Is AMD faster than intel

[kirkyg]

Originally posted by Moridin


I demonstrated that on properly optimized code the P4 performs similar to the Athlon clock for clock. How dose that support your point in any way? As for "microanalyzing" all I have done is take you unsubstantiated claim and refute it. You have yet to provide any evidence to counter that.

No they don't because the games you are pointing to are based on game engines who's development began before P4 optimization guidelines were available, so they couldn't possibly be taken into account. This is why application benchmarks CANNOT provide evidence for your claim. By definition an application benchmark cannot give you any information beyond that specific application using that specific data. (If you have questions reguarding what benchmarks can and cannot tell you feel free to ask)

I think the lightspeed results are clear indication that the Athlon does not necessarily "destroy" the P4 clock for clock in processor intensive graphics applications.

My point is that its never going to be optomized to its full potential and it ALWAYS turns out to be marketing for the chip maker. I dont buy that anymore. I want to know if its going to give me a more solid framerate in warcraft3, counterstrike, bf 1942 etc.

kirkyg

[Chronus]

Originally posted by Moridin


Do you have some links to back that up?

You should have read carefully what I posted.

I said "hear", there're no links that I can remember.

However, last November, INTEL's director of operations (IIRC) answered to a question asked by a CNBC anchor (about the slow recovery of the sector and the competition INTEL was experiencing from AMD) in this way:
"There's really no competition in the CPU market. We dominate this market and set its trends. Without a doubt, the gap between our technology - in terms of speed and eficiency - and our competitors's will continue to build up."

This was the second time I hear INTEL representatives speak in this way. And on the same network (MSNBC/CNBC).

Besides, take a look at this

"A desktop system based on the Intel® Pentium® 4 processor at 2.80 GHz delivers six times the productivity performance gain of commonly installed systems that use the Intel® Pentium® III processor at 500 MHz, as measured by Productivity Performance SYSmark 2002*. "

This is from http://www.intel.com/ebusiness/upgrade/index.htm.

6*500 is 3000, ie, six times the performance of the Pentium III at 500 MHz should be equivalent to a Pentium at 3 000 MHz. The fact that they managed to have that kind of performance at 2.8 GHz is of great significance to them. Is it any wonder that most analysts say that INTEL still speaks and thinks in terms of "MHz wins all"? If they present such a reasoning ("2.80 GHz delivers six times the [...] performance gain of [...] the Intel® Pentium® III processor at 500 MHz"), then they must think that way.

Plus, read this

" Surpassing the 3 GHz mark, the Intel® Pentium® 4 processor at 3.06 GHz offers higher levels of performance, creativity and productivity.
Based on Intel® NetBurst™ microarchitecture, the Pentium 4 processor offers higher-performance processing than ever before. Built with Intel's 0.13-micron technology, the Pentium 4 processor delivers significant performance gains for use in home computing, business solutions and all your processing needs.",

which can be found at
http://www.intel.com/home/desktop/pentium4/index.htm

Plus, read this very smart piece, this very smart way of going after AMD:
"I was using a 2100+ Athlon XP*. I thought it was giving me good game performance and that I was playing to the best of my ability. But then I received the Pentium 4 processor from Intel. Playing UT2003* with that processor was a completely different definition of smooth. Imagine no choppiness whatsoever. Imagine playing these games at the speeds they were meant to be played. Intel has made this possible. ",

and

"Several friends have asked me which kind of computer they should buy: AMD processor-based or Pentium 4 processor-based. I've told them I own both, but that my main computer is now the Pentium 4 processor machine, and it looks to be that way for a long time to come. "

which is available at http://www.intel.com/home/maximize/s...anbarchian.htm

Now showing off a guy comparing a XP2100+ to a 3GHz processor! is laughable.



You should really think twice before you post anything like this. You dislike people who post things with nothing to back them up. So do I. Whatever I post, I try to make it based on facts or links or articles or TV appearances.

Chronus

[steppy76]

And that has what to do with the price of tea in China? The lack of "model" names means that the only thing really differenciating p4's IS their clockspeed. Having them not say it(to demonstrate which P4 processor they're talking about)would be like AMD always just using Athlon XP without the model numbers. The rest of it is marketing speak, and both companies engage in it equally. I don't see it showing Intel with a Mhz is all that matters mantra.

As far as the comments by Intel's director of operations, he's RIGHT. Intel DOES dominate the market and set it trends...the BEST competitor they have has 1/4th of their marketshare(this would be considered "not having much competition" in a lot of businesses books). Whether the "Without a doubt, the gap between our technology - in terms of speed and eficiency - and our competitors's will continue to build up." ends up being true has yet to be determined, but could be very prophetic if A64 flops. As for the p4 2.8 is 6 times faster than p3 500...look at the bench, the p4 has a much faster FSB...faster memory...SSE2(which I'm sure is probably in that bench)...on-die L2 cache...it very well may not need 6 times the clockspeed to get six times the performance when you take into account all the factors.


