P4 2.4Ghz - or - Tualatin 1.4Ghz OC to 1.575Ghz?

[Fun Guy]

I am looking to build the best machine possible for my main box. I was going to go for a P4 at the highest rating, but when I saw the below article I realized that having a PIII-S 1.4Ghz at 1.575 (easy overclock to 150Mhz bus) would seriously challenge the P4 for performance (maybe even beat it), I thought again.

Tualatin vs. P4 vs. Athlon

Any opinions?

Oh yeah, one more thing, I am wondering when the new CPUs and chipsets are coming out to support the 533Mhz bus for the P4. Anyone know? And will the RAM bus be overclockable do you think?

 

[ahsia]

While the Tualatin seems impressive, it is not worth the money IMO. The CPU alone cost $300+, and while the performance is comparable to a P4 1.8GHz, it is comparable to only an Athlon 1.4GHz. For about $120 less, you can get the XP 2000+ or a very overclockable P4 1.8A that will smoke the Tualatin, even if the Tualatin is overclocked to 1.575GHz. Tom's review doesn't tell you whether it is using the 1.8 Williamette or the Northwood, but I'll be it isn't the Northwood.

Stick the either the 1.8A or XP 2000+, the motherboards are better for those processors, and the Tualatin is way overpriced.

As for the second part of your question, P4 with 533MHz bus should be out soon. SiS645DX has official support for 533MHz, and memory is easily overclockable. Even with my SiS645 Gigabyte 8SRX, I am running my DDR333 PC2700 at DDR362=181MHz, and some people with Corsair or Kingmax have gone as high as DDR380-390=190-195MHz.

Good luck!

 

[Athlon4all]

The Tualatin is a decent CPU, but it really, is a dead end upgrade wise and can't compete with a 2GHz NW, much less a 2.4 NW.

 

[Fun Guy]

I think you guys are missing the significance behind the benchmarks.

Look at the graphs on all pages. When the 1.26 Tualatin is operating on a 133Mhz bus, as tested, it beats the P4 1.8 handily in a few areas. Overclock it to a 150Mhz bus (CPU at 1.425) and not only will the raw CPU speed increase its performance relative to the P4, but the increased bus speed will give you an additional significant boost.

Stick a 1.4 in there and OC it to 1.575, and I think the competition for the P4 is stiff. The P4 might even lose, and by a big margin! Why do you think there is not a clock-for-clock Tualatin out there to compare directly against the P4? Many people suspect it is because Intel knows the P4 would lose. The P4 is simply a much less efficient processor than the PIII, especially a PIII operating with 512K of cache on a small die (the Tualatin).

Regarding upgrades - I can count on one hand (actually one finger) the times I upgraded the CPU without upgrading the motherboard too, so upgradability is a non-issue for me.

Oh yeah, and regarding the 'high price' of the Tualatin 1.4 (~$300), just look at the price of the 2.4 P4 (~$600). Which one is overpriced?

 

[BD231]

I'm running a Tulatin at 1.5ghz(158mhz fsb), it dose 1600mhz, but I dont know where to find ram that supports such a high FSB clock. Once Intel dose the price drops, and gets rid of the 1.6a thru 1.8a's, the Tualatin will become much cheaper. Once the price gets to about 150 or lower, I'll buy a 1.4 and oc it, and hopefully a few months after that, I can pick up a nice new Hammer set-up

 

[ahsia]

<< I think you guys are missing the significance behind the benchmarks.

Look at the graphs on all pages. When the 1.26 Tualatin is operating on a 133Mhz bus, as tested, it beats the P4 1.8 handily in a few areas. Overclock it to a 150Mhz bus (CPU at 1.425) and not only will the raw CPU speed increase its performance relative to the P4, but the increased bus speed will give you an additional significant boost.

Stick a 1.4 in there and OC it to 1.575, and I think the competition for the P4 is stiff. The P4 might even lose, and by a big margin! Why do you think there is not a clock-for-clock Tualatin out there to compare directly against the P4? Many people suspect it is because Intel knows the P4 would lose. The P4 is simply a much less efficient processor than the PIII, especially a PIII operating with 512K of cache on a small die (the Tualatin).

Regarding upgrades - I can count on one hand (actually one finger) the times I upgraded the CPU without upgrading the motherboard too, so upgradability is a non-issue for me.

