Comparing the technical merits of the Athlon and Pentium 4

JustHalf

I'd like to see people compare the technical merits of the Athlon and Pentium 4 chips that AMD and Intel (respectively) have released. The cost of the chips is not to be considered.

I don't want people saying that "AMD rocks, Intel sucks" or the like. If this is the case, explain why it is so, and your reasoning.

I'd like the following grounds covered at least:
* Performance at stock speeds
* Overclocking headroom in current and previous processes
* The impact of RAM/chipset/motherboard choice on the performance of the system
* Chip features and support

Thanks.

Gonzo

Ive been waiting to buy the new 2.53ghz pentium 4 from Dell which was released on may6th but they still havent brought it out to Dell ireland yet all other dell countries have had it for past weeks.

PPC

My xp 1900 running at 1.754 (12*146):

Quorthon

Originally posted by Gonzo
Ive been waiting to buy the new 2.53ghz pentium 4 from Dell which was released on may6th but they still havent brought it out to Dell ireland yet all other dell countries have had it for past weeks.

hehe they passed on the price cuts tho (which I was waiting for) so the 2.2 dropped €300 and I spent this money on a 21" trinitron monitor instead!!


Q

Gerry

How about discussing it, instead of spamming pictures of benchmarks?

JustHalf

Okay PPC. You want to show some real-world benchmarks?

PPC

Originally posted by JustHalf
Okay PPC. You want to show some real-world benchmarks?

Define realworld and i'll do em.
3dMark2001?
Quake3?
Halflife?

Just say what.

zAbbo

No matter what u do PPC, i think you`re gonna end up losing. The fact is that Gerry`s feathers have been ruffled in an ealier thread on gaming machines. I still can`t understand why Gery advised a guy on a budget to buy a big feck off P4 when the chap was on a budget(http://boards.ie/vbulletin/showthread.php?threadid=54302)

It will be interesting so see what happens here?

Having read into this a little i`m siding with PPC his Sisoft Sandra benchies are pretty good. Although that is his own machine which is prolly done up to the hilt.

Gerry

What do you mean my feathers have been ruffled? I made some points, tried to back them up, and got completely ignored by PPC and yourself as well by the looks of it. I have stated that the p4 is more expensive, by about 60 - 70 euro. In my opinion its worth that extra cash. I don't call a p4 1.6 costing 205 euro a big "feck off" p4. I put together a full system based on that combination for under 1100 euro.

I was trying to put across the point that the p4 will overclock more, and end up faster overall. Note that the starting post in this thread asks for a discussion. You are siding with PPC because the benchies look pretty good? Why do you feel the need to take sides? This is supposed to be a discussion board, so why don't you read the first post again and post your opinions about the technical merits of the 2 chips.

elexes

i see both points on this but i think the post want physical and mathamatical proof that the p4 runs faster then the amd xp's

JustHalf

Originally posted by PPC
3dMark2001?
Quake3?
Halflife?

Unless people are going to use their computers to run SiSoft Sandra or 3DMark2001, then and example real-world benchmarks would be Quake III (at lowest detail and resolution settings to rule out graphics card bottlenecking as best you can).

And now, here comes the important part. This has to be COMPARED to a P4 system, preferably a Northwood. Is that "2GHz Pentium 4" even a Northwood? If not, you're comparing the latest Athlon core to a much older P4 core. And I've never been one to trust those built-in results, ever since my KT133A board beat their built-in score for memory performance on that chipset substantially at tweaked speeds, and lost badly at default settings (KT7A-RAID).

Quigs Snr

Quake 3 is not a great benchmark for todays huge machines. If you are going to compare, then you need identical hard disks, RAM, and Video Cards at a minimum.

PPC

Originally posted by JustHalf
Unless people are going to use their computers to run SiSoft Sandra or 3DMark2001, then and example real-world benchmarks would be Quake III (at lowest detail and resolution settings to rule out graphics card bottlenecking as best you can).

And now, here comes the important part. This has to be COMPARED to a P4 system, preferably a Northwood. Is that "2GHz Pentium 4" even a Northwood? If not, you're comparing the latest Athlon core to a much older P4 core. And I've never been one to trust those built-in results, ever since my KT133A board beat their built-in score for memory performance on that chipset substantially at tweaked speeds, and lost badly at default settings (KT7A-RAID).

