How Crap is the Pentium 4 !!!!!!

[Bojo]

I just cannot believe how bad this new processor is! I though the pentium 3 was pretty lame but Intel has outdone themselves this time. Am I the only one astounded by how totally ridiculous this chip is?

I remember the days when you bought a cpu and you new that at the worst it would be about 50% better than it's predecessors. Gee it must have been a few years ago.

Not only is it far slower than the athlon and the pentium 3 in most benchmarks but their screwing around with two different versions!!

And yeah sure it'll get faster in 6-12 months. Great!! fantastic then we'll be able to reap the rewards of a few percentage points of greater performance as the gigahertz go through the roof. And what about the size!!! what a piece of crap!!!! Thats it, my prediction is that AMD is going to destroy Intel over the next 2 years.

If anyone can tell me why Intel shouldn't be run out of business by their totally lame chips please let me know.

 

[sciencewhiz]

competition

 

[Hawker]

Hi!


I can't say I agree with you. The PentiumIII is one generation older than the Athlon. Still it performs very close to the Athlon. The P4 will probably offer better performance later. I think it is a bit early to judge. Wait a few months before you cast your verdict.


Vidar
Editor
Temple Of Technology
www.templeoftech.com

 

[RoadRuner]

your just like the peeps that called the duron crap when it came out.


Until the market exploits the pentium 4 (ie whistler, new versions of linux etc) you will see poor performance.

same thing happened with pentium-pro when it came out.

I've still got a p-pro running today crunching rc5 and routing my cable modem.

Wait for compiler optimizations, then you can speak the truth about the processor.

 

[BurntKooshie]

<< I remember the days when you bought a cpu and you new that at the worst it would be about 50% better than it's predecessors. Gee it must have been a few years ago. >>



I think these are some of the things you are missing:


*) The Athlon. It was some good 50% faster in some situations. Not every. On average, it was perhaps 20% faster than the P3 Katmai. Then the coppermine came around. Performance was pretty much on an even keel, except for things like mp3 encoding, and 3dstudiomax. With Q3, the P3 STILL TO THIS DAY beats out the Athlon. The athlon could do better, with stronger compilers....

*) The Pentium 3. Was it 50% faster than a P2? Even if you were to compare a hypothetical 450mhz P3 coppermine to a 450mhz P2, the difference would be a lot less than 50%. Infact, the P2 450 and P3 450 had very, very little performance difference, and with good reason: there wasn't much changed with it! To get the very most out of it, the P3 needs to have SSE optimizations, which means recompiling. Wow, similar to older days....

*) The Pentium 2. Was it 50% faster than the Ppro? No, infact, the Ppro was slower in some limited situations. The extra L1 cache helped out, but the L2 cache being half the clock speed, and with higher latencies didn't help out performance. Oh, but you want to compare it to the P55 (pentium mmmx) right? Ok. Yes, it was faster. Intel introduced the P2 266 BEFORE the 233. Why? To make sure there was a big enough performance delta so that people would buy it. The P2 233, except in some gaming situations, and really FPU intensive code, was NOT that much faster than a P233mmx. Many reviews (read: magazines) said if you wanted to buy &quot;one behind the bleeding edge&quot; to just get the P233mmx, and save a wadd of cash compared to the P2 233.

*) The Ppro. This was actually significantly SLOWER in 16 bit code than the P2 200. Is that a + 50%? No....But yeah, I know, anyone using a Ppro didn't spend all that cash to use windows 3.1! They used windows NT, and there, it was king. But not in all places. And it was horribly expensive! It was NOT an economical decision when buying a comp with a Ppro when they first came out. A lot of code had to be recompiled with a &quot;free&quot; FXCH in mind to get the best performance. Recompiling, and subsequent coding, was done to take specific advantage of the &quot;new features&quot; of the Ppro. It wasn't something that magically just happened. But the P6 core, which includes the Ppro, P2, P3, and Cumines (and all celerons) has been so heavily optimized for at this point, that nearly everything you buy is already pre-optimized for the P6 core (it could be better, but its a lot better than any K6-x optimizations you'll ever see).