Oh yeah, and regarding the 'high price' of the Tualatin 1.4 (~$300), just look at the price of the 2.4 P4 (~$600). Which one is overpriced? >>



Sorry, I accidently clicked reply w/o typing anything.

Here is what I think:

Performance

P4 1.6GHz (OC'ed to 2.2GHz) = XP 2000+ > PIII Tualatin 1.4 (OC'ed to 1.575GHz)

Price

PIII Tualatin 1.4 (OC'ed to 1.575GHz) > XP 2000+ > P4 1.6GHz (OC'ed to 2.2GHz)

If you take cost into account, than the Tualatin is not a good buy. It cost way too much. If you think it will perform better than the P4 or Athlon, then just get the Tualatin. I just don't think it is worth the money.....

 

[ntrights]

P4A 1.6 around 130 $ 1.8 around 180 $ (pricewatch) @ 2.2 or 2.4 and you got a price/performance winner.

'nuff said.

 

[imgod2u]

<< I think you guys are missing the significance behind the benchmarks.

Look at the graphs on all pages. When the 1.26 Tualatin is operating on a 133Mhz bus, as tested, it beats the P4 1.8 handily in a few areas. Overclock it to a 150Mhz bus (CPU at 1.425) and not only will the raw CPU speed increase its performance relative to the P4, but the increased bus speed will give you an additional significant boost. >>



The 1.8 P4 is one of the very lowest end P4's currently available. It costs significantly less than a comparable P3, so to say that a P3 overclocked to its maximum yield compared to a P4 at the lower end of its yield is an unfair comparison. Overclock a P3 .13 micron die to its maximum yield and then use the same (or equivalent) cooling to overclock a P4 to its maximum yield, then make a comparison. Clockspeed really doesn't play any role in this, it's all about maximum yield of an architecture.



<< Stick a 1.4 in there and OC it to 1.575, and I think the competition for the P4 is stiff. The P4 might even lose, and by a big margin! Why do you think there is not a clock-for-clock Tualatin out there to compare directly against the P4? Many people suspect it is because Intel knows the P4 would lose. The P4 is simply a much less efficient processor than the PIII, especially a PIII operating with 512K of cache on a small die (the Tualatin). >>



There is no clock for clock comparison because most knowledgable people know clock for clock comparison will only prove one thing, one design has a higher IPC. That says nothing about overall speed. It's like comparing MHz among different processors, has absolutely no effect on overall speed.
As for efficiency, I think your mentality is flawed. IPC is not efficiency. Efficiency is when you utilize a large amount of the resources available to you rather than have it sit there idle. Most of the P4's die does not stay idle, in fact, I'd say it's fair to assume that the meager 2 FPU units that it has is working constantly. What resource is the P4 not using that would make it less "efficient" than a P3? It has less execution units and most of its caching is put to good use.



<< regarding upgrades - I can count on one hand (actually one finger) the times I upgraded the CPU without upgrading the motherboard too, so upgradability is a non-issue for me.

Oh yeah, and regarding the 'high price' of the Tualatin 1.4 (~$300), just look at the price of the 2.4 P4 (~$600). Which one is overpriced? >>



Considering how much the 2.4 outperforms the Tualatin 1.26 (it's not 1.4 unless you overclock), I'd say the overpricing is about the same.

 

[EMAN]

Funguy, you also have to realize that you can overclock the p4 to 133fsb too. Then it will crush the tualatin into bits.

 

[Fun Guy]

My mentality is flawed? Hahaha. Thanks for the entertainment.

All you've managed to do with your post is to show how you have no life in addition to no knowledge of what you're talking about.

Move along.

 

[subhuman]

Actually Fun Guy I think imgod2u's post was exactly right on, some great info.. you might want to reread it and learn something.

 

[Fun Guy]

I read it and re-read it and still have yet to find value in it.

This guy is talking about which processor can flex its muscles more and I really could not care less about that. What I care about is how much weight it can push - how much real work it can do. Tom has shown that the Tualatin even running at 1.2 beats the P4 1.8 in many instances.

The P4 is an inferior architecture, proven over and over again. Why do you think there are no direct comparisons available? Because Intel does not want the average Joe to know this. Intel's agenda is to sell P4s, while my agenda is to get the processor that does the most work.

I said it before, I'll say it again, I don't care how pretty it is or how 'new' it is, I'm not buying the newest, I'm buying the best.

