Does RAM have anything to do with the proccessor or MB?

[dwchang]

Well, that's not really comparable, since CPU clock speed is measured in cycles per second while RAM is measured in a number of binary digits. You can't really compare the two. However, you could say he has more GigaHertz in the CPU than GigaBytes in the RAM... it's still meters and liters, but at least you now have a quality to attach to your quantity.

CPU speed is starting to become meaningless anyway. A 2.0 Ghz Centrino performs like a 3.0 Ghz P4.

A 2 Ghz Centrino doesn't exist yet. They are up 1.6 Ghz and just announced a 1.8 Ghz for Q3'03. You might be talking about 'possible' benchmarks. Also a 2 Ghz doesn't perform like a 3 ghz P4. Given, I know the argument being that the Centrino architecture is far superior to that of the P4 (because...it is...if you want me to go into technical details like pipeline stages and so on, I can. Suffice to say Centrino ~ P3 and P3 > P4 architecturally...not speed). If you mean in that extrapolation of speeds, then you *may* be right. We won't know till the 2 Ghz part comes out.

[the Black Monarch]

That's odd. My friend claims that his IBM laptop ("Mr. Lucky") has a 2.0 gig Centrino... maybe IBM overclocked it and didn't tell him... or maybe he's bullshitting us...

[dwchang]

That's odd. My friend claims that his IBM laptop ("Mr. Lucky") has a 2.0 gig Centrino... maybe IBM overclocked it and didn't tell him... or maybe he's bullshitting us...

I'd go with the latter.

If he has a 2.0 Ghz Notebook, it's probably just a "normal" P4 mobile part as opposed to a Centrino.

Here is a link that has their roadmap:

Centrino

As you can see they are not even close to 2.0 Ghz and won't be for some time. In fact, since the Centrino is just a rework of the PIII core, it will be some time before they get "speed" out of it. As I stated earlier, the PIII core is a lot more efficient than the P4 core (which is one of the most laughable architectures in terms of efficiency), but at the same time cannot be clocked at higher frequencies easiliy.

Given, off-topic, but I'd rather have a more efficient core than "speed" since "speed" isn't true speed or performance. They just did that because most consumers think a bigger number means better performance. If I can do 3 "things" at 1.8 Ghz or 1 thing at 3.0 Ghz...you tell me which is "faster?" Then again, I'm getting into the old performance != speed, performance = speed * instructions/clock, which will bore the hell out of you guys.

Anyway...as you can see your friend is probably bullshitting you or doesn't know he has a P4-mobile part. I highly doubt IBM overclocked it for him since no firms do that. It's stated in all warranties that no companies (IBM, Intel, AMD, etc.) condone overclocking (not that we don't know you guys are going to do it anyway).

[the Black Monarch]

The laptop has the little Centrino sticker and a battery life of over ten hours while playing DVDs continuously, so I'm guessing it's not a P4

I'll have to talk to him about this.

[alternatefutures]

First, there is no chip called "Centrino", it's the Pentium M. Centrino is a P-M, the chipset, and wireless. Secondly, the Pentium M is only architecturally superior to the P4 if, after both have ramped as far as they can go, the Pentium M is faster and at least as stable (if you go by performance, and it does look as if the final Banias will be better than the final Northwood). If you go by power efficiency and heat disipation, then yes, the Pentium M is better (hence why it is in a laptop and the Pentium 4 isn't). I know our AMD friend likes to rate work done per clock cycle as the sole factor, but that isn't so.

[the Black Monarch]

First, there is no chip called "Centrino", it's the Pentium M. Centrino is a P-M, the chipset, and wireless.

Same fucking thing, different haircut.

Secondly, the Pentium M is only architecturally superior to the P4 if, after both have ramped as far as they can go, the Pentium M is faster and at least as stable (if you go by performance, and it does look as if the final Banias will be better than the final Northwood).

There is no "as far as they can go." You can make a chip as fast as you want if you have the right cooling system and motherboard. What you should look at is price: does a $500 Centrino outperform a $500 Pentium 4?

If you go by power efficiency and heat disipation, then yes, the Pentium M is better (hence why it is in a laptop and the Pentium 4 isn't).

