Search
Sign in with
Sign up | Sign in

P3's better than early P4's? is it true or false?

Last response: in CPUs
Share

Hi y'all...

I think I read somewhere sometime ago that Pentium 3 processors were better performers than 1st generation Pentium 4 processors. Is it true? If so, can you tell me where can I find more info. about this issue?

A friend of mine is a big Intel fanboy and I want to give him a "present", if you know what I mean... :D  :D  :D 

Thanks.... :!:

More about : early true false

Register or log in to remove

Yes, it's true. A 1.4GHz Williamette P4 only performed as well as ~1.1GHz P3. The whole "more work per clock" thing that had AMD leading in performance up until Northwood C.

Quote:
Yup, it's true in lots of cases. Intel with the high clocks on the P4, to sell them, that's probabl then main reason.

But, why talk about the past, we can talk AM2 and Conroe, not P3 and P4.
Cause the p3 was great and didnt generate alot of heat like the p4

The Pentium 3 was a better and more efficient Arcetecture than the Pentium 4, But Intel got greedy and just wanted to Follow "Moores Law" and they gave up Efficiency and performance for heat and Speed

But seems like they've realised that performance is'nt everything it's efficiency and workload
A thing that AMD has been following for the entire Pentium 4 legacy
Related ressources
Ask the community

Quote:
The Pentium 3 was a better and more efficient Arcetecture than the Pentium 4, But Intel got greedy and just wanted to Follow "Moores Law" and they gave up Efficiency and performance for heat and Speed

But seems like they've realised that performance is'nt everything it's efficiency and workload
A thing that AMD has been following for the entire Pentium 4 legacy


Moores Law has nothing to do with greed it was a claim that every 18 months performance of IC's should increase by 2x.

Quote:
Arrrrggg Shiver me Timbers! K8MAN be speaking the Truth~ arrrggg!!!!! I also found a P3~1.6Ghz that performs about as good as some of the 1.8GHz to the 2.2GHz P4s


I was under the impression they capped at 1.4, it a special contract model?

Thank you guys for all your replies.

I was hoping to get some references to sources in the industry that might have detailed documentation about this topic, so that my friend can't deny this fact.

Anyway, you already confirmed what I was investigating, so you've been very helpful!

Thanks again...!

Quote:
Moores Law has nothing to do with greed it was a claim that every 18 months performance of IC's should increase by 2x.


I believe Moores Law is only loosely related to performance; namely that Moores law is related to the density of transistors in a given die space would double every 18-24 months.

http://www.webopedia.com/TERM/M/Moores_Law.html

Quote:
The observation made in 1965 by Gordon Moore, co-founder of Intel, that the number of transistors per square inch on integrated circuits had doubled every year since the integrated circuit was invented. Moore predicted that this trend would continue for the foreseeable future. In subsequent years, the pace slowed down a bit, but data density has doubled approximately every 18 months, and this is the current definition of Moore's Law, which Moore himself has blessed. Most experts, including Moore himself, expect Moore's Law to hold for at least another two decades.


Also: http://www.intel.com/technology/magazine/silicon/moores...

netburst can handle high clock speeds at the cost of performance. if we had a p4 and p3 with exactly the same clock speed, the p3 would be a lot faster.
but the market likes high numbers, like 2, 3ghz... that's why amd came up with the PR stuff, coz if they were selling based on their mhz, people would think that a 1.8 ghz amd is slower than a 2ghz p4 for example.

by the way, p3 is so much better than p4, that core architecture is based on... p3

The P3 Tualatin with 512 cache were superb processors. They really challenged the early Willamette processors. Tom's published an excellent article here:

http://www.tomshardware.com/2001/09/19/last_passing_man...

Unfortunately they were overpriced and in direct competition with the P4's. They were cool and efficient. I had the 1.26 Ghz version and while SSandra gave my P4 2.0 Ghz C a significant "performance" advantage, I couldn't see it in real world gameplay.

I used 2 different PIII chips and was so discusted with how bad all of Intel's netburst products have been that for my desktop computing needs i've used AMD chips of late, though i'm very interested to see just how good Conroe will be, but hey, i'm digressing a little, thats right - the PIII.

