Whats nm?

[Guest]

How does it affect chip performance? Smaller nm means it can do more instructions, right?

And what is the proper way to ask at what level a chip is designed. Something better than "how many nm's is this chip?"

[David (PassMark)]

nm is a nanometer.

See Wikipedia for the symbols that denote decimal multiples and submultiples of the metre.

Smaller chips can contain more transistors, and run cooler, and so potentially go faster.

You could ask, "Is this a 45mn chip?", and it would be a reasonable question.

[2006]

Ok I'm not THAT dumb. But I guess it's my own fault for asking in such a noobish tone.

I know what a nanometer is, I was just wondering how processors benefit from a smaller "semiconductor design" if that's the term, hence the other question.

So basically, a chip with smaller transistors dissipates less heat? Why dont we see chips clocked at 5 ghz then?

[David (PassMark)]

Chips can be made to run at 5Ghz... but liquid nitrogen was required a few years ago. Now it can be done with just water cooling or dry ice. Do a search on Youtube.

[2006]

Are you saying 5 ghz will be possible with simple air cooling in a few years from now?

[David (PassMark)]

Do some Googling. People are already claiming to have hit 5GHz on air with the i7 CPU.

[David (PassMark)]

Further, this article,
http://www.atomicmpc.com.au/News/151892,cpu-gigahertz-battle-resumes.aspx
claims 7Ghz is obtainable for short periods.

[2006]

Short periods of time, right.

That article doesn't impress me, as that six-core AMD hasn't proven to be faster than the i7. And "Yes, the new Phenom II X4 965 Black Edition runs at 3.4GHz out of the box, a tad tick higher than the 3.33GHz Intel Core i7 975 Extreme Edition." what? The highest clocked commercial CPU in history was the Pentium 4 at 3.8 ghz, so whats so new about 3.4?

I guess my question was, since the start of the thread: does a smaller semiconductor design directly allow higher clock rates, because if so then why aren't clock speeds increasing, why do they stay at about 3 ghz?

[David (PassMark)]

I think the explaination is that better ways to improve the processor power were developed, than simply increasing clock speed. Larger caches, more cores, faster front side buses, more efficient piplines, better branch prediction, new SiMD instructions, etc, etc.... some of these other improvements probably worked counter higher clocks speeds. But the trade off was clearly worth it as current CPUs are far faster than a P4.

[2006]

I know that. In fact, the latest Core2 quad clocked at 800 mhz would be way faster than P4 at 3.8 while consuming 10 times less power. But I still can't seem to comprehend the role of a smaller semiconductor design.

Like, the P4 is 130 nm right? And the latest Core CPUs are 65/45. If the Intel Core extreme was designed at 130 nm instead of 65 then how would that affect performance? Would it dissipate more heat and take more power? Say 130 watts instead of 95? Or would it be slower at the same wattage?

[David (PassMark)]

Ok I'm not THAT dumb.

That article doesn't impress me

I know that

If you know all the answers, then you are asking the wrong questions, or asking the wrong person.

[2006]

No, I'm just receiving irrelevant answers. First off, I wasn't asking what a nanometer was, but asking what the role of smaller transistors were but I guess I worded it wrong and gave the impression I was as clueless as you assumed me to be.

Second, you said that smaller transistors take up less power, permitting a higher clock rate. So I asked how come commercial CPUs aren't increasing their clock rates then. Then you said it was because they were focused on improving other features of the CPU such as more cores. Irrelevant.

Then I ask this

Like, the P4 is 130 nm right? And the latest Core CPUs are 65/45. If the Intel Core extreme was designed at 130 nm instead of 65 then how would that affect performance? Would it dissipate more heat and take more power? Say 130 watts instead of 95? Or would it be slower at the same wattage?

to get a final confirmation if I got it right, and you tell me "you know everything." I dunno, do I? I'm asking you because I assume you know better. I'm a computer geek, but sure as hell ain't no engineer.

[David (PassMark)]

First off, I wasn't asking what a nanometer...

Then why did you title the thread "Whats nm?" if you already knew?

You might get a more detailed response if you post you questions in the Intel developer forum. We aren't semiconductor engineers.

[2006]

Then why did you title the thread "Whats nm?" if you already knew?

Because I wanted to know what the entity was that was xx nanometers, I seriously didn't know that it was a transistor. I know I worded my question wrong in my first post, but I have clarified myself numerous times, so I don't get what you're accusing me of. Instead of nitpicking, you should've simply said "I don't know exactly, ask on the Intel developer forum."

However, I don't think my question is that complicated. All I wanted to know was how chips benefit from smaller transistors. This is the fifth time I ask this.

But since you're an administrator of an important website and everything, you probably have a personal reputation to uphold and don't really like to admit there's something you don't know. I understand.

Thanks anyway.

[David (PassMark)]

Sigh.
The thread is closed.