Can someone explain why p3 is faster clock for clock with a northwood?

[mrgoblin]

I see these prescott benchies and Intel just seems to have awful ipc. WHy is it intel chips are like this?

 

[Nebor]

What does it matter as long as the end result is the same.... You chose between effeciency and clock speed.... Either way, comes out about the same.

 

[mrgoblin]

So with current die shrinks, wouldnt a p3 core with new extensions on 90 nano technology at 2.4 ghz be able to compete with a 3.0 prescott?

 

[Accord99]

No, since the fastest 0.18um P4 Williamette was significantly faster than the fastest 0.18um P3 Coppermine, and the fastest 0.13um P4 Northwood signifcantly faster than the fastest 0.13 Tualatin.

 

[dexvx]

P3 clock for clock faster than a Northwood is a myth.

The mighty 1.4Ghz/512KB Tualatin-S was about a match for a 2.0A Northwood/400/DDR266. Right now we have 3.0C Northwood/800/HT/DC DDR400.

 

[robcy]

It had a shorter pipe, and much less branch prediction error. In short it does not waste as many cycles as the Williamette/Northwoods. But, the longer pipes allow Intel to scale the core speed much higher, enought to make up the diffrence in efficiency and then some. With other improvements like 200fsb, and HT they have also increased efficiency, but the most important fact is that they are currently 1.8ghz faster than the fastest Tualatin, and that covers a lot of ground.

 

[RanDum72]

P3 clock for clock faster than a Northwood is a myth.

The mighty 1.4Ghz/512KB Tualatin-S was about a match for a 2.0A Northwood/400/DDR266. Right now we have 3.0C Northwood/800/HT/DC DDR400.

Aren't you contradicting yourself here? Yo usaid yourself the 1.4ghz Tualatin 'S' is equal to a Northwood 2.0ghz CPU, which is 600mhz faster in clock speed. So its not a myth at all.

 

[THUGSROOK]

yes, P3 clockspeeds are more inline with AMD PR ratings then P4 clockspeeds.

 

[FishTankX]

p3 Max .13 clockspeed: 1.5GHZ
P4: 3.4GHZish..

Nuff said.

Which is gonna win? A 1.5GHZ p3 or a 3.4GHZ p4?

 

[FishTankX]

A p3 tualatin 133MHZ SDR FSB is very rarley a match, clock for clock, for something like a P4 EE or a P4C with HT, espically in mutltasking enviornments.

 

[mrgoblin]

So then how long can intle keep ramping? Woudlnt they hit a wall soon? Why arent they taking an amd like approach since its obviously working better as we can see with prescott, an early view be as it may.

 

[dexvx]

Originally posted by: RanDum72

P3 clock for clock faster than a Northwood is a myth.

The mighty 1.4Ghz/512KB Tualatin-S was about a match for a 2.0A Northwood/400/DDR266. Right now we have 3.0C Northwood/800/HT/DC DDR400.

Aren't you contradicting yourself here? Yo usaid yourself the 1.4ghz Tualatin 'S' is equal to a Northwood 2.0ghz CPU, which is 600mhz faster in clock speed. So its not a myth at all.

I meant the old arse Northwoods yes, but the new ones, no.

Tomshardware did that huge 68 CPU shootout (from a Pentium-133Mhz), and the higher end, better equiped Northwoods can match it clock for clock.

 

[Amorphus]

Originally posted by: dexvx

Originally posted by: RanDum72

P3 clock for clock faster than a Northwood is a myth.

The mighty 1.4Ghz/512KB Tualatin-S was about a match for a 2.0A Northwood/400/DDR266. Right now we have 3.0C Northwood/800/HT/DC DDR400.

Aren't you contradicting yourself here? Yo usaid yourself the 1.4ghz Tualatin 'S' is equal to a Northwood 2.0ghz CPU, which is 600mhz faster in clock speed. So its not a myth at all.

I meant the old arse Northwoods yes, but the new ones, no.

Tomshardware did that huge 68 CPU shootout (from a Pentium-133Mhz), and the higher end, better equiped Northwoods can match it clock for clock.

even if you increase the quantity, the performance:speed ratio will still remain the same. and we're talking about northwood cores (assuming A or B, but C with its higher FSB would only show gain because of memory).

with equivalent RAM, the P3 would own a P4 at the same speeds, C, EE, or not. however, technology progresses beyond what we have.

 

[FishTankX]

Originally posted by: Amorphus

Originally posted by: dexvx

Originally posted by: RanDum72

P3 clock for clock faster than a Northwood is a myth.

The mighty 1.4Ghz/512KB Tualatin-S was about a match for a 2.0A Northwood/400/DDR266. Right now we have 3.0C Northwood/800/HT/DC DDR400.

Aren't you contradicting yourself here? Yo usaid yourself the 1.4ghz Tualatin 'S' is