Just out of curiosity, why do they list processors, memory and buss speeds in terms of Hz. I have always considered Hz as an Analog signal (AC voltage sine waves), and DC voltage as a digital signal ( the voltage never varies, well you either have full +VDC, 0VDC, or perhaps full -VDC if the leads are reversed). So if computers process information using bits, ( digital signals, 0's and 1's, "on" or "off" "true" or "false") why do they use Hz to represent the speed of of these components?

Hz means... times per second.
in anologue (like sound haves) Hz is specifyinh the number of waves (sine ?) per second.

processors also run in cycles.... the "Fetch -> Decode -> Execute" cycle.
in this case Hz specifys the number of processor cyles the processor performes each second.

If you're dieing to know a little more detail, the processor cycles are derived from digital waveforms. Like you mentioned a digital signal from a DC source is one of two values: +Vdc and 0Vdc. Sometimes -Vdc is used instead of 0Vdc for differential digital circuits. Anyways, with a digital signal, you create a square wave, right? Something like this:
That type of signal is usually referred to as the "clock" and the amount of time it takes to go from one "rising edge" to the next (the time difference between points A and B) is the period of the waveform just as it is for an analog signal. Take that period and turn it into a frequency and that's where the MHz and GHz numbers come from.

None of the components used in digital circuits respond instantaneously; they need time to adjust. That is the prime use of the clock signal; it serves to synchronize all the digital circuits in a given portion of the system. Typically, the rising edge of the clock is considered the event that causes the next step in processing. So digital circuits are designed around time constraints to finish one "step" of a process before the next rising edge occurs. The faster your clock, the more steps your computer will take toward whatever goal, but it's also that much more difficult to design circuits to finish what they need to do in that shorter amount of time. Consequently, if you ever talk to a digital logic designer, they'll probably tell you that meeting timing requirements is the number one hurdle to getting a product out the door; not coming up with the idea, not coding the logic, but making sure the circuits do what they need to do fast enough. Heh... unfortunately, most people (translation: marketing) think that all you need to do is "just double the clock speed" to make things go faster. Tell that to a logic design team to see them groan/wince/cry.

i wonder where people got the idea of more clock ticks that faster it goes?

also the circuits are made to perform there task X amount of times per click, if you have a 1Ghz system that only performs each task 1 time a tick. and a 800MHz system that performs its task 3 times a tick, what do you think will run faster?

i like AMD's method of giving numbers to processors, the higher the number, the more power, but also can mean slower clock in some cases... Intel is all about telling people about there high and fancy clock speeds

It is a mistake that many people make to think that CPU frequency is the most important parameter for the speed of a computer. First of all, some things aren't done on the CPU, like hardware-accelerated graphics. Secondly, all results of what the CPU does are stored in the RAM, and the RAM is usually a lot slower (my RAM is 333 MHz, while my CPU is 2600+ MHz). So the speed of the RAM is important, but also the amount of cache on the CPU, as data that is cached doesn't necessarily have to be stored to the RAM every cycle. Of course, a smart design of the CPU can help, as mentioned by SciYro, because then it can do multiple things in one cycle. Then, the amount of RAM matters, because modern applications use huge amounts of it and you don't want it to be swapped to your hard drive. And, especially in Linux, if you have lots of RAM, disk access will be cached in your RAM. Linux + huge amounts of RAM = speed. Now I could also start talking about OSes and software, but that would bring me too far off-topic.

I believe the reason they do that is so that people don't see that and AMD runs at 2.2ghz and an Intel is running at 3ghz and just buy the Intel one. This illusion that the only thing that matters is the clock speed must be causing AMD some problems when it comes to selling stuff, so they just give it a number which indicates the approximate equivalent Intel clock speed.

Reading Dark_Helmets description has brought back all my first year degree electronics courses flooding back, I'm not a massive fan of anything low level. Although I did pretty well in the electronics modules... unlike ARM assembly, stupid ARM.

sciyro hit it on the head, its how much the processor does with each clock tick, rather than how fast the clock goes....if you chk some of the results from "benchmark" tests that various pc mags and sites do, you find that AMD does many things equal or better, with a slower clock speed, like opening a 15k line Word document, hehe

From a "brute force" standpoint, increasing the clock speed is the most direct method of increasing performance. Let's say it takes 10 "steps" to perform a typical floating point division operation. Let's also say you have two identical processors with one exception: one runs with a 400MHz clock, and the other at 800MHz. The processor with the 800MHz clock will finish the division in half the time needed by the 400MHz clock. That is simply raw speed and is roughly analogous to a car race. If you put two cars on a track with one capable of going 60 mph and one at 30 mph, then the 60 mph car will win.