The first I used was the PIII 500E (coppermine core) which overclocked really nicely and with it's on-die CPU speed L2 cache it was also very effective, I ran it on a VIA Apollo Pro 133 chipset as the aging BX chipset wouldnt clock properly to 133 (BX only supported a 3x divider for the PCI bus, running it above 112MHz FSB wasnt really possible) and the less said about the i820 chipset the better - it plainly didnt work!

The second PIII I used was the 1.4GHz chip that used Intel's superb Tualtin core (it was one of the first processors to be based on a 130nm process). I benchmarked this against a mate's then brand new 2GHz P4 Prescott based machine and the PIII won convincingly, so the later PIII's were only really beaten once netburst's clock speed got into the range of being a full 1000MHz faster!

The Tualtin's life didnt end with the PIII though, it went on to become the basis of the Pentium M, which in turn is the basis of Intel's new Core arcitecture, conveniently brushing the whole netburst mistake under the carpet!

No wonder my P3-800EB (512k, 133FSB) ran XP just about as well as my P4 1.8 did on the same ram/graphics card. PC800 RDRAM was this shit on that sys (f'ing expensive though) and I had a motherboard for both the slot 1 and the socket 478 that used it. Long gone now though...well, the P4 1.8 is still hanging around (HTPC).

My old p3 at 800 mhz didnt even need a heatsink fan... :lol: 

though, I would never trade it back for my northwood c.
When you add 2 ghz of headroom on a p4 it leaves the p3 in the dust.

Quote:
as the aging BX chipset wouldnt clock properly to 133 (BX only supported a 3x divider for the PCI bus, running it above 112MHz FSB wasnt really possible)


Probably splitting hairs here and getting off the subject, but the BX supported 4x devider at 133. It did not however support anything but 1:1 and 2/3 AGP, so at 133 the AGP ran at 83.

I have several BX boards overclocked, so that's the only reason I know that. :wink: 2 of them I've had for over 8 years and they still work well, for what they are (1.1 celerons oc'd to 1.45G)

Those first P4's were terrible. I remember having a 1Ghz Athlon and it running circles around 1.4 Ghz P4's. It felt like I was working on an "old" computer on brand new P4's.

RDRAM - that was a fiasco bordering on conspiracy which was driven by nothing more than Intel's greed and was totally oppposed to consumer demand. The story goes that Intel forged a deal with RAMBUS whereby if Intel could push thier RDRAM standard to the biggest share of the market they would get a 50% stake in the company and various other favourable deals.

The first attempt that this was Intel's i820 chipset for the PIII-E with a 133MHz FSB. This platform was originally designed to work with RDRAM and RDRAM only. Unfortunately for Intel, board makers and consumers were not prepared to purchase RDRAM which at the time was extremely expensive and had very limited availability and appeal. In a rushed last minute gasp to try and save the i820 platform they converted the chipset to PC133, unfortunately they didnt do a very good job and users experienced horrific reliability problems which eventually led to a product recall, refunds for all purchasers and the canning of i820.

Wind forward to the release of the P4 - Intel now had a golden opportunity to monopolise RDRAM for thier own gains, so when the P4 was released the only RAM supported was RDRAM, which was not just expensive but also slow. Theres actually very little about RDRAM which was good, it had a very narrow bus, just 16bits, which meant to match SDRAM (64bit) it had to run 4x faster, but thats just the start. RDRAM was very difficult to manufacture and many manufacturers were reluctant to do so due to low yields and having to pay royalties to RAMBUS for the privelage. RDRAM was never popular with consumers, it was expensive, it was slow, it ran very hot, Intel tried to push it by bundling it with the Pentium 4 but in the end boards came to market for the P4 that supported SDRAM - SDR SDRAM at first too, which although a better and cheaper alternative to RDRAM was hardly state of the art as AMD platforms were already running DDR266 SDRAM. RDRAM faded away and in desperation RAMBUS tried to claim that they invented SDRAM (a complete fabrication) and tried to charge royalties to those who make it. Some of the companies stupidly paid up but others contested it - and won. The directors of RAMBUS have now been tried for fraud and Intel has tried to distance themselves from them ever since.

