3.2 Prescott vs. 3.2 Northwood at "The INQ"

[mamisano]

The Inquirer

Northwood beats Prescott in all tests but PC Mark 2004. Editor has it wrong with Cinebench. They state that Cinebench shows Prescott faster, but the benchmark is shown in seconds(lower better): Prescott 78.8 - Northwood 69.

Also states that Prescott was 10c warmer at same speed.

 

[hytek369]

Prescott is losing, lol

 

[pakotlar]

once again, I'm glad about my 3.5ghz p4

 

[SilverBack]

Well take the Inquirer with a grain of salt.
The chips should ramp better as far as frequency and the SSE3 isn't taken into consideration.

I'm still going to get one
Should do nicely in my water rig.







3dMark 2003

 

[JonnyBlaze]

remember when p4's first came out? it was the same deal. they were slow. wait till those things are 4.0ghz over clocked to 5ghz

JB

 

[beatle]

Originally posted by: JonnyBlaze
remember when p4's first came out? it was the same deal. they were slow. wait till those things are 4.0ghz over clocked to 5ghz

JB

Still there was a lot more to the P4's scaling than just the increased MHz. L2 Cache increased, FSB increased, and now HT are all improvements (I'm sure I'm missing a few others as well).

What is Intel calling these upon release? Will they be P4D's?

 

[DerekBaker]

Pentium 4E


Derek

 

[DAPUNISHER]

Once Prescott changes sockets and the clockspeed, boards, bios, and some apps are better optimized for it, the northwood will start to fall behind no doubt. I'd also like to see some HT testing with it to see if it does indeed have any advantage over the P4c right out of the gate or if it too will require optimizing for it before it shows it's stuff?

 

[Genx87]

Once Prescott changes sockets and the clockspeed, boards, bios, and some apps are better optimized for it, the northwood will start to fall behind no doubt. I'd also like to see some HT testing with it to see if it does indeed have any advantage over the P4c right out of the gate or if it too will require optimizing for it before it shows it's stuff?


In other words about 12 months from now?

 

[mamisano]

Originally posted by: DAPUNISHER
Once Prescott changes sockets and the clockspeed, boards, bios, and some apps are better optimized for it, the northwood will start to fall behind no doubt. I'd also like to see some HT testing with it to see if it does indeed have any advantage over the P4c right out of the gate or if it too will require optimizing for it before it shows it's stuff?

I guess that leaves the door WIDE open for the A64. They are going to rolling out the 3700+ and FX53 by the end of Q1.

 

[videoclone]

Intel are having big problems with these CPU's they are having a hard time getting them to 3.2Ghz let alone 4, 5Ghz

I think AMD Athlon64 & FX will have another whole Pentium 4 Revision to stomp all over until next core enhancement comes out from Intel ....

Intel need a new CORE ASAP if they want to stay competitive to the new Performance leader " AMD "

WindowsXP 64 and Windows lonhorn 64 are going to kill Intel's sales if they dont get off there high horse and admit they were wrong and implement 64bit NOW !

 

[tallman45]

It's a good bet that the Prescott will not fail, Why ? The Nasdaq and the stock market as a whole looks very closely at Intel. AMD is lesser of a force when it comes to it's impact on the economy.

If this is a flop then plan on seeing a huge cover-up. There are a lot of forces that want the economy to recover even if reports are fudged a little again (ie: MCI and Enron).

Somehow Intel will claim that this new proc is a huge improvement and subsequent success.

 

[PetNorth]

I'd also like to see some HT testing with it to see if it does indeed have any advantage over the P4c right out of the gate or if it too will require optimizing for it before it shows it's stuff?

Cinebench 2003 is official bench for Cinema 4D, and the CPU rendering bench of Cinebench take a lot of advantage from HT. Well, in this INQ tests Prescott is about 13% slower than Northwood in Cinebench CPU render...

 

[Duvie]

Originally posted by: PetNorth

I'd also like to see some HT testing with it to see if it does indeed have any advantage over the P4c right out of the gate or if it too will require optimizing for it before it shows it's stuff?

Cinebench 2003 is official bench for Cinema 4D, and the CPU rendering bench of Cinebench take a lot of advantage from HT. Well, in this INQ tests Prescott is about 13% slower than Northwood in Cinebench CPU render...

Actually the only enhancement HT will give in cinebench is if you run the renderer with multix processors...Thay have both a single cpu test and mutiple cpu test...I don't know what the heck the inquirer is doing reporting the times instead of the scores...If they had done that we could have looked at some of the past reviews to see if they are doing single or multiple.....

I see my p4 2.4@3.5 does the single test at 74sec and a score of 351 in multi cpu test I have a time of 62sec and a score of 421

With that in mind it appears they ran the HT test or multiple cpu test....Not a good sign....



Also what is the chance they have an early silicon chip like the ones floating around awhile back??? Some guy said he got one recently and he said after he got setup all in order his temps were about the same at same speed....So can we trust the Enquirer??? Do you all remember their articles about the A64 before release??? they were not that good as well, though I can't remember them actually testing one themselves.....

The score they listed above for multiple cpu test I think is worse then my 2.4@3.0c.....


