It's because of the binary system, it's more logical to have and easier to design a system that follows the 1, 2, 4, 8, 16 etc. route. That's not to say it's impossible to have a 6 core processor but it doesn't make much sense as you'll still need to use 3 bits to represent the processors regardless:
0: 000
1: 001
2: 010
3: 011
4: 100
5: 101
6: 110
7: 111
If you had six processors you simply wouldn't use 6 and 7 but to reach 4 and 5 (which are actually 5 and 6 as we start at 0) you'd still need the extra bit to address these processors.
Quad core processors...
Not sure what you mean, essentially a core is little different to a processor in terms of what it provides, that is a dual core processor is just a tidier, more compact and more efficient form of a plain old dual processor system. You can get dual dual-core processor systems if that's what you mean such that you have 2 processors with 2 cores each? The same will be true of quad cores - even then you're working in the good old 2, 4, 8, 16 route in terms of the number of processors though.
It's probably also worth noting that things inside processors themselves are also far from simple nowadays in terms of performance, in the older pentium days and before (pre-MMX) it was just all about the number of clock cycles a processor could do pretty much. Each clock cycle referred to an instruction to be performed, an 100mhz system could process twice as much data as a 50mhz system (although that doesn't necessarily mean software would run twice as fast but that's a different story). With MMX and now SSE, SSE2, SSE3, SSE4 and SSSE in processors amongst other things a single cycle can actually perform an operation on multiple sets of data (hence the name given to these technologies - SIMD, Single Instruction, Multiple Data). Pre-SIMD if a processor wanted to translate a 3D coordinate it would have to use up 3 clock cycles:
translate x
translate y
translate z
With SIMD, it can carry out the translate instruction (that isn't a real instruction, it's just as an example, if you're really geeky enough to be interested look up the x86 & SIMD technology instruction sets) in one clock cycle ala:
translate x y z
It's this kind of technology that makes graphics cards work too - they're specialised for processing graphics related data such as 3D coordinate or texture mapping algorithms and such although now they're becoming much more generic such that you can perform physics calculations and such on the newer GPUs also. Having solutions like SIMD means a processor that runs at 100mhz might run slower than one running at 50mhz if we're carrying out 4 instructions per cycle and the 100mhz is only running one, this is of course an unrealistic scenario as not all data is paired up with 3 other sets of data that require equivalent operations.
The point is that pumping out more mhz and more cores isn't necessarily the best solution in the long term, this is essentially the brute force route and the brute force route does have it's boundaries in that if you just keep pumping mhz into the problem you're going to end up with a processor which runs as hot as the surface of the sun
This is the underlying reason Intel has switched from NetBurst architecture which saw them through the last 5 - 10 years or so to the new Core architecture.
Much of what I've said is an oversimplification, there was more to the older processors than just clock speed even then but this illustrates the main concepts in performance computing such that you can understand there are other and better ways of solving the problem of faster computing and to be fair, multiple cores in themselves are an example of this so long as the industry doesn't get too carried away with cores such that we have too many needlessly generating heat.
Of course if you're really daring, have a look into quantum computing
It's probably also worth noting that things inside processors themselves are also far from simple nowadays in terms of performance, in the older pentium days and before (pre-MMX) it was just all about the number of clock cycles a processor could do pretty much. Each clock cycle referred to an instruction to be performed, an 100mhz system could process twice as much data as a 50mhz system (although that doesn't necessarily mean software would run twice as fast but that's a different story). With MMX and now SSE, SSE2, SSE3, SSE4 and SSSE in processors amongst other things a single cycle can actually perform an operation on multiple sets of data (hence the name given to these technologies - SIMD, Single Instruction, Multiple Data). Pre-SIMD if a processor wanted to translate a 3D coordinate it would have to use up 3 clock cycles:
translate x
translate y
translate z
With SIMD, it can carry out the translate instruction (that isn't a real instruction, it's just as an example, if you're really geeky enough to be interested look up the x86 & SIMD technology instruction sets) in one clock cycle ala:
translate x y z
It's this kind of technology that makes graphics cards work too - they're specialised for processing graphics related data such as 3D coordinate or texture mapping algorithms and such although now they're becoming much more generic such that you can perform physics calculations and such on the newer GPUs also. Having solutions like SIMD means a processor that runs at 100mhz might run slower than one running at 50mhz if we're carrying out 4 instructions per cycle and the 100mhz is only running one, this is of course an unrealistic scenario as not all data is paired up with 3 other sets of data that require equivalent operations.