Rightio. I should be home tonight so i'll becnh it then if i can get that damn timedemo to work.

Anyone tried the new HL one from Madonion?

Gerry

Well who has a p4 to run benches on? I built that p4 for a friend, so I no longer have it.

Quake3 scales to todays machines, I don't see the problem with it. It will keep running faster the more horsepower you throw at it, in a linear fashion.

PPC

Will i get my friend to run it on his 1.8Nw @ 2.1Ghz?

Gerry

Ya that sounds good. If the machines have slightly different graphics cards, you can minimize the effect by running everything on bottom detail, we only want to test the cpu/motherboard/ram combination.

PPC

He's got a 1.8NW, 512Mb DDR and a TH7 II mobo and a GF2.
I just need to clear the details and get him to get Q3.

sutty

Originally posted by JustHalf
If not, you're comparing the latest Athlon core to a much older P4 core.

I'd just like to point out that although the XP is a new chip, it is still based on the first athlons, there have not really been any changes made to the core it's self, more to what was used to make the chip, extra SSE codes put on it.....but under all that, it is a 4+ year old core made to compeat with the P3.....so infact it is the athlon that is getting the unfiar test. All this test shows is that the athlon although old is still going strong. If you want this test to be truly fair, it should be Athlon V P3. Again there are somethings that the athlon is good at and others that the P3/4 is good at. Also sandra is a real world test, as you do run it on your PC. However you are right in saying dont trust the pre-set scores, you should use a PC made with the other chip type to get a fair reading on both CPU's/chip sets/ram
also the P4 should be using DDR ram, as RDRAM can slow its CPU proformance down and give the P4 a unfiar memory bandwidth.
as stated in a post above, you need to have the componets of both systems the same....ie: Ram, HDD, 3dcard, ect.....

Gerry

Erm, the xp is about 10% faster at the same speed, it includes hardware prefetch as well as sse. Theres no point in saying that its unfair, the p3 is based on the 10 year old p6 design, so wheres the logic in that. The comparison compares the current chips from both companies. Intel are concentrating on the p4, not the p3. Having said that, if they wanted to, they could have a 1.8 - 2.0 ghz p3 out now, with a fast bus to make good use of ddr. A 1.4ghz p3 Tualatin keeps up quite well with an athlon xp 1600 - 1800, but it costs way more also.

sutty

Phil what I meant was that the Athlon was made to compeat with the P3, and not the P4. But you are right, the P3 was always a great chip, We done some tests our self when I first got my 850 athlon and we both set our chips to 1Gz.....if I remember right you kicked my athlon around

PPC

Yeah P3 was great. Had one till i went to my 1GHz athlon.
Always Intel before that.

JustHalf

Originally posted by sutty
Phil what I meant was that the Athlon was made to compeat with the P3, and not the P4. But you are right, the P3 was always a great chip, We done some tests our self when I first got my 850 athlon and we both set our chips to 1Gz.....if I remember right you kicked my athlon around

If the Athlon XP was released to compete with the PIII, then AMD are seriously retarded.

Inspector Gadget

Firstly, there have been three different Athlon cores:

* K7 ("Classic")
* Thunderbird
* Palomino (XP)

As well as three Pentium III cores:

* Katmai
* Coppermine
* Tualatin

...and two Pentium IV cores:

* Williamette
* Northwood

As they all have different performance characteristics relative to each other, it's specious to say that the Athlon was designed to compete with the Pentium III, because every device involved has evolved, more than once at that, not to mention there being the occasional new entry to the scene (the Pentium IV), so basically the Athlons have been designed and updated to keep pace (sometimes leading, sometimes trailing) with whatever Intel spew out, and vice versa for Intel with it's PIII/PIV's competing with whatever AMD produce.

Additionally, especially when comparing the PIV family to the Athlon family, they're both designed to go fast, but in different ways. The Athlon family executes far more instructions per clock cycle than a P4 ever will (many shallow pipelines as opposed to fewer, longer ones), but the P4 goes through an awful lot more clock cycles, so they (more-or-less, depending on what applications you're using and who's fastest this week) line up.