Maybe I'll just build two machines and run my own test for the heck of it. A 1.4Ghz PIII-S @ 1.575Ghz/150Mhz bus, and a new P4 @ 2.53Ghz/533Mhz (133Mhz Quad-Pumped). Then it won't be a 'my dad - your dad' argument, it will be a comparison.

 

[Sid03]

sounds like you already had your mind made up. if so, why'd you bother asking? or did you only want to hear "tualatin" answers?

 

[Fun Guy]

That's a good question. I have a Tualatin 1.26 OC to 1.425 and it is zippy. The Crucial RAM I have only allows me to go 3-2-2 at 150Mhz so if I were to build a 1.4 -> 1.575 rig then I would need to get something like the Mushkin 2/2/2@150 RAM.

I was wanting to go for the P4 but cannot find any evidence it will be better. I was welcoming input and expecting some evidence one way or the other - as you can see, I supplied a little data myself to get the ball rolling in the right direction.

Instead it seems what I got was some hardcore P4 fans who appear to be hardcore for no reason other than they like it, or in imgod2u's case, because of.... umm, well, all I was able to glean from his post was a lot of gibberish that doesn't mean anything to me. Maybe I'm looking too hard for real information or evidence or data - shame on me.

Perhaps my mistake was in asking for opinions instead of information. I'm sure that was my mistake. In any case, I think I will solve the problem myself and build the boxes I outlined in my last post and test them myself. Could be really interesting, actually.

 

[DT4K]

Fun Guy, you ask a question, get some answers, then procede to tell everyone they are idiots.
Since you are so much smarter than the rest of us and already have all the answers, why did you ask?

 

[Wolfsraider]

please supply benchmarks for us uninformed i'd truly like to see the results as i can't afford the option of building both thanks

 

[DT4K]

I looked at those benches.
First, it is being compared to the Willamette, NOT the Northwood.
Second, it is being compared to the Athlon Thunderbird, NOT the XP.
Third, look at more than one set of benchmarks before you make up your mind.
Try These from Anandtech

True, the Tualatin 1.2 beats some of the P4 Willamettes in some benches.
But they also show the Tualatin 1.2 getting beat by the 1.8 Willamette and the 1.4 Thunderbird in virtually all of the benchmarks. And we aren't even talking Northwood or XP here.
Yes, the first P4 Willamettes were disappointing and the Tualatin is a very nice chip.
But for much less money, an Athlon XP 1800+ or a Northwood 1.8a would crush that Tualatin. And they can be overclocked too.

 

[Diable]

Did you look at all the test THG ran in that review you linked to? The reason I ask is the 1.26GHz Tualatin got spanked by the 1.8GHz P4 in all the processors intensive test (CinBench, Divx4 encoding, Quake3 and AquaMark). It only beat the P4 in SysMark (which is office apps), a couple of Sandra benches (which are worthless), 3DMark 2000 (also worthless) and UTBench (if UT is one of your favorite games its worth something). After looking at all the benches the Tualatin looks like a overprice POS compared to the cheaper P4 and Athlon.

If I were to guest the outcome of a 1.57GHz Tualatin vs. a 2.4GHz P4 bake off I'd say based on THG's numbers the P4 would win and easily. Even if your running apps that are "unfriendly" to P4's like 3DStudio Max or Maya they will get a boost from the extra bandwidth and mhz of a 1.8GHz P4 running on a 133mhz fsb plus more and more apps are using SSE2 so your best bet is to get a 1.8GHz P4 and overclock it, like I did .

 

[Fun Guy]

I don't believe I got personal until someone called my mentality flawed. Up until that point I simply was playing devil's advocate. If I saw or was provided with evidence that a P4 at 2.4 could best a PIII-S at 1.575, I'd go for it. I'd GLADLY go for it.

Bottom line is I was willing to entertain the idea of getting a P4 and wanted someone who had truly put some thought into it and also did some research. Clearly I was expecting way too much from the crowd around here.

That's fine. I'll do my own comparison - and Wolfsraider, I will post the results when I have them. Then we'll see about all of the 'crushing' that will transpire. I bet the difference will be very slim.

As far the die shrinks, Shanti, the only things die shrinks enable on the P4 and the AMD are faster speed and cooler operating temperature. I don't believe a 1.8 and 1.8a will have any difference in performance, although I am more than willing to be wrong on this (i.e. show me). So the argument about Willamette vs. Northwood or TBird vs. XP essentially has no teeth.

