|
| Thread: Your PC needs an upgrade. :P | |
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
posted June 28, 2006 02:37 AM |
|
|
Your PC needs an upgrade. :P
I received an e-mail from a friend of mine saying that current CPUs run at 500 GHz....
Do you believe it or not? Check this out!
I also did some googling. I believe the initial announcement is this.
____________
The empty set
|
|
friendofgunnar
Honorable
Legendary Hero
able to speed up time
|
posted June 28, 2006 03:01 AM |
|
|
You know pardon me for being pessimistic but software designers are just going to think of a way to bloat it up so that programs run just as slow as ever. I've seen this several my entire life.
We need better software, not faster computers
|
|
Roland
Adventuring Hero
Guardian of light
|
|
posted June 28, 2006 06:56 PM |
|
Edited by Roland at 18:58, 28 Jun 2006.
|
500 GHZ
Yes I read a little about it on another site.
Since I'm not an expert no comment for now.
But I agree with our friend, that :
Better software we need
yes surely indeed
____________
"all I know is I know nothing" Socrates.
"This nation will not thrive unless philosophers rule or the governors philosophize." Plato
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
|
posted June 28, 2006 09:13 PM |
|
|
Users also have a share ...
Probably I agree with both of you that what we need is good software. But in many areas I think we have the best software we can get... Besides, every time someone develops a software, he (the company) does so, so that certain needs and requirements are fullfilled. Sometimes even simple problems are not so easy to be sovled (though they can be posed in a very simple way), or, if something is solved the easy way then you reach the problem that the software is (a bit?) crap since it takes lots of time for something to be computed.
E.g., the way we learnt to multiply two integers at primary school is not the fastest as the length (number of decimal digits) of the number increases. However, this problem belongs to a class of problems that are characterized that are easily solvable.
On the other hand, the other main category has problems that are not easily solvable (at least up to today), but if someone (God?) whispers you a possible solution to the problem, then you are able to verify easily if this is actually a solution or not of your problem. First class of problems is called P and second one NP. Actually, this is the greatest open question in Theoretical Computer Science and one of the 7 greatest in Mathematics in general. For more info on these problems, check Million Dollar Babies, aka Millenium Prize Problems. 
Point is, that most (in numbers ) beautiful problems belong to class NP. And not only that, but usually, users or companies have outrageous demands on the time required to solve their problems. For example, one of the problems with the greatest reputation in NP is Travelling Salesman Problem yet, although in the last decades people have found better solutions (in the number of steps required to solve the problem) still there has not been the breakthrough we want mainly because we have to prove the P = NP conjecture (while most scientists believe and are towards the opposite direction). On the other hand, an idea that recently is in the air (and seems to have more chances than P = NP conjecture) is quantum computation, which changes dramatically the definition of computation. But only time will tell on this as well ...
____________
The empty set
|
|
Ecoris
Promising
Supreme Hero
|
|
posted June 28, 2006 10:27 PM |
|
|
Yes, if quantum computers are developed they will make the impossible possible. Consider integer factorization. Some guy already wrote an algorithm that can factorize integers in polynomial time! Say goodbye to RSA...
OFF TOPIC:
WHich avatar should I choose:
Smaug
or this frog
or something totally different?
|
|
Roland
Adventuring Hero
Guardian of light
|
|
posted June 30, 2006 09:20 AM |
|
Edited by Roland at 09:26, 30 Jun 2006.
|
Kalimera dimis, thank you for your math lesson.
But I have to say I left the school long time ago.
And I wasn’t that good with mathematic. 
One example about bad software:
Creative Labs, they have most of the time great sound cards, and on the other hand mostly Bad drivers to work with.
Back to CPU news, AMD is developing a quad core CPU witch might consume the same power of dual core ones, see:
Here
Off topic
for ecoris, Roland recommends that you choose your avatar for a reason not randomly, so if you are a nature friend the frog might be a good idea.
____________
"all I know is I know nothing" Socrates.
"This nation will not thrive unless philosophers rule or the governors philosophize." Plato
|
|
Binabik
Responsible
Legendary Hero
|
|
posted June 30, 2006 10:45 AM |
|
|
Quote:
I received an e-mail from a friend of mine saying that current CPUs run at 500 GHz....
It's not a CPU, it's only a single transistor. I'd like to see them put a billion of them in a compact CPU and do anything with it other than boil that liquid helium. Even if a CPU ran that fast, the speed would only be internal. Unless/until some unknown technology is discovered, you could never interface with the CPU at anything close to that speed.
Even modern PCs are pushing the limit of the speed running on the motherboard itself. On circuit boards, at GHz+ speeds, you'd better know what you are doing designing the board. Actually anything over 100 MHz, you have to start being real careful with the circuit board design. At GHz+ speeds, you are fighting against all kinds of laws of physics. Electricity doesn't like going that fast. Well, electricity loves going fast. It just doesn't like being told what to do at those kinds of speeds. It wants to rebel and do something other than what you want it to do.