*) The Pentium. A lot of code had to be recompiled. The original Pentium 60 was slower than a 486DX4 100mhz. Where's your magical 50%? Because the Pentium had a dual issue super-scalar design, code had to...you guessed it, be RECOMPILED to get the most performance. If you don't understand why that's the case, let me know, and I'll write a short diatribe on it for you...or maybe someone else will. The point being, it took awhile before the fact that the Pentium could ramp in mhz better than the 486 came to light, and the Pentium prooved to be the faster solution.

Whew. Do you really want me to go on? No, the P4 isn't smoking the P3. Sure, its scores are all over the place, where in some, it absolutely rules, and in others, it chokes badly. Other intel chips have done the same exact thing, and some you've probably never heard of, because it was 10 years ago, and they flopped SO BAD that intel doesn't talk about it anymore.

The P4's performance varies more than many expected. But here's a dirty little secret: the scores from a 1.5ghz P4 are, on average, about the same as a P3 1ghz. 1.5 ghz is EASY for intel to get for the P4. Is > 1ghz easy for the P3? Nope. 1ghz STILL ISN'T IN VOLUME. The P4's performance will only go up, even at the same clock speed, because of better compilers, and also because it ramps higher. THIS IS ONLY THE BEGINNING for the P4. The P3, on the same process technology, will not go NEARLY as high in mhz as the P4. The P4 will be able to have a performance advantage (on average, some it will win, some it will lost, but it will win more than it loses) on EVERY process technology.

Care to take a similar, more indepth perspective? Read this, which is hightly technical, and above my head in some places

Intel is too huge to be taken down over something like this. If AMD does well, and Intel doesn't execute well with the P4, then yes, Intel could be hurt. But they are far from out of the market because of the P4. The P4 architecture has such an incredibly high lowest common denominator, its scarry. 3.2gbytes/sec of bandwidth is something that only the high end Alphas have right now. The P4 has it too.... The P4's performance will only go up from here, and relative to the P3, even with the P3 scaling higher, the P4 will continue to show a performance delta.

BK.

[EDIT]spelling

 

[BurntKooshie]

BTW - if you think I'm an Intel zealot, know that I've never owned an Intel chip in my life. Know that I'm not an AMD zealot. Know that I'm not a Cyrix zea....wait a minute, for those of you who know me...dang, I am a Cyrix zealot

 

[Jubjub]

BurntKooshie, way to sling the facts! Man, you get paid by the word, right?
Seriously, well said. I was curious though, when you said, &quot;But here's a dirty little secret: the scores from a 1.5ghz P4 are, on average, about the same as a P3 1ghz.&quot;
To what scores are you referring? I hadn't read anything to this effect yet, and I'd be grateful if you could point me in the direction of that reference.

Tx,
Jubjub

 

[BurntKooshie]

P4 is faster than P3 in :http://www.anandtech.com/showdoc.html?i=1360&amp;p=16

http://www.anandtech.com/showdoc.html?i=1360&amp;p=17

http://www.anandtech.com/showdoc.html?i=1360&amp;p=20

http://www.anandtech.com/showdoc.html?i=1360&amp;p=21(split between the two benchmarks).

http://www.anandtech.com/showdoc.html?i=1360&amp;p=22


P4 is slower than P3 in:http://www.anandtech.com/showdoc.html?i=1360&amp;p=18

http://www.anandtech.com/showdoc.html?i=1360&amp;p=19

http://www.anandtech.com/showdoc.html?i=1360&amp;p=21

I'll find more later from different sites, but you get the idea (I don't wanna be late for classes )

 

[Jubjub]

Thanks BK! Compelling stuff...

'Course, to the average consumer, all that matters is that big fat MHZ (or GHZ) number, and the number after the &quot;P.&quot; A P4 1.5 Ghz box would outsell a P3 1 Ghz box in retail channels even if the P4 was half as fast in real-world performance. Oh well, so it goes.

 

[BurntKooshie]

<< 'Course, to the average consumer, all that matters is that big fat MHZ (or GHZ) number >>

That's exactly what VIA is betting on with their CPU's.