Quote:
RDRAM - that was a fiasco bordering on conspiracy which was driven by nothing more than Intel's greed and was totally oppposed to consumer demand. The story goes that Intel forged a deal with RAMBUS whereby if Intel could push thier RDRAM standard to the biggest share of the market they would get a 50% stake in the company and various other favourable deals.

The first attempt that this was Intel's i820 chipset for the PIII-E with a 133MHz FSB. This platform was originally designed to work with RDRAM and RDRAM only. Unfortunately for Intel, board makers and consumers were not prepared to purchase RAM which at the time was extremely expensive and had very limited availability and appeal. In a rushed last minute gasp to try and save the i820 platform they converted the chipset to PC133, unfortunately they didnt do a very good job and users experienced horrific reliability problems which eventually led to a product recall, refunds for all purchasers and the canning of i820.

Wind forward to the release of the P4 - Intel now had a golden opportunity to monopolise RDRAM for thier own gains, so when the P4 was released the only RAM supported was RDRAM, which was not just expensive but also slow. Theres actually very little about RDRAM which was good, it had a very narrow bus, just 16bits, which meant to match SDRAM (64bit) it had to run 4x faster, but thats just the start. RDRAM was very difficult to manufacture and many manufacturers were reluctant to do so due to low yields and having to pay royalties to RAMBUS for the privelage. RDRAM was never popular with consumers, it was expensive, it was slow, it ran very hot, Intel tried to push it by bundling it with the Pentium 4 but in the end boards came to market for the P4 that supported SDRAM - SDR SDRAM at first too, which although a better and cheaper alternative to RDRAM was hardly state of the art as AMD platforms were already running DDR266 SDRAM. RDRAM faded away and in desperation RAMBUS tried to claim that they invented SDRAM (a complete fabrication) and tried to charge royalties to those who make it. Some of the companies stupidly paid up but others contested it - and won. The directors of RAMBUS have now been tried for fraud and Intel has tried to distance themselves from them ever since.
True so true if intel had kept the p3 and if it supported ddr man i would still be running with intel fanboys :lol: 

Early P4s were no match for P3s
When Northwood arrived, P4s finally managed to strike back but until then everyone went off to buy P3 instead of P4.
P6 is the best processor design on this planet and also other planets (*lol*).

Just consider this: a Pentium-M at 2.13GHz is quite as fast as an Athon64 running at 2.0GHz and the Pentium-M internal architecture is derived by the P3, not the P4!

Toms has published a benchmark review some months ago about this, making the tests with an adapter board that makes possible to mount a Pentium-M on a standard desktop ATX mobo (ASRock makes such boards): search for this of you need documentation.

I had the good fortune of acquiring a matched pair of Pentium III-S (Tualatin) 1.26 GHz CPUs from a Dell server. The trouble is finding a motherboard that supports the Tualatin chips. Any suggestions? I would love to overclock 'em to see what they can do.

Interestingly - I popped one in my dad's Dell OptiPlex GX150 and it recognized the CPU properly in the BIOS and later I confirmed it using CPU-Z. Too bad that there's virtually no tweaking allowed by the Dell BIOS/mainboard.

Early P4s were terrible, especially when paired with SDRAM and i845 chipset. 1 GHz P3 could easily keep up with 1.7 GHz P4 + i845 + SDRAM.

I will never forgive Intel for releasing i845 SDRAM chipset. So many poor n00bs bought such crappy systems... :cry: 

Quote:
Moores Law has nothing to do with greed it was a claim that every 18 months performance of IC's should increase by 2x.


I believe Moores Law is only loosely related to performance; namely that Moores law is related to the density of transistors in a given die space would double every 18-24 months.

http://www.webopedia.com/TERM/M/Moores_Law.html

Quote:
The observation made in 1965 by Gordon Moore, co-founder of Intel, that the number of transistors per square inch on integrated circuits had doubled every year since the integrated circuit was invented. Moore predicted that this trend would continue for the foreseeable future. In subsequent years, the pace slowed down a bit, but data density has doubled approximately every 18 months, and this is the current definition of Moore's Law, which Moore himself has blessed. Most experts, including Moore himself, expect Moore's Law to hold for at least another two decades.