To kirkywhatever...you're right that the Athlon does outperform the P4 clock for clock and this point in time on the majority of things. Nobody is saying differently. What they are saying is that it is NOT a universal thing, and could change in time. A clock for clock comparison is really irrelevant though as Intel's top end is clocked several hundred Mhz higher than AMD's at any given point in time(this is why it is a worthless argument Colossus). If Intel DID stand pat for the next year or two to let AMD catch up in clockspeed it would become a valid base of comparison...but this is the real world and Intel is NOT sitting still.

[Moridin]

Originally posted by kirkyg


My point is that its never going to be optomized to its full potential

That's what they said about the P5 and P6. It wasn't true then, what do you think has changed?

[Moridin]

Originally posted by Chronus



However, last November, INTEL's director of operations (IIRC) answered to a question asked by a CNBC anchor (about the slow recovery of the sector and the competition INTEL was experiencing from AMD) in this way:
"There's really no competition in the CPU market. We dominate this market and set its trends. Without a doubt, the gap between our technology - in terms of speed and eficiency - and our competitors's will continue to build up."

Speed != clock rate.

Originally posted by Chronus


"A desktop system based on the Intel® Pentium® 4 processor at 2.80 GHz delivers six times the productivity performance gain of commonly installed systems that use the Intel® Pentium® III processor at 500 MHz, as measured by Productivity Performance SYSmark 2002*. "

What part of that do you disagree with? Are you saying the 2.8 GHz P4 does not run at 2.8 GHz or that it does not offer 6 times the performance of a 500 MHz PIII?

Originally posted by Chronus

6*500 is 3000, ie, six times the performance of the Pentium III at 500 MHz should be equivalent to a Pentium at 3 000 MHz. The fact that they managed to have that kind of performance at 2.8 GHz is of great significance to them. Is it any wonder that most analysts say that INTEL still speaks and thinks in terms of "MHz wins all"? If they present such a reasoning ("2.80 GHz delivers six times the [...] performance gain of [...] the Intel® Pentium® III processor at 500 MHz"), then they must think that way.

Not one in that quote did Intel say "MHz wins all". What they said was a currently shipping chip outperforms a previous chip by 6X, nothing else. You read the nonexistent clock rate is all important stuff into it all by yourself.

BTW, it is easily demonstrable that clock rate is far and a away the most important factor in performance when you look at the big picture as this quote does. So, in this case clock rate would indeed be highly important, and deserves to be mentioned.

Originally posted by Chronus

" Surpassing the 3 GHz mark, the Intel® Pentium® 4 processor at 3.06 GHz offers higher levels of performance, creativity and productivity.
Based on Intel® NetBurst™ microarchitecture, the Pentium 4 processor offers higher-performance processing than ever before. Built with Intel's 0.13-micron technology, the Pentium 4 processor delivers significant performance gains for use in home computing, business solutions and all your processing needs.",

Again what part of that do you disagree with. The chip does run at 3.06 GHz, and most people agree it is the best performing chip available.

Originally posted by Chronus


"I was using a 2100+ Athlon XP*. I thought it was giving me good game performance and that I was playing to the best of my ability. But then I received the Pentium 4 processor from Intel. Playing UT2003* with that processor was a completely different definition of smooth. Imagine no choppiness whatsoever. Imagine playing these games at the speeds they were meant to be played. Intel has made this possible. ",

and

"Several friends have asked me which kind of computer they should buy: AMD processor-based or Pentium 4 processor-based. I've told them I own both, but that my main computer is now the Pentium 4 processor machine, and it looks to be that way for a long time to come. "

I personally do not like anecdotal evidence, but testimonial is a well established sales technique that many companies use. This certainly does not support your claim that Intel pushes clock rate as the only thing that matters.

Originally posted by Chronus

You should really think twice before you post anything like this. You dislike people who post things with nothing to back them up. So do I. Whatever I post, I try to make it based on facts or links or articles or TV appearances.

Chronus

I'm trying to get you to think a little before you post things that are unsubstantiated and largely false. Historically, designing for clock rate is the best way to achieve performance, and if you look at the big picture clock rate is in fact the best proxy for performance there is. It is only when you start to make small comparisons across different architectures that clock rate as a proxy for performance breaks down.

This is why most processor companies put clock rate front and center in their processor designations. It is simply to important to ignore, but I have NEVER seen or heard an Intel rep claim it is the only thing that matters, despite your claims that they do it all the time. In most cases therefor I must believe it is your misinterpretation of statements made by Intel that are at the root of this claim.

[Chronus]

Originally posted by Moridin
...
I'm trying to get you to think a little before you post things that are unsubstantiated and largely false. ...

Largely false?

Your arrogance shines through all the showing off. Plus, rudeness incorporates pretty well into your character and behaviour.

Enough talk.

Chronus

[kirkyg]

Originally posted by Moridin


That's what they said about the P5 and P6. It wasn't true then, what do you think has changed?

Your going to sit here and tell me the pentium benefited from mmx and mmx2? LMAO...the instructions were never used in anything beneficial.

through with thread. Waste your money how you see fit.

kirkyg

[steppy76]

Kinda of the "I'm taking my ball and going home" debate strategy you two? Once your arguments get debunked you leave?

BTW, thanks for the backup Moridin.