Finally, when it comes to benchmarks, surely if you're using an application (such as Quake III) to test the relative performace of CPUs, won't the CPU family that the compiler used to build the application is optimised for (in all likelihood, this means Intel) be at (at the very least) a slight advantage?

Only my 2c, but...
Gadget

PPC

Thats why i used Sandra cause i dont think it was designed for Intel's.
Should we bench it on the system aswell?

Dazza01

Originally posted by Gonzo
Ive been waiting to buy the new 2.53ghz pentium 4 from Dell which was released on may6th but they still havent brought it out to Dell ireland yet all other dell countries have had it for past weeks.

Gonzo,

Just to let you know, The 2.5ghz is actually available from Dell ireland at the moment.....

I saw few being sent back for damage by the muppets in Dell

JustHalf

Gadget, if every application is optimised for one chip and not for the other, then the "theoretical performance" of the chip doesn't matter.

_CreeD_

From a different perspective (ie. Real world, non-gaming). I use 3 different PCs for music, Sonar 2.0 as the sequencer.
2 are Pentium 4s, 1.7 (work machine) and 2.0Ghz-(Northwood core), the other is an Athlon XP running at ~1.64Ghz.
The most intensive song I've done to date uses 22 tracks, with 11 instances of DXi's (Virtual Synths, VERY cpu hungry, the ones I used dont use much ram or stream from the HDD so Im ignoring those for the mo.). If you're not used to using a sequencer then to put it simply when you run out of CPU power your DXi's start to dropout/stall (also affected by the soundcard, but in this case all 3 have Audigy's).
On playback the 1.7Ghz P4 drops out about 3/4 of the time.
The 2.Ghz P4 drops out rarely, maybe 1/5.
The AthlonXp 1.64Ghz never drops out.


For game performance the machine configs are too different for a fair test.

Inspector Gadget

Originally posted by JustHalf
Gadget, if every application is optimised for one chip and not for the other, then the "theoretical performance" of the chip doesn't matter.

Well, yes and no - the fact remains that they're both trying to execute the same instruction set, more or less - it's just a matter of how good each chip is at doing that, isn't it?

All I'm saying is that an application that's been written with the intention of being run on a specific CPU may not be an ideal model for a benchmark. Not that I believe there is an ideal model, mind you.

Gadget

JustHalf

Of course not. For example, Intel's SysMark wouldn't be the best tool to use when comparing the P4 to the Athlon (it would still be fairly useful when comparing different P4's though)

However, if you're using a widely used application (like Quake III and the nVidia drivers of course) then you don't need to worry about whether or not the compiler used was optimised for either platform. This is because you should care far more about how well the application you want to run on your machine runs on your machine.

Personally, I don't give a damn if my machine has a lower SiSoft Sandra Multimedia benchmark than a given Athlon if, because of the chip, it plays Quake III faster.

It's also interesting to see people compare stock speed P4's against overclocked Athlons... given that your FSB is probably higher than standard.

Inspector Gadget

Originally posted by JustHalf
It's also interesting to see people compare stock speed P4's against overclocked Athlons... given that your FSB is probably higher than standard.

It certainly is - although again that's not black-and-white either as stock Athlons can have FSBs up to 266MHz where stock P4's can have FSBs up to 533MHz! Either way you look at it, it's not going to be "fair", is it?

To my mind, anyway, the whole matter of benchmarks is a bit dodgy to begin with. I think the "speed" of a machine is subjective - if it feels fast when I use it, I'm a happy bunny.

Gadget

_CreeD_

Originally posted by Inspector Gadget



To my mind, anyway, the whole matter of benchmarks is a bit dodgy to begin with. I think the "speed" of a machine is subjective - if it feels fast when I use it, I'm a happy bunny.

Gadget

Exactly. Whatever does the best job at the things you want it to do, for the money you're prepared to pay for it.
It doesn't matter a damn which one gets a higher benchmark if it won't run the specific things you want as it should. Why bother running things like Q3 at lower settings to stress the CPU if you're going to play it high enough to make your VidCard the bottleneck (Which everyone will logically enough).