Hey, my laptop uses a P4 that was designed for desktops. So the Pentium 4 IS in a laptop.

I know our AMD friend likes to rate work done per clock cycle as the sole factor, but that isn't so.

No he doesn't. He brought up clock speed as an equally important factor:

performance = speed * instructions/clock

[alternatefutures]

There is no "as far as they can go."

If that were true then we could have chips running at insanely high speeds. Significantly overclocked Pentiums and Athlons can be maintained at near freezing and still be so unstable you can't boot to Windows. Cooling is not the only limit to a chip's design. Yield factors are another thing and a good reason why Barton chips still have lower frequencies than the vapor 2800 and we have a Pentium 3.0 Ghz with an 800MHz FSB instead of a 3.06 800Mhz FSB.

Hey, my laptop uses a P4 that was designed for desktops. So the Pentium 4 IS in a laptop.

That would be the MOBILE Pentium 4, which is a lower voltage P4. The desktop version wouldn't be in a laptop unless there was a market for a laptop/heating blanket.

[dwchang]

First, there is no chip called "Centrino", it's the Pentium M. Centrino is a P-M, the chipset, and wireless. Secondly, the Pentium M is only architecturally superior to the P4 if, after both have ramped as far as they can go, the Pentium M is faster and at least as stable (if you go by performance, and it does look as if the final Banias will be better than the final Northwood). If you go by power efficiency and heat disipation, then yes, the Pentium M is better (hence why it is in a laptop and the Pentium 4 isn't).

Uhm, I hope you didn't misunderstand anything. I think I even stated that the Pentium-M architecture if superior to the P4 Architecture. Hell it has like 7 less pipeline stages. It may not reach the same frequencies, but it still the superior architecture. I might also add that the backbone of the Pentium-M architecture was the PIII, which further supports what I have said about Intel making the P4 architecture inefficient for the sake of frequency (since ppl think it's performance).

This is a well-accepted fact among the Computer Architecture community.

I know our AMD friend likes to rate work done per clock cycle as the sole factor, but that isn't so.

Uhm...yeah as Black Monarch said...the equation is:

Performance = Frequency * Instructions/clock.

Perhaps you read over it too quickly? This is a fundamental "equation" for computer performance and all Architects are taught this in their fundamental class.

The P4 focuses on frequency and is known for it's inefficiencies architecturally. Hell there are even pipeline stages where all it does it transfer data from one stage to the other...THAT'S IT. No FPU/ALU instructions, no memory reads, etc. That's probably why it's 20+ stages.

The P3 architecture if something I actually respect. It was a very solid pipeline and fairly efficient, however it couldn not go beyond a certain frequency and thus Intel opted to design the P4 with that in mind.

As for AMD, their architecture is very similar to the P3 architecture and is 13 stages where none just transfer data. This is obviously the reason why they can't reach those high frequencies, but what they do with in that time is more efficient than a P4.

This is basic architectural fact and not a bias. I even have the designs for both on my work computer, but unfortunately, I can't show you that .

Hey, my laptop uses a P4 that was designed for desktops. So the Pentium 4 IS in a laptop.

Yeah, the P4 laptops just use normal P4 chips. Basically they go through tests and one of them is a "mobile" test. If it can run at those voltages and at an adequate speed, it goes to the mobile sector. I believe alternatefutures went into this in his last post.

If that were true then we could have chips running at insanely high speeds. Significantly overclocked Pentiums and Athlons can be maintained at near freezing and still be so unstable you can't boot to Windows. Cooling is not the only limit to a chip's design. Yield factors are another thing and a good reason why Barton chips still have lower frequencies than the vapor 2800 and we have a Pentium 3.0 Ghz with an 800MHz FSB instead of a 3.06 800Mhz FSB.

Unfortunately the Black Monarch, he's right. Each architecture is limited to a certain speed. Namely, this is based on the transistor technology that each chip is on. If the transistors can only "switch" at say 3 Ghz., that is the limit of the chip-technology. Oh and of course if the technology itself can support it. I mean if you had 1 pipeline stage and it had a TON of logic...there's no way you could get 10 Ghz or something. Obviously each architecture itself is limited by whatever logic is used and that logic is made up of the transistors (as said about the switching speed).