The point is that pumping out more mhz and more cores isn't necessarily the best solution in the long term, this is essentially the brute force route and the brute force route does have it's boundaries in that if you just keep pumping mhz into the problem you're going to end up with a processor which runs as hot as the surface of the sun
This is the underlying reason Intel has switched from NetBurst architecture which saw them through the last 5 - 10 years or so to the new Core architecture.Much of what I've said is an oversimplification, there was more to the older processors than just clock speed even then but this illustrates the main concepts in performance computing such that you can understand there are other and better ways of solving the problem of faster computing and to be fair, multiple cores in themselves are an example of this so long as the industry doesn't get too carried away with cores such that we have too many needlessly generating heat.
Of course if you're really daring, have a look into quantum computing
OFFICER XEST - PROTECTING YOU AGAINST FORUM CRIME
Che Xefan, el presidente.
Che Xefan, el presidente.
Yeah, I'm not sure the easiest way to use one either tbh seeing as they apparently work better turned off...Gandelf wrote:I had actually considered looking into going down the route of getting a quantum computer, but decided against it because I would never be certain exactly where my computer would be when I came to use it!!
OFFICER XEST - PROTECTING YOU AGAINST FORUM CRIME
Che Xefan, el presidente.
Che Xefan, el presidente.
Xest wrote:Yeah, I'm not sure the easiest way to use one either tbh seeing as they apparently work better turned off...
It could be the solution to energy consumption... as more and more computers come into use, quantum PCs would require less (or even no) energy. Maybe someone could invent a quantum engine for motor cars, or quantum power stations even? Then we could solve the entire world's energy problems in a single stroke, as well as provide a solution for preventing greenhouse gases from changing the global climate! Who knows?
I'm sure there's easier ways of solving the problems of greenhouse gasses, nuclear power stations are by far the best option right now but as we wont let every nation have them it's not something that'd work worldwide - I can't imagine Iran would be willing to buy power off us from our nuclear power stations somehow
OFFICER XEST - PROTECTING YOU AGAINST FORUM CRIME
Che Xefan, el presidente.
Che Xefan, el presidente.
I heard the Europeans (or "us and that lot on the continent") are working on a power station that works by nuclear fusion. If they succeed, it will solve a lot of energy problems, without the associated danger of fission reactors. It would also be completely environmental.
Some scientists I believe, are trying to solve the problem of creating "cold" fusion reactors. For normal fusion, as you are no doubt aware, we would require temperatures that are found only in the Sun, but some people believe that "sono-luminescence" may be a way to achieve fusion at considerably lower temperatures, although so far there is some doubt if it's actually been achieved, primarily due to the accuracy of the equipment used to measure the fusion event.
Some scientists I believe, are trying to solve the problem of creating "cold" fusion reactors. For normal fusion, as you are no doubt aware, we would require temperatures that are found only in the Sun, but some people believe that "sono-luminescence" may be a way to achieve fusion at considerably lower temperatures, although so far there is some doubt if it's actually been achieved, primarily due to the accuracy of the equipment used to measure the fusion event.
I am not sure how much of a percentage power use is for computers, but reversible computing can in theory operate without using any power. Programming a reversible computer is a little more difficult though. You can't say a=a+b, you have to say something like (a,b) becomes (a,a+b). It is then possible to work out the input from the output.
Quantum computers must use reversible logic (for the quantum part), so the it would use very little power. However, the system for isolating the quantum computer would likely overwhelm any savings in energy.
Quantum computers must use reversible logic (for the quantum part), so the it would use very little power. However, the system for isolating the quantum computer would likely overwhelm any savings in energy.
Prydwen
Lairiodd Level 50 Nightshade and Legendary Grandmaster Smith (1065) check prices here
Lairirian Level 50 Mana Mentalist and Legendary Spellcrafter (TDD)
Lairgreybark Level 50 Arb Animist
Lairmindlock Level 50 Bard (TDD)
Camlann
Lairthall Level 35+ Friar
Stocking one 99% of most of the useful spellcrafting gems at Houses 3304 and 3306
Over 150 gems at 99% stocked
Lairiodd Level 50 Nightshade and Legendary Grandmaster Smith (1065) check prices here
Lairirian Level 50 Mana Mentalist and Legendary Spellcrafter (TDD)
Lairgreybark Level 50 Arb Animist
Lairmindlock Level 50 Bard (TDD)
Camlann
Lairthall Level 35+ Friar
Stocking one 99% of most of the useful spellcrafting gems at Houses 3304 and 3306
Over 150 gems at 99% stocked