Also: http://www.intel.com/technology/magazine/silicon/moores...

Please don't assume I don't have access to the internet.

Quote:
RDRAM - that was a fiasco bordering on conspiracy which was driven by nothing more than Intel's greed and was totally oppposed to consumer demand. The story goes that Intel forged a deal with RAMBUS whereby if Intel could push thier RDRAM standard to the biggest share of the market they would get a 50% stake in the company and various other favourable deals.

The first attempt that this was Intel's i820 chipset for the PIII-E with a 133MHz FSB. This platform was originally designed to work with RDRAM and RDRAM only. Unfortunately for Intel, board makers and consumers were not prepared to purchase RDRAM which at the time was extremely expensive and had very limited availability and appeal. In a rushed last minute gasp to try and save the i820 platform they converted the chipset to PC133, unfortunately they didnt do a very good job and users experienced horrific reliability problems which eventually led to a product recall, refunds for all purchasers and the canning of i820.

Wind forward to the release of the P4 - Intel now had a golden opportunity to monopolise RDRAM for thier own gains, so when the P4 was released the only RAM supported was RDRAM, which was not just expensive but also slow. Theres actually very little about RDRAM which was good, it had a very narrow bus, just 16bits, which meant to match SDRAM (64bit) it had to run 4x faster, but thats just the start. RDRAM was very difficult to manufacture and many manufacturers were reluctant to do so due to low yields and having to pay royalties to RAMBUS for the privelage. RDRAM was never popular with consumers, it was expensive, it was slow, it ran very hot, Intel tried to push it by bundling it with the Pentium 4 but in the end boards came to market for the P4 that supported SDRAM - SDR SDRAM at first too, which although a better and cheaper alternative to RDRAM was hardly state of the art as AMD platforms were already running DDR266 SDRAM. RDRAM faded away and in desperation RAMBUS tried to claim that they invented SDRAM (a complete fabrication) and tried to charge royalties to those who make it. Some of the companies stupidly paid up but others contested it - and won. The directors of RAMBUS have now been tried for fraud and Intel has tried to distance themselves from them ever since.


Actually the chipset worked as intended with regards to RDRAM and SDRAM, it was later pointed out that the MTH riser cards provided by the motherboard manufactures was not built to specs provided by Intel, causing electrical noise during the process of changing the signal from RDRAM specifications to SDRAM specifications. Which in turn caused machines to hang, spontaneously reboot; in response Intel provided a replacement for the customers that managed to get boards with the MTH.

Quote:
Early P4s were terrible, especially when paired with SDRAM and i845 chipset. 1 GHz P3 could easily keep up with 1.7 GHz P4 + i845 + SDRAM.

I will never forgive Intel for releasing i845 SDRAM chipset. So many poor n00bs bought such crappy systems... :cry: 


Thankfully your forgiveness is worth the same as my opinion to Intel.

My first attempt at a computer build was September 2001.
At that time things were getting interesting.
The ATi 8500 made it's debut (DX 8.1)
XP was right on time in its debut and the P III Tualatin core processor for the Celeron was here.
The days of having to deal with a meager 800mhz cut core cele were over! (P3 was running $350, Cele was $125 w/256 on die cache)
I was able to get an ASUS TUSL2-C motherboard and a Celeron 1200 mhz Tualatin core processor.
With the mobo unlocked that cele reaced 1450mhz without a hitch.
I'll never forget my friend getting a new P4 1.6 HP.
After doing some research I found out that my Cele actually outperformed his 'early' P4.
That wouldn't last too long as Intel revamped the P4 shortly thereafter.
At the same time they killed the Tualatin P3 core altogether (except to go mobile....that core lasted a while on Intel M)
because...............it was VERY popular and probably outselling the early P4. (just a guess)
Why else would intel kill a very good die? Age? Competition.
That Tualatin was the bomb and definitely something to grow on.
To answer your question, Until the P4 reached 1.8 and had it's second core change, the Tualatin P3/Cele kicked it's meager butt.
That Cele was the one to have if you couldn't afford the P3. Performance was Very Close.