With digital electronics, the big thing that limits the speed isn't so much the frequency, but the rise times, or transition time from a low state to a high state. The equivilent speed the hardware "sees" is several harmonics higher than the actual frequency.
____________
|
|
Ecoris
Promising
Supreme Hero
|
|
posted June 30, 2006 01:50 PM |
|
Edited by Ecoris at 13:51, 30 Jun 2006.
|
|
The speed of electricity is constant, right? I've been told the limits are actually the size of the chips and cooling. With very high clock rates the chips have to be very small in order for the signal to reach its destination before the next tick of the clock. With 500 GHz a signal has only got 1 / 500*10^9 seconds to reach its destination. Even ligth will only travel 0.06 cm during that time.
|
|
Binabik
Responsible
Legendary Hero
|
|
posted July 02, 2006 01:27 AM |
|
|
I don't remember exactly the speed of electricity. It's less than the speed of light. I think maybe something like 2/3 the speed of light. If I remember right, the "rule of thumb" for the speed of electricity on a circuit board is approx 15cm/ns. So on the motherboard of your computer it might take 2 nanoseconds to get from one end to the other.
So the shorter the distance, the faster it gets there. So yes, making the chip smaller means things happen faster. But getting there fast is the easy part. There are other limitations. Imagine a car that has instant acceleration. Even though it accelerates instantly, it still takes time to push the gas pedal and go through all the mechanical linkage to tell the car it's supposed to go fast.
In electronics, the change from a logic 0 to a logic 1 means a change from zero volts to some higher voltage. As an example I'll use a slow clock rate and the old style 5 volt system. At 1 MHz, the inverse is 1 us. In 1 us, it must go from 0V to 5V, stay there long enough to be recognized as a logic 1, then go back to 0V and stay there long enough to be a valid logic 0. In the theoretical model, it goes from 0V to 5V instantly. In reality it takes time. This is what I referred to in my last post as the "rise time".
Of necessity, very high frequencies must also have fast rise times. But a slow frequency can also have a fast rise time. It can take 1 nanosecond to go from 0V to 5V, then stay there for a year before it goes back to 0V (that's not a very fast frequency). It's the 1 nanosecond you are more concerned about, not the year.
I don't know much about the internals of CPUs. But on a circuit board, any time you have a rise time faster than 1ns, you have to be VERY careful. Electricity doesn't like change, it just wants to go, and keep going. It's for a different reason, but you can compare this to momentum with physical objects. And like momentum, the faster the change in electricity, the less it likes it.
There are a couple main things you have to fight in high speed electronics. One of them is inductance. Inductance occurs because of magnetism. If you take the north pole of two magnets and try to touch them, they will push apart. Any time electricity travels through a conductor, it creates a magnetic field. There's a lot going on in electronics and all these magnetic fields cross each other. They do NOT like that. When the flow of electricity is constant, the magnetic fields are not really a problem. But when the flow of electricity is rapidly changing, it becomes a very big problem. As the speeds continue to increase, the more of a problem this is.
Electricity just doesn't like change. And with modern digital systems, it's changing billions of times a second.
I was going to go into the other main reason high speed causes problems. And it's more interesting in my opinion. But I'm sure this is already way more info than you want. (and maybe you know all this anyway)
____________
|
|
dimis
Responsible
Supreme Hero
Digitally signed by FoG
|
|
posted July 20, 2006 12:11 PM |
|
|
Sorry guys for not participating for that long. Perhaps this is the story of my life: Starting things and leaving them unfinished!
Anyway, I am very optimistic for the future of computers in general. First of all, I think that games and their "minimum"-requirements is the major reason for pc-upgrades. Fortunately (?), we are in 3D-era already and I do not expect that the dimension will increase. On the other hand, novelties on algorithms and cpu-architectures will still appear which in turn is going to have a dramatic impact on human-computer interaction. In other words, one would upgrade his pc so that a game simply runs faster, while up to now one wouldn't be able to run a game if his pc couldn't meet minimum requirements. Moreover, as time goes by and one buys a new pc, the "speed-up" gain from a further (future) pc upgrade will be less recognizable by a simple user/gamer.
Now, regarding the drivers-problem for hardware peripherals, I think one should be patient. After all most problems with the drivers appear when someone wants to be on the cutting-edge of technology. But this is a usual tradeoff between stability/reliability and cutting-edge technology. Again, I believe that this is a problem induced by pc-gaming industry.
Seems to me that I love games but I generally hate gaming companies!
offtopic @ Ecoris:
I really don't know which one you should choose as avatar. Both seem very nice. You know, I am having a similar problem the last ... 18 months ... and as you realize I 've not ended up to a specific avatar (refer to my second sentence on this post)!
____________
The empty set
|
| |
|
|