[steppy76]

Originally posted by kirkyg


Your going to sit here and tell me the pentium benefited from mmx and mmx2? LMAO...the instructions were never used in anything beneficial.

through with thread. Waste your money how you see fit.

kirkyg

MMX was used quite a bit, and may still be today(I'm not sure if SSE totally superceded MMX or not). The reason you don't see MMX enabled or something is because every target processor HAS MMX. Anyway, he not even talking about MMX(which came out a few years after the Pentiurm debuted), he's talking about how both the Pentium and P6 needed code tweaked for their basic architectures rather than old legacy 486/386/286 code.

[Moridin]

Originally posted by Chronus


Largely false?

Your arrogance shines through all the showing off. Plus, rudeness incorporates pretty well into your character and behaviour.

Enough talk.

Chronus

I don't know how much more polite I can be. You made a claim I disagree with, and I gave you every opportunity to support your claim with evidence and so far, you have failed to do so. under these circumstances, how can we accept that claim as true?

BTW, such attacks such as labeling people rude and arrogant simply because they disagree with you are very bad form. If you disagree with someone, attack the ideas not the person.

[Moridin]

Originally posted by kirkyg


Your going to sit here and tell me the pentium benefited from mmx and mmx2? LMAO...the instructions were never used in anything beneficial.

Who said anything about MMX? The first versions of the P5 and P6 didn't even support MMX. Using new instructions are only a small part of optimizing for a new processor. It doesn’t even make sense to bother with these new instructions until you have dealt with things like scheduling.

What I was referring to the fact that early P5s were frequently outperformed by highly clocked 486s, and early P6s were frequently outperformed by contemporary P5s running at the same clock rate.

On newer software this advantage is completely reversed, but early in their life these processors suffered the same criticism you are laying against the P4.

You are correct in one thing though, mmx was relatively uncommon early on. This doesn’t weaken my claim at all though, since the P5 and P6 eventually managed large performance increases over their predecessors even without it.

Originally posted by steppy76
I'm not sure if SSE totally superceded MMX or not

SSE did not. MMX handles integer operations while SSE handled floating point. IIRC, SSE2 does both, and therefore does supercede MMX, though the capabilities are similar. SSE2 is roughly a combination of MMX + 64 bit Integer, and SSE with 64-bit floating point. The throughput for 32-bit integer and 32-bit floating point hasn't changed however, so if you are working with 32-bit operations SSE2 doesn’t offer a significant advantage over SSE + MMX.

Originally posted by kirkyg

Waste your money how you see fit.

Where have I ever talked about money in this thread? I have been discussing the technical merits of these processors, nothing more nothing less.

[steppy76]

Originally posted by Moridin
1. On newer software this advantage is completely reversed, but early in their life these processors suffered the same criticism you are laying against the P4.

2. SSE did not. MMX handles integer operations while SSE handled floating point. IIRC, SSE2 does both, and therefore does supercede MMX, though the capabilities are similar. SSE2 is roughly a combination of MMX + 64 bit Integer, and SSE with 64-bit floating point. The throughput for 32-bit integer and 32-bit floating point hasn't changed however, so if you are working with 32-bit operations SSE2 doesn’t offer a significant advantage over SSE + MMX.

1. As did the PPro and PII vs. the Pentium, and just about every other Intel processor at release. It amazes me that we have the same arguments about the exact same thing every 5 years or so.

2. Thanks for the clarification...I knew one of them (sse or sse2)superceded MMX.

[Moridin]

Originally posted by steppy76


1. As did the PPro and PII vs. the Pentium, and just about every other Intel processor at release. It amazes me that we have the same arguments about the exact same thing every 5 years or so.

I don't recall the 486 having that difficulty. It did take a long time before on die FPU was used much though. (In fact the FPU wasn't used much until long after the P5 was established, and just before things began to shift on mass to the P6.)

The 386 was such a big jump over the 286 that it took nearly 15 years before the new capabilities (386 enhanced mode, which included 32-bit operation, and memory protection) were used much, but IIRC it did still offer a significant performance advantage over the 286.

The criticism of the 386 was more along the lines of "why would anyone need that much power in a PC? This chip is really only good for servers." That seems to be what IBM though when they waited two years before putting it into a PC.

That is of course when Compaq stepped in. They beat IBM to the punch with the 386, and we never again cared if a PC was 100% IBM compatible, only if it was MS compatible. IBM never regained control of the PC.

[steppy76]

Well, when the 486/dx2's and dx4's(still don't know why they weren't dx/3's) the complaints were that only the processor speed doubled or tripled but the rest of the system didn't. Little did people know just how much that little thing called the clock multiplier would change the processor world. Things like that,(ok not the EXACT same arguments but you get my drift.)

[azns_kickass]

Originally posted by Chronus


Largely false?

Your arrogance shines through all the showing off. Plus, rudeness incorporates pretty well into your character and behaviour.

Enough talk.

Chronus

What a (not so) grand way to admit defeat to an argument. All your 'facts' are unsubstantiated, and you know it. When someone proves you wrong, you move on to personal insults.

Shameful.