Thanks, ejay, for the tip about the DFI board. Newegg won't ship to Canada but I found a few of the boards on ebay, one of which I bid on.

I'm not going to bother with a dual processor setup. I'll let my dad keep one of the Tualatin's and I'll build a system around the other one just for fun. I already have lots of computers at my place: an A64 @ 2.48 GHz, a Celeron D 345 @ 3.91 GHz, an Athlon XP 1800+, a couple of Pentium 4-M notebooks... you get the idea.

I have always enjoyed getting relatively high performance from cheap processors via overclocking, but I've never gone overboard with water cooling or things like that. Everything is done with stock fans and air cooling. I don't like to drop too much money on any one system because I get bored after a year or so and like to change things up. I'm not a gamer either which is fortunate because good video cards are bloody expensive. I have had my Winchester 3000+ for about a year now and I'm considering replacing it with an Opteron 165. It's cheap, has a good reputation for overclocking and would drop right into my Asus 939 board.

Quote:
hmmm... i have i845 that came with my sony paired with a 2.66 p4 and 1gb DDR ram. does this make me a noob?


Of course not. He specifically said the i845 chipset for SDRAM. The DDR i845 is a different revision (in total there are 5 different i845 chipsets I believe).


TO EVERYONE ELSE: The Tualatin PIII was better than a 1.8 GHz Willamette mainly in synthetic benchmarks, and that includes 3dmark and winstone, as one cannot "play" 3dmark, or do productive office tasks using Winstone. Most REAL applications, such as games and audio/video encoding, photo and graphical design apps, and even more serious programs such as autoCAD and scientific apps would be better served with a 1.8 ghz Willamette using DDR ram than the 1.4S with pc133 sdram. Taking cost into consideration and it becomes a no-brainer. The P4 wins that portion of the contest hands-down. On the performance level, the 1.4S and the 1.8 Willy trade blows almost equally, as it's about 50/50 win/loss ratio for both. One thing to note, however, is that the 1.8 P4 is never far behind the Tualatin 1.4S when it does lose in a benchmark. The same cannot be said for the opposite.

Saying that a 2 GHz williamette P4 (the Northwood P4 is completely out of the Tualatin's grasp) would be stomped by the 1.4S is shear lunacy. What I find perhaps most humorous of all is the rabid love for the Tualatin PIII by some people, who also happen to have intense hatred for P4-class Celerons, especially the Willamette and Northwood Celeron line. I suppose it would break their hearts to know that the 2.0 GHz Celeron that they find so vile actually performs equal to the revered 1.2 GHz Tualatin PIII most of the time, which brings up a very interesting concept: in a die size comparison (both at .13 Micron) the Tualatin vs P4 Northwood is no contest at all. The Tualatin is actually much more comparable to the Northwood Celeron, if one looks at the two from the die size standpoint, making the Tualatin look incredibly poor.

Feel free to peruse THG's own benchmark marathon to prove my point.

http://www.tomshardware.com/2003/02/17/benchmark_marath...

Quote:
Socket 423 P4s all sucked.

What's the difference in performance between Katmai and Tualtin core'd P3s?


The difference is huge, as you've skipped the entire Coppermine generation with that comparison. Most important differences were the L2 cache: on die full speed on the Tually and Coppermines, while the Katmai's L2 was at half cpu speed. Other difference of course is slot vs socket design (one again due to L2 cache differences really). The Tualatin was the first .13 Micron Intel chip. Katmai was what? .18 Micron? Katmai only went up to 600 MHz, while the Tualatin started at 1.0 GHz. (or was it 1.1?) The Tualatin's L2 cache was either 256 KB (Celeron and regular PIII) or 512 KB (Server); the Katmai always had 512 KB L2. That's all I can think all off the top of my head. There are more differences of course, most being a lot more technical.

Quote:
Hi y'all...

I think I read somewhere sometime ago that Pentium 3 processors were better performers than 1st generation Pentium 4 processors. Is it true? If so, can you tell me where can I find more info. about this issue?

