[Neojag]

OK, Metalmurphy told me it was completely useless to buy a processor that has more than 1 GHz because it was a total waste of money. Why? Well, because 1 GHz is like loght speed and as our closest friend einstein told, nothing is faster than light, and the CPU would never reach that speed.

ok, i personally dont believe in this, mainly i dont belive that in Nasa, or to make the FF movie they used lots of 1GHz CPUs, and if it was, why do they keep developping faster CPUs (talking about the MHz, not the cache or anything)
but maybe im completely wrong! who knows...

so??

 

[Arpanet]

It's true that using today's technology, processors aren't going to get much faster. There's only so much stuff you can pack onto one microchip... and unless we develop a completely new system of making processors, we'll be hitting a wall in terms of speed soon. However, I don't think 1GHz is that wall. If you look at benchmarks of the new P4 2GHz vs. the P4 1GHz, I think you'll see a big difference... although not as much as, say, a P2 200 MHz and a P4 1GHz.

 

[Neojag]

yup, thats exactly my point!

 

[Betamax]

Damn if only scotty had have known that he wouldn't have had to mess around with all that antimatter/matter reaction stuff (its rather explosive you know)

 

[Neojag]

ur not answerting to the question

kewl, my pc is running for about 8 hoours and it still hassssssnt crashed
im amazed, lolz

what time is it there in N ireland Betamx?

 

[takwu]

well, i wouldn't be so sure about the 1GHz = light speed theory.

the clock speed, in this case 1GHz, only means there are one trillion cycles in a second within the CPU. how fast the electrons travel in the cpu, has some affects on the clock speed limit, but is definitely not the ONLY factor.
for instance, by shortening the paths in a circuit, the electrons can get to their destinations in less time even if they're traveling at the same speed as before, thus allowing a faster clock speed.

and altho i'm not sure, i think electrons traveling thru a conductor is still a lot slower than the light speed. i heard that theoretically an optic circuit would be a whole lot faster than an electronic circuit, even with today's advances in electronics technology.

 

[Betamax]

Its about 3:15 am

Who needs sleep anyway?

ps thats nothin'. This machine here has been on now since 11am yesturday.
Win 2K M8. Win 2K.

 

[Neojag]

that light speed thing was a metafor PP just a way to say that the speed cant go up anymore, lolz

its the same time as here betamax.
damn, i didnt know w2k was THAT solid...

 

[Arpanet]

Pfft... my friend runs a Hotline server... Win2k stays up 3+ months at a time for him... whereas Win98 rarely goes more than 10 hours for me without dying horribly. :fingers:

 

[takwu]

man, you got me
i totally made a fool out of takwu :emb:
lets hope no one noticed

 

[Neojag]

lololol
no prob dude ur just a tad too scientific

 

[Metalmurphy]

lol, I didn't though u would put this in here Neojag! Well i might as well defend my self!

What I said is that altough the proceessor supports speed up to 1Ghz it is totaly useless cause the electricity on the circuts ain't gonna go faster than that light speed and 1Ghz IS light speed! And u know how the processing of the 1 and 0s works don't u? 1 is for existence of electricity and 0 for non existence, so, the electrity won't go faster then 1Ghz and so neither the processor! About nasa PC and all that, as I had already tould u, they have multi processors and split up the processing task, but all of them will go at max 1Ghz!!

Anyway... i could be completly wrong! But I don't think so...

 

[KillerShots]

A bit to chew on...

There's nothing 'instant' about a 1 or a 0, ESPECIALLY in a processor. You have to think of the other components involved. To get a 1 or a 0, the circuit goes through an amplification period that, at a zoomed out level (say about 10 ns) for a processor, looks like a square wave (1s and 0s). However, if you start zooming in (even to around 5 ns), a lot of electronics will start looking like a sine wave... and that isn't good for a digital device. I'm not talking about the noise, either. I'm talking about the main surge upward no longer being vertical...that's not a problem with the speed of electricity so much as the conductance of the material and whether there's devices that need charging (i.e. caps), etc. The trick is to cause the charge up to 5v or whatever your high is in a quicker period of time, and vice versa. In this area, there is PLENTY of room for improvement.