Now, beyond that you get into the world of "tricks". There are things you can do to help speed things up. Typically the most widely used trick is parallel processing (not in the traditional sense of the computer's actual processor). Say in the floating point division example you could separate the process of calculating the division into two distinct, independent parts of 5 steps each. To give the illusion of faster raw speed, you could build two separate circuits; one to handle each 5-step part. (note: it's not quite that simple, but that's a discussion for another day) If you modified the 400MHz processor with this new floating point design and then compared it against the original 800MHz processor, then they would behave identically. At least, where floating point division is concerned. To use the car race analogy again, it would be as if the race coordinator said "Get from point A to point B any way possible." The 60mph car sees there's a nice paved track available and takes it, but the 30mph car realizes that there's a shortcut that's half as long as the track and takes it. Both cars arrive at the finish line at the same time.

The thing is, you're not quite comparing apples-to-apples though. If you equipped the 800MHz processor with the new floating point design, it would again be twice as fast as the 400MHz processor in all respects.

I'm not endorsing Intel's or AMD's method of labeling processors. Each method has it's own benefits and drawbacks. For Intel, they are reporting raw speed and you can't argue with that. If they tweaked their design like AMD, then they could probably be even faster. AMD decided to go the route of squeezing as much performance out of each clock cycle as possible, and they could be faster if their circuits ran at the same native clock speeds Intel's does. It's just two different sides of the same coin.

Bash intel all your want but the fact remains they are the ones solely responsible for the processors you are using now. Second they make fine processors that outperform AMD's. No matter how much you want to believe AMD is always faster they are not.

amd isnt faster, i dont think anyone here said amd has a faster clock than the intels....?

I would like to say thanks all, for the great answers, you folks sure have an easy way of explaining things.

So what causes these "pulses"? the quartz crystal (if my memory serves me right i heard it was quartz)

I just find it so amazing that something can vibrate billions of times in a second, but than again putting millions and millions of transistors on a
2" X 2" piece of "silicon" is hard to believe too!!!

Technology, women, and linux...once you figure 'em out...its a beautiful thing!!!

PS My apologizes to any women out there i know you could have just as easily substituted men!

2in ?!!?!, last time i checked the actual place where the millions (i believe it was either P3 or P4 that has about 25 million transistors, or maybe it was 60 million for the P4? i forget)

ok, back to the point: the place where all that computing power is at is the die, a little metal tab in the middle of the processor, basically they stuck 24-60 million transistors onto a piece(s) of silicon (the s is because there are multiple levels arnt there?) roughly the same size as your thumb nail (thats trimmed)

Stack: no one here is bashing Intel, they make great processors for games (if your hardware can keep up ), and no one is denning that they made the modern processor

but for the performance/cost ratio AMD is better then Intel (unless you want to do big time gaming, a P4 with a 800MHz FSB will be a great addition to your PC )

unless you want 64-bit processing, then i think AMD's $800 processor will be better (yea, like i have that much money!)

You've got it. Usually at the heart of each clock signal is a crystal with a known vibration frequency. Beyond that, I'm not exactly sure how that is translated into an electrical pulse. That gets into materials science, and that's the realm of the brainiacs at Intel and IBM.

It is staggering to think of the sheer number of times you get a pulse with today's processor speeds. 3 billion pulses per second when talking about a 3GHz clock. As for transistors, I worked on a part that had over 3 million gates (and a "gate" is roughly equivalent to 2.6 transistors) manufactured into a chip no bigger than the size of your pinky's fingernail. We're talking a "package size" measured in millimeters square.

The die is made of silicon. I honestly don't remember it's classification (metal or non-metal and I'm too lazy to dig up a copy of the periodic table). It has both electrically insulating and conducting properties under the right circumstances. That's where the term "semiconductor" comes from. As a bit of trivia, each die is cut from a larger piece of silicon called a wafer (usually a thin, circular disc). Each die cut from the wafer is tested to make sure it's electrical properties fall within acceptable ranges. When I studied all this, the approval rate was not impressive. I think it was somewhere around 70% to 80% fell within the needed ranges. After that, then transistors are placed on the die. Then, the die goes through testing again to make sure it operates correctly. Again, approval rate was not stellar. I think it was roughly the same, but slightly lower (60% to 70%).

Layering is the term given to stacking digital circuits on top of each other. I know it's a common practice for circuit boards (like the motherboard, expansion cards, and pretty much anything electronic and green ). I cannot say with certainty that it's performed on the actual die. I wouldn't think so, because that would add another test-and-throw-away phase described above to make sure each layer was positioned correctly. When you're talking about transistors measured in micrometers, alignment is no simple task, but it may be possible.

I'd be interested to see a percentage layout of Intel or AMD's chips. I would bet that a significant chunk of the gates/transistors they use would be used to make their level 1 and level 2 on-board cache. Internal memory is a space/transistor hog for most custom made parts.

silicon is a metalloid (i think it was, either that or nonmetal, diffidently not gas or metal)

silicon is the poorest conductor known, an example: on TV they showed after they took some silicon blocks out of the oven, and they were still red hot in the middle, the people could actually touch them! (bare handed)

silicon also cant conduct electricity, it another element (i forget what one, but there are 2, one is abbreviated p, and another n (i don't know if thats there periodic obviations, but thats what they were called when i was studying transistors one day)