Make sense?

[alternatefutures]

If the point of a CPU was just efficiency (which does play a huge part in the laptop arena), then I would agree with your definition of superior architecture. However, if the point of a CPU is performance (as in workstations), then efficiency, while good, does not a superior architecture make. Right now I'd say Intel's and AMD's current highend 32-bit offerings are at parity overall, but it depends on what you need to accomplish. If you're an engineer working with Autocad then the Athlon has the superior architecture. If you're a special effects guy working on 3DStudio Max, then the P4 has the superior architecture even though it is less efficient per-clock. The question is, can you, as sales likes to say, make up the efficiency difference in bulk (and hopefully some extra). If you can, then I don't see how you can call it inferior. It might not be as elegant, but for something that's pretty much invisible to the end-user, that's not of much concern.

[dwchang]

If the point of a CPU was just efficiency (which does play a huge part in the laptop arena), then I would agree with your definition of superior architecture. However, if the point of a CPU is performance (as in workstations), then efficiency, while good, does not a superior architecture make. Right now I'd say Intel's and AMD's current highend 32-bit offerings are at parity overall, but it depends on what you need to accomplish. If you're an engineer working with Autocad then the Athlon has the superior architecture. If you're a special effects guy working on 3DStudio Max, then the P4 has the superior architecture even though it is less efficient per-clock. The question is, can you, as sales likes to say, make up the efficiency difference in bulk (and hopefully some extra). If you can, then I don't see how you can call it inferior. It might not be as elegant, but for something that's pretty much invisible to the end-user, that's not of much concern.

Y'see you're getting into the argument about specific operations and programs. This has nothing to do with a superior architecture. The superior architecture is that which can do the most in the least amount of time. Thus performance = speed * instruction/clock. Since the instructions are very similar (i.e. JMP, ADD, SUB, IMUL, RET, TRAP, etc), one can thus make this comparision in architectures.

As for specific programs, that *is* related to architecture, but not in the broader sense I am arguing. You are going into specifics like say the FPU within each chip and their data-routing methods. Although is part of the architecture, I am talking about the big picture. Even though these specifics may exist program wise....the PIII/Athlon architecture is far superior to that of the P4. This is an accepted fact within the Computer Architecture field. Given, the P4 can execute at a higher frequency and thus the Athlon and it are similar in performance (look at equation), but the more efficient and superior architecture is that of the PIII/Athlon since it executes things more efficiently. Now of course, one could argue that the P4 has some superiority since it makes up for it's inefficiencies by being able to run faster, but...well...that's another argument.

Basically, what I am getting at, as said above, is that you are comparing specific programs for users and I am talking about the big pictures...from a Computer Architects standpoint. As an Engineer, I don't think I could ever say a processor with 20+ pipeline stages and where multiple ones only transfer data from one stage to the other is superior to something that has a cleaner design and doesn't waste stages like that just for the sake of frequency. So, what I'm getting at with "superior" is from a design perspective. Again...this is a fairly well-accepted fact within the computer industry.

At the same time, one will probably wonder why is Intel clobbering us? That answer is easy....MOST consumers equate bigger numbers with better performance. This is due to Intel's marketing department and even I will say that they do a GREAT job in convincing the consumer that Ghz = performance. I mean most educated consumers see the benchmarks and can see a 2.25 Ghz (3200+) Barton is comprable to that of a 3.06 Ghz P4. Why? Because of the equation above...

Then again I could go into benchmarks and how, unfortunately, the consumer can't even trust those, but I won't since most of you will be bored to death and I know a number of you are probably just thinking "oh he works at AMD and that's why he thinks this."

Not that you'll believe me, but as an Engineer I *chose* to work at AMD because of their designs and whatnot (yes I had offers from both)...although their work-life is what really got me . I'd like to think my engineering brain is beyond such things as "company loyalty" and that I can see a "bad" design and whatnot. Hell, even in my advanced architecture classes we went into such things, so again...for the most part the Computer Architecture field feels this way although not as laughable as say the Itanium (We've sold more Opterons in one deal (Cray Super Computers ~10,000) than Intel has Itaniums in 3 years haha).

Anyway...sorry about that last night...I find it funny...

Later!