A friend of mine is a big Intel fanboy and I want to give him a "present", if you know what I mean... :D  :D  :D 

Thanks.... :!:


OK now you fraggin AMD fanboys are just asking for it!!...

But yes that was true. Although isn't that true with a lot of things? Take the GeForce 6 & 7 series for example. The late "generation" 6 series cards caned the early 7 series generation. It just takes a while for the new technology to get itself into gear, this time in doing so is the time where the companies really make the most money off people hungry for the newest technology, no matter how it performs.

man stop being a pompous brat about it.

The reason I asked was because I've been having troubles with my dual Katmai-550 server and in other parts of the forum people have all gone a bit nostalgic saying how much they liked the Katmai, so I was wondering more in terms of 'at-the-time-of-release' which was a bigger step forward.

I'm quite surprised that everyone is so into the P3s to be honest. Last year when Tom's did their April Fool on Microsoft/Intel buying into a Time Machine and Intel 'quoted' on saying they would 'erase the P3 from history' sorta gave me the impression that it was something of a dog/lemon.

Certainly the Slot 1/Socket 378 thing was a big lesson to Intel, but darn, they repeated it again with the next generation (423/478). It's a shame they had to keep the Pentium 4 brand alive for so long, but I guess Pentium 5 would have sounded a bit strange.

Quote:
I had the good fortune of acquiring a matched pair of Pentium III-S (Tualatin) 1.26 GHz CPUs from a Dell server. The trouble is finding a motherboard that supports the Tualatin chips. Any suggestions? I would love to overclock 'em to see what they can do.

Interestingly - I popped one in my dad's Dell OptiPlex GX150 and it recognized the CPU properly in the BIOS and later I confirmed it using CPU-Z. Too bad that there's virtually no tweaking allowed by the Dell BIOS/mainboard.


Ebay has quite a few dual cpu server PIII boards that are Tually capable. That might be your best bet. The prices aren't cheap though. :cry: 
CPUs Expert

P3 1400 Tualatin was the last P3 and was better then most P4 wilamettes while outputting 1/2 the heat, didnt require a new psu or case and was pretty overclockable.

1st gen P4's should never have been released - expensive hot and slow.

Even to today the P3's are better then the P4 - Pentium M should have been called the pentium 4 cause thats where the P6 design went and within 6 months its coming back as the baises of conroe.

and btw:

Dual Core Yonah 2ghz = ~30w
Dual Core P4 with same performance = 130w

go figure.


P6 forever!


Quote:
Hi y'all...

I think I read somewhere sometime ago that Pentium 3 processors were better performers than 1st generation Pentium 4 processors. Is it true? If so, can you tell me where can I find more info. about this issue?

A friend of mine is a big Intel fanboy and I want to give him a "present", if you know what I mean... :D  :D  :D 

Thanks.... :!:

Quote:
Hmm does anyone here know where i can find a dual p3 mobo with agp and 133mhz fsb?


I think the Abit VP6 could handle what you need. Also I recall seeing a dual MSI MB that had support for dual P3s and DDR!!!!!!! ram.
CPUs Expert

Quote:
1st gen P4's should never have been released - expensive hot and slow.


hey, how do you think intel makes their money?

exactly

oh im an idiot, o look a 3ghz P4, thats loads faster then a 2ghz AMD, those AMDs sure stink.

Quote:
Actually the chipset worked as intended with regards to RDRAM and SDRAM, it was later pointed out that the MTH riser cards provided by the motherboard manufactures was not built to specs provided by Intel, causing electrical noise during the process of changing the signal from RDRAM specifications to SDRAM specifications. Which in turn caused machines to hang, spontaneously reboot; in response Intel provided a replacement for the customers that managed to get boards with the MTH.


Didn't know that, but I had one with the riser card and it worked fine. Its actually still running fine, I handed it off to a friend. P3800 Coppermine and a gig of SDRAM on the riser card. I guess I'm one of the lucky ones (or asus was good about building them).

Anyway there are several cheap dual 370 boards with agp on ebay. And don't talk bad about joefriday's rig, that thing is sweet! (except for that ME part *shudders*)
Register or log in to remove
 
Ask the community
Read discussions in other CPUs categories
!