 

[Metalmurphy]

Ok... i said a lot that I didn't understand

Do u mean that processing can be faster then light speed?

Sorry for my ignorance

 

[KillerShots]

No, processing (at least with current technology and theories) cannot go faster than light speed. Current processing is not even hindered by light speed, however, it is hindered by the relatively low conductance of materials used to bring an electric signal from one spot to another, and with various devices waiting for other devices to finish (i.e., capacitors charging - could take several ns to charge up for an extremely small one (thank god we don't need large ones)). Also, if I remember correctly, RAM is nothing more than a HUGE series of capacitors storing energy (charged = 1, not charged = 0...power cut, no longer charged). Right there is a huge bottleneck, and bear in mind that a processor needs to use RAM of some sort for EVERY SINGLE operation (internal cache RAM or external RAM). Now, for read-only purposes, this system is fine - the caps are charged, they send out a flatlined 1 or 0 that something sees whenever it checks, getting rid of the charging problem...however, when you CHANGE the value to either a 1 or a 0 from another state, it will take time. This, too, can be improved. If the processor starts using superconductors (none have yet been created), the speed will (theoreticall) increase exponentially.

Also, how is it that you think that a signal at 1 Ghz must be at light speed? There are Terrahertz signals and beyond that exist, we aren't anywhere near the power of a light-speed processor.

 

[Metalmurphy]

Then my theory is somehow correct... the only thing wrong was that i thought that 1Ghz was = to light speed... sorry

I'm reading on the wrong sources

Thks KillerShots!! U really know about this stuff!

 

[Neojag]

eheheheheh...

 

[takwu]

lets hope i'm not making a fool of myself again this time:

KillerShots,
i think memory modules that use capacitors are called Dynamic Memory, and when used in RAM, called DRAM. i think they're used only in main memory (well ok, recent advances in certain types of DRAM made them appear on video cards). yes they take time to read/write/recharge, but they take less space and power to maintain.

and not every single instruction needs to use the RAM, or cache. actually i think very few of the executed instrustions load main memory. most are operations on _registers_, which are the units of memory within the cpu, made of flip-flops.

 

[KillerShots]

Originally posted by takwu
lets hope i'm not making a fool of myself again this time:

KillerShots,
i think memory modules that use capacitors are called Dynamic Memory, and when used in RAM, called DRAM. i think they're used only in main memory (well ok, recent advances in certain types of DRAM made them appear on video cards). yes they take time to read/write/recharge, but they take less space and power to maintain.

That could well be. I don't know really what the newer RAM uses, and if they got away from using caps, then...GREAT! One bottleneck down .

and not every single instruction needs to use the RAM, or cache. actually i think very few of the executed instrustions load main memory. most are operations on _registers_, which are the units of memory within the cpu, made of flip-flops.

Remembering my classes, I stand corrected. Internal registers do indeed use flip-flops which is simply transistor-style logic...meaning no charging/discharging. It's my turn to look like a fool

 

[Betamax]

He got confused with the terminology

Under S.I. Standards the unit Hz is a measurement of frequency and is the same unit as per second (or s^-1).

Thus any quantity that is an arbitary number that is divided by time results in a frequency.

In the case of the microprocessor the clock speed is rated by the number of calculations that it can perform in a give second. As the no. of calculations is an arbitary number this reults in a frequency hence microprossor speeds are rated in Hz.

Still with me? Good. Now the problem occurs when we look at a COMPLETELY different field of physics that of Electromagnetics. Em Radiation moves in waves. As we cannot measure the lenght of the whole wave (as it would in essance be infinate) it makes sense that we measure the lenght of one single cycle or the wave length. Now EM waves move at the speed of light and when the speed is divided by the wave lenght the resultant unit is in fact (s^-1) or a frequency.

So in other words although in both case we have a frequency they are in fact not related at all!

Its a good job that they aren't related. If this was the case then you processor would be emmitting enough gamma radiation for all of us to be turning green and ripping are shirts off!