Now the Tualatin die shrink - especially in the case of the 1.26 and 1.4 PIII-S's - is a completely different story. The L2 cache on these two chips doubles to 512K, and this on-die cache runs at clock speed, so it was more than just a simple die shrink. This is where the playing field got changed. Couple this with P4 problems (too many to list) and you actually have a good race going.

I am leaving this thread simply because it is useless. I will be back when I build the two boxes and have some results.

 

[DT4K]

It's not exactly just a die-shrink.
And there is definitely a performance difference between the Northwood and Willamette, and between the Thunderbird and Palomino cores.
Look at these:
Willamette vs Northwood
Thunderbird vs Palomino

But nobody is going to stop you from buying a Tualatin. You asked for our opinions and we were trying to help. I would really like to see some screenshots of your Tualatin @ 1.575 beating a Northwood @ 2.4Ghz or an XP @1.8Ghz

 

[codehack2]

<< Now the Tualatin die shrink - especially in the case of the 1.26 and 1.4 PIII-S's - is a completely different story. The L2 cache on these two chips doubles to 512K, and this on-die cache runs at clock speed, so it was more than just a simple die shrink. >>



How so??

You do realize that

1) With the die shrink on the P4, the L2 cache doubled to 512k, just as it did with the PIII-S...

2) PIII's have had l2 cache running at clockspeed way before this die shrink...

3) All 3 chips, p4, Athlon XP and PIII-s have L2 Cache running at clockspeed...

 

[Fun Guy]

Okay, now I am learning something.

So you guys think my test will be a rout in favor of the P4, huh?

 

[Wolfsraider]

i really would be intrested in this but then i also think the p-4 would beat a tulatin.

that said hmmm.. it would be nice to know what the price/performance differences are
i for one am lost in the numbering schemes and as i don't have a degree or first hand knowledge
i have to rely on those that do. most here know the differences and debates based on someone else's observation's.
not all as there are a lot of people here who know tons because of their personal experiences.

so the real question boils down to "do you really want/need the comparison"if the answer is yes go for it but please post it for this
computer junkie to add to his collection lol

good luck in your quest fun guy and have a nice day

 

[RaynorWolfcastle]

Things to keep in mind:
A) The P-III's pipeline is much shorter than the P4's, all else being equal, the P-III-S would smash the P4 on branchy code (such as office apps). The P4's higher clockpeed help close the gap, but for 95% of users of office apps the bottleneck is between the chair and the monitor.

B) The P4's higher clockspeed and SSE2 optimization and high memory bandwidth make it a better bet for apps with predictable code (streaming media, media encoding, etc). These are the places where you would most likely see a difference in real life.

C) Finding SDRAM that's good for higher FSBs than 150 MHz may be difficult. Finding PC2700 memory to accomodate a 133+ MHz bus for the P4 is much easier.

D) There is no such thing as a "best processor", different processors have different strengths. Choose the one that's best for what you do, in most cases the P4 Northwood at 2.4/133 will likely be faster than the PIII Tualatin at 1575/150.

Good luck on the build

-Ice

 

[PH0ENIX]

The notes about the IPC, length of pipline, and the complete uselessness of clock speed ratings are certainly true.

imgod2u was certainly not spouting 'useless gibberish', and I cant say i've ever seen him do so.

Now to say that the willamette - northwood transition wasn't anything special, and that a 1.8 would be the same as a 1.8a, is just plain wrong - I think everyone's covered that fairly well. One major indicating factor is how much the Athlons trounced the P4, previous to the inception of the Northwood core.

Now a p3 tualatin would be my preference over a willamete P4, for certain.
But you're comparing a 1.13 micron p3 to a 1.18 micron p4 (one that never really showed much strength in any department), and an outdated athlon to boot.

Lets see that same test with the right processors.

If you're planning on building these 2 systems, I think you'll find that everyone here has been right from the outset.

And on another note

[Flame] I agree with the suggestion that you come here to seek peoples opinions. I do not see how any of the opions put forward could be classified as P4 zealotry, let alone anything less than informative. By asking for someone's opinion on something, then being told the same thing by multiple people, only to proclaim that they're a bunch of idiots, and the FACTS that they are stating dont mean anything, only serves to suppliment that statement about your mentality.
Thats all i'm going to say on the matter - you can do whatever you want on this forum, after all it's not real life, so arrogance means little. [/Flame]