I will wait for real review sites....I like PC Tech Report, Anandtech, and Tomshardware.....PC is the best overall in terms of conclusions and well rounded test, while Tom has the best testing for my likes and uses, and anandtech has sound conclusion just a very one sided testing suite...These 3 molded together can give you a nice picture....

 

[THUGSROOK]

so far every Prescott preview has shown it to be a POS.

 

[Freejack2]

That's kind of odd. The prescott wins in the non game stuff but loses in the games. Wonder if they tried this with a different video card.
I'm suprised about the heat though. The prescott should be running cooler.

 

[Duvie]

Originally posted by: Freejack2
That's kind of odd. The prescott wins in the non game stuff but loses in the games. Wonder if they tried this with a different video card.
I'm suprised about the heat though. The prescott should be running cooler.

Why is that??? It is still a .13micron chip....It has added cache which I heard usually makes the core hotter....I would have expected it to be the same or just a hair higher with all the rumored steppings they did to reduce heat and power leakage...

Was thischip supposed to have strained silicon process or just the upcoming .09 prescott version???

 

[SilverBack]

There are no .13 micron Prescotts

 

[PetNorth]

I see my p4 2.4@3.5 does the single test at 74sec and a score of 351 in multi cpu test I have a time of 62sec and a score of 421

So obviously INQ has tested Cinebenech in multi cpu test (with HT optimization):

Prescott 3.2: 78.8 seconds
Northwood 3.2: 69 seconds

Yours Northwood @3.5: 62 seconds

10% faster yours than North INQ, and this is (+- error 1-2%) the different we can expect between a 3.2 and a 3.5 clock.

Sincerely, if these results are confirmed (games as Quake and Comanche included), this is very pathetic...

And more, they say Prescott runs 10º hottest than equivalent clock Northwood, wow... this is horrible...

 

[Duvie]

You are rght!!! I guess I was thinking the socket change to 775 marked the actual .09 transformation...

If the heat issue is true on all of these chips, and there are mention of multiple versions floating about, then this is bad news and I don;t see how they get to heights they originally thought...

Usually when we have stepped dowe we have been dropping vcore and thus we got a slight reprieve from the heat...To have heat issues right out of the gate...YIKES!!!!

 

[Duvie]

Originally posted by: PetNorth

I see my p4 2.4@3.5 does the single test at 74sec and a score of 351 in multi cpu test I have a time of 62sec and a score of 421

So obviously INQ has tested Cinebenech in multi cpu test (with HT optimization):

Prescott 3.2: 78.8 seconds
Northwood 3.2: 69 seconds

Yours Northwood @3.5: 62 seconds

10% faster yours than North INQ, and this is (+- error 1-2%) the different we can expect between a 3.2 and a 3.5 clock.

Sincerely, if these results are confirmed (games as Quake and Comanche included), this is very pathetic...

The fact is my single cpu test is faster then the prescott granted at 3.5ghz but HT offers a 18-20% boost....Its single time must have been pathetic....It would have been nice to have ran that as well so we can see if HT scaled equally....


Looking back at my benches my 2.4@3.0ghz ran the HT test in 76 sec.....So it acts like a 3.0ghz northwood????

 

[PetNorth]

yeah... I suppose Prescott 30 stage pipeline are guilty lol

 

[Duvie]

I see a couple of things here...


1) this is starting to smell like the launch of the P4 willamette...you remember the on where the 1.3 P4 couldn't handle the P3 1ghz chip...yes it has SSE2 optimizations but like this SSE3 this time no programs currently taking advantage of them at the start....I starting to think Intel knew these things would perform worse and maybe hence the price out of the gate on them ...The optimzations to SSE3, HT and added L2 cache were needed to keep it close...When will they arrive to save the chip??? Who knows...Maybe once we are 600mhz higher from here they will start to trickle out...

2) Intel did the increased pipelines to add longevity to the chip in terms of ramping speed....Well if the heat is any idea this is going to be one short trip and retail Intel P4 will come with a Water cooling kit!!!! 10c over stock 3.2ghz means these things are not gong to oc well with air cooling alone....

3) I figure we will have to wait until 3.6-3.8ghz chips to feel like much of an upgrade....

 

[videoclone]

We are all into Computers we all know AMD are better so why do you people still buy Intel ? i dont see how anyone in these forums would still buy intel and now even an overclocked intel CPU to 3.5Ghz + only just keeps up with the Athlon64 in games.

 

[HurleyBird]

SSE3 offers far fewer instructions over SSE2 than SSE2 did over SSE1. Most of these seem to be for 'better hyperthreading'.

Its possible that the instructions for better hyperthreading need software support. If so, we should see improved versions of hyperthreaded apps taking advantage of this, and maybe a real big performance improvement.

Another rumor flying around is that the ALU's are no longer double-pumped, in order to keep the heat down. If thats true than that could be contributing to the poor performance.

Also, after Intel gets its heat problems solved (I'm guessing there pouring loads of money into SOI right now) the 30 stage pipeline should scale very well.

Lets not forget, theres also rumors that prescott is 64-bit compatible too.

What we are seeing right now is the worst case scenario. If it is this bad at launch, or even a month or two after launch than intel is going to loose marketshare to AMD (unless maybe if they start spending even more on marketing) but its entirely possible this deformed kitten might turn into a ferocious lioness, I just wouldnt bet any money on it