear on modern hardware.

[XcOM]

moden processors are stupidly fast now, this is why my current 1.8ghz can out peform my 2.6,

The GHz wars are well gone, now we are more concerned with going green and number of instructions per clock cycle, hence the above mentioned 1.8 chip can beat a 2.6 previous gen.

[Vexxx_2007]

Hmm... I kinda see your guys' points. But another question. In 20 years, will this particular old processor have like an exponentially larger amount of transistors and become exponentially faster?

I'm kinda confused and may be taking this the wrong way. But that's what it seemed like to me.

[XcOM]

to an extent yes,

as processors get faster that don't need to have double the transistors to work at double the speed of the last gen, its all about how them transistors work and how fast they can process an instruction.

As times go by CPU design changes and becomes more efficent and therefor needs less MHz to do a job so to speak.

[Drum Thumper]

I believe the cap speed is either 4 Ghz or 5 Ghz. Regardless, the true power of the future lies in expanding the other way--number of cores. xmastree posted this gem (http://www.thebestcasescenario.com/forum/showthread.php?t=15916) explaining how parallel processing works--increase the number of cores available, optimize the software, and all of a sudden the quad (or quint or octo or dodeca, or, well I think you get the idea) core becomes not only very powerful, but allows for CPU manufacturers to give us more bang for the buck.

When will we see the octocores or the dodecacores? Well, Intel just announced a 6 core server chip the other week, so as die sized get smaller (Intel is currently at 45nm iirc), the number of cores that they can place on a slab of silicon increases at a comparable rate.

So, in theory, since a game such as Scorched Earth is not (nor will it ever be) optimized for multiple cores, you should see no difference on a single core 2.8 Ghz system when compared to a multiple core system. However, if you use software that has been optimized for multiple core machines (and has been coded in such a way that it can run on a single core machine, albeit sluggishly), the difference will become night and day.

[Gizmo]

They are already designing and testing newer types of processors ever hear of a quantum computer?

Well they have made them, http://computer.howstuffworks.com/quantum-computer.htm

http://www.wired.com/science/discoveries/news/2000/03/35121

Quantum Leap in Computing
Leander Kahney Email 03.23.00
Federal researchers say they've created the most robust quantum computer ever, indicating that the concept is rapidly moving from theory to practice and could create the most powerful computing devices ever dreamed of.

If the trend of increasing performance continues, a quantum computer that triples today's fastest computers could be built in five years, according to physicist Raymond Laflamme, who helped build the world's first 7-qubit computer described in the most recent issue of Nature.

"Right now it's impossible to say if we can scale these technologies," said Laflamme, the project's lead researcher. "But if you asked me five years ago if we could build a 7-qubit computer in five years, I would have said it was impossible."

Quantum effects give sub-molecular computers great power. Still, quantum computers may never be general-purpose computing devices and are more likely to be targeted at massive number-crunching problems like encryption and decryption, searches of huge databases and simulations of quantum physical states.

While the theoretical foundations of quantum computing were set in the 1980s, scientists had been unable to build quantum computers until recently.

The first 3-qubit quantum computer was created just 18 months ago at the Department of Energy’s Los Alamos National Laboratory in New Mexico. The lab's researchers describe in the Nature paper how they used a test tube of trans-crotonic acid and a powerful nuclear magnetic resonance spectrometer to create the 7-qubit (pronounced kew-bit), or quantum bit, quantum computer.

Physicist David Wineland of the National Institute for Standards and Technology, who is working on a competing quantum computing technology, said the work was important but ultimately the NMR technology will hit a roadblock.

"It's significant because it's the most complicated system people have been able to do something interesting on," he said. "But people in the field generally feel that the final quantum computer won't be an NMR computer."

Laflamme's quantum computer was created by manipulating the nuclei of seven molecules in a test tube of trans-crotonic acid, hence 7-qubit. Like a spinning magnet, the molecules' nuclei can be lined up with electromagnetic pulses from the nuclear NMR spectromotor, which is a specialized version of the imaging devices commonly used in hospitals.

"It's like trying to manipulate needles with bulldozers," Laflamme said.

The lining up of a nucleus parallels the encoding of information in conventional computers as binary ones or zeros. However, unlike a traditional bit, which is either on or off, the nuclei are subject to the very weird laws of quantum physics that allow them to simultaneously be in multiple states. In other words, they can be a one or a zero at the same time.

Wineland said that the NMR approach will run out of steam at 15 qubits because key interactive effects between the quantum particles start to disappear.

Wineland is working on one of several different approaches to quantum computing that utilizes trapped ions instead of fluid liquids.

Laflamme compared his research to the early days of computing, when computers weighed 30 tons and were built from thousands of vacuum tubes. Back then, computer scientists predicted that one day computers would weigh only 5 tons and be built from hundreds of tubes.

"What do we have today?" Laflamme asked. "Laptops and Palm Pilots."

"On my optimistic days I think we will have quantum computers in 20, 30, 40 years maybe," he said. "On my pessimistic days, I think quantum computing is crazy."

So just wait it's about to get insane crazy real soon.