 

[Bojo]

BurntKooshie - Some of your points I agree with but others I find totally bizaare. I think you are bending the truth quite a lot in some of your arguments.

Why are you comparing the Athlon to the P3 for a 50% performance improvement? Their direct competitors in MHZ and performance updates! The Athlon was not meant to be a generataion ahead of the P3.

Your point about the P3 not being much faster than the P2 I completly agree with, it's also a complete joke (not so much as the P4). by &quot;a few years ago&quot; I meant 3-5 years.

Now you compare the Ppro to the P2! You're completely bending the truth to support your argument. Yes indeed in 32-bit apps the Ppro is about the same as the P2, I completely agree but again you are not comparing the correct CPU generations and market segments. Ok yes you mention that this is the wrong comparison to make but I want to reiterate that point.

Yes I agree that the MMX technology had to be adopted by software to get much performnce updates but I think here I must simply question your facts here. It is not unfare to say that in general the pentium 2 is 50% faster than the pentium. I think most people would agree here.

But here it gets weird. You talk about comparing the 486-DX4 to the Pentium 60? What tests are you talking about??? for starters lets get realistic with comparisons. compare the same MHZ pentium and 486 processors. the 486-DX4 100 faster than a Pentium 100 ?? Dont make me laugh. Even when the pentium was released in most cases it was at least 50% faster than the 486. You are obviosly talking about select cases not in general.

But to get back to the point I find it rediculous that you are defending the P4. We've seen that the P4 is significantly slower in many tests than it's former CPU the P3. In many tests it's about the same and in some its quite a lot faster. Forget about software optimization, the fact is that up until the P3, Intel celerons were in general a fair deal faster than the ones before them, and now the P4 isn't even close. Thats a fact, get real about it.

In part this is not just Intel's fault, lets face it, It's the laws of Physics that are making it harder and harder to keep up Moore's law. But that doesn't change the fact that the P4 is a joke.

If you really aren't an Intel fan then this is pretty sad that you can't face facts. Yes you have some very valid points but really their just not enough to turn the P4 into anyhting like a decent CPU. Its big, yes it's smart, but it just ain't fast.

I'm very interested to hear what you have to say for yourself BurntKooshie, you are obviously very well informed but I just don't think your basic arguments are very well thought out.

On a different note, why on earth are you a Cyrix fan?


-----------------

 

[StickHead]

When the average Joe goes out to buy a computer, the thing he knows to look for is a big number in front of the letters Mhz. And he thinks it's fast. Americans have a facination with big things)) it's like I want a car with a V8, because it's fast.

 

[Zedfu]

right there. i've seen pc's on the market with high mhz cpu for extremely cheap prices, but the average consumers are too dumb to figure out that &quot;you pay for what you pay&quot; without realizing all other components in it are barebone.

 

[BurntKooshie]

<< Why are you comparing the Athlon to the P3 for a 50% performance improvement? Their direct competitors in MHZ and performance updates! The Athlon was not meant to be a generataion ahead of the P3. >>

Because AMD calls it their &quot;7th&quot; generation CPU. The P3 is still a super beefed up &quot;6th&quot; generation



<< I completely agree but again you are not comparing the correct CPU generations and market segments. Ok yes you mention that this is the wrong comparison to make but I want to reiterate that point. >>

After thinking about it, you're right. Different market segments, different prices, different performance levels.



<< Yes I agree that the MMX technology had to be adopted by software to get much performnce updates but I think here I must simply question your facts here. It is not unfare to say that in general the pentium 2 is 50% faster than the pentium. I think most people would agree here. >>

I would certainly agree....what's your point? You're comparing the first iteration of a 5th generation CPU v. a 2nd iteration of a 6th generation CPU. See my point?



<< But here it gets weird. You talk about comparing the 486-DX4 to the Pentium 60? What tests are you talking about??? for starters lets get realistic with comparisons. compare the same MHZ pentium and 486 processors. the 486-DX4 100 faster than a Pentium 100 ?? Dont make me laugh. Even when the pentium was released in most cases it was at least 50% faster than the 486. You are obviosly talking about select cases not in general. >>

I make this point because the two chips were out at about the same time. There were still plenty of 486dx4 100's being sold when the Pentium was introduced. In this case, the roles were reversed, where the older generation held a clock advantage (due to heat, etc, not due to architectural design) over the newer one. The 486 never scaled much past about 160mhz (AMD's 586, really nothing more than a higher clocked 486 with more L1 cache, would overclock to that quite nicely), whilst the Pentium made it to 200mhz. Of course the process technologies were different, but even still, it was the case that, with unoptomized compilers for the Pentium, the 486 at a higher clock rate could win. The P4 could be a like a 486, but DESIGNED to ramp better than the previous generation, even though it had lower average IPC (not like a 486 in reality, just analogously), which is what will make up the difference in performance. In x86 code, there are so many branches, so many stalls, etc, that its getting really, really tough to get more IPC. But a higher clock speed is still possible, quite obviously.

Here's where the T-bird athlon comes in. On paper, it should brutally destroy the PIII coppermine. But it doesn't. See aceshardware for a really, really good explanation. So the P3 should be destroyed. But its not. So really, the one that has a higher clock speed wins, right? Right. And that is where AMD has been winning, because they have a 1.2ghz chip, while Intel still doesn't have volume 1ghz. AMD wins.



<< But to get back to the point I find it rediculous that you are defending the P4. We've seen that the P4 is significantly slower in many tests than it's former CPU the P3. In many tests it's about the same and in some its quite a lot faster. Forget about software optimization, the fact is that up until the P3, Intel celerons were in general a fair deal faster than the ones before them, and now the P4 isn't even close. Thats a fact, get real about it. >>

Slower at the same clock speed, I totally agree. Please, please read the article I quoted. I'm not trying to say that I'm right, you're wrong, get over with it, but by you reading it, you'll see where I'm coming from. Basically, the P4 can only get better with better compilers, and more optimization.

I'm going to go out on a limb, and say that the P4 1.5ghz is just as fast as a 1ghz P3 (on average, some it wins, some it loses). So its a lot slower per clock. But 1.5ghz is easy for the P4. 1ghz is really tough. So who will be the winner be on the same process? Eventually, the P4 will win out in absolute performance. That is what its all about, right? Take a look. The P4 turns out SPEC2k scores that are among the best in the world. The EV67 833 could top it, but oops, its not out (politics suck in chip manufacturing....poor API). Oops, the 833 isn't out yet. So who cares if it takes intel 1.5ghz to do what the Alpha can do in 833? The fact is, there is a consumer chip, namely the P4, which can do very well right now.

I don't like that fact, because I dislike Intel as a company moralistically (not technologically). I like AMD, because they are the underdog. I loved cyrix because they were the ultimate underdog with a chance (the Jalepeno core was markedly similar to the Athlon's....but it was never finished ).

I hardly think the P4's a joke. It is for now, wait another year. Then tell me.

BK.

[EDIT]Oh yeah. Forgot to add this: I'm a Cyrix &quot;fan&quot; because I'm the only one on earth to be a fan. Back in the early 686 days, they were the fastest x86 on the consumer market for integer code. Then people found quake....It was still good technology until about PR 266, after which point, they were no longer very competative.

 

[bonkers325]

waste of money and crap even though i love intel i must say the p4 is suck

 

Intel has too much money and resources to be destroyed.
Yea AMD is great and so is Intel.
Its good to see the leading company get bested, keeps them on their toes.
Oh gee look were getting beat, well lets remove a few billion from our advertising campaign and give it to R&amp;D.

 

[frustrated2]

I think we are way to premature to judge the p4. Once software gets optimized for it it is going to be a monster. The memory bandwidth is incredible and allows this processor to have unlimited potential. At Tom's hardware he already overclocked the processor to 1.7 ghz which means that it scales well and that there will be a 2 ghz processor on the market before we know it. The price is incredibly high right now but don't forget when the new 1ghz athlons were released they were $1000 as well. Platforms and rdram will all come down soon enough. I am sure that we are in for a treat in the next year observing the processor race between intel and amd

 

[Zedfu]

and you can say that for AMD also. their athlon is only around 2-3 years old. and you can also say that they're still at their premature level. other revisions can be done to this core to further upgrade its potential, just like what Intel did for the P6 core. they went from ppro, p2, p3, to cumine, with each step improving its performance. the tbird has not even nearly reached its structural limit yet, so i can expect further refinements coming from this core in the future.

 

[Zedfu]

also, AMD already has a lineup for future cpus to be released already. they've been bitten once before, but won't be bitten again!

 

[FastD]

Just want to say that.. You are arguing that the p4 will be faster in the future.. yes this is true. BUT.. Won't Amd's processors also be faster in the future?.. Ofcourse they will be.

Someone said that the average joe only looks to the megaherz. I don't agree. There is one more thing he asks for and that is PRICE! Here in sweden a 64mb RIMM costs about $400.. (a Mosel Vitalec 128MB cas 2 dimm is $120) and you have to get two of them. That's $800 just for the memory. That's pretty close to what many here make in a month after taxes.

And if you are looking at buying a high end system then you will want 256MB or more.
That means at least $1600.. Just for the memory. That is a little less than you will pay for a complete Ahtlon 1200 sdr system which is AVAILABLE TODAY and that is a system which performance will be on par with a p4 box(that costs almost twice as much) in all applications except quake 3 in 640*480 resolution.

By the way.. who plays Quake 3 at 640*480?.. Do you?.. I don't.. And I refuse to believe that noone except those with very old, low end systems play it at this resolution. But then again.. We are disussing high end, not lowend now.

My conclusions are these:

1. The p4 will NOT sell well on it's high megaherz alone because it is simply too expensive coupled wiuth RDRAM.

2. It is pointless to compare quake 3 scores at the lowest resolution since most ppl play it at a much higher resolution anyway.

3. Since most people with high end rigs play games at 1024*768 or even higher and we all know that at resolutions higher than 800*600 the graphics card is the bottle neck and the only benchmark we know of were the p4 does well is q3 at low resolution there would be absolutely no reason to get a p4 today or anytime this year and most likely not until sometime late next year when software has been optimized for the p4.
And that is provided that that will happen AND if AMD won't release faster cpu's!!!

Bottom line. The p4 is bad value now and tomorrow, low performing and won't sell.

Cheers from Sweden!

 

[gplracer]

I think the bottom line here is that the p4 simply is not a good upgrade at this point. I am sure it will get faster with time, but the amd chips will too. Only time will tell who is going to come out on top. Intel seems to be banking on the future at the expense of the moment. I just built a duron system and I would do the same again. It STILL is the best value for the money. When it is time for me to build another, I will go intel or AMD. I do not care which. Keep in mind that most of us here are pretty knowledgeable about computers. The average consumer will see 1.4 gigs and go wild. They will buy because it has the highest rated speed. Why would you want a thunderbird i gig when intel has the 1.4 gig and the name recognition. I can hear the sales people now. Intel does not care about us. We are too small a segment of the market. They only care about $$$$ the same as AMD. It is just the state of our society now that claims like this are being made. It is the same as the store that offer 50% off. That is do do. 50% off of what? Intel will not get my dollar now but they know what they are doing.

 

[Bojo]

Hear Hear!!


----------------------

&quot;How do you like me NOW!!!&quot; -- Will's Friend

 

[SuperTool]

I am going to try and keep my mind open about the P4. I think intel put it out there for early adapters, and I think any consumer buying it now is nuts. They cut a lot of corners to get this thing out, smaller Icache, weak FPU. I think this CPU is too big for the .18u process. I think once they move to a smaller process, they will get both the MHz up, and fit put better FPU (they better have one by then) and more cache on the die. SSE programs will be here eventually, with both Intel and AMD planning to support SSE. Mainly of course, they need to get away from Rambus to be able to compete with Athlon on price.
I am more interested in the IA-64, and how that's going to fan out. If that fails, that's hundreds of manyears down the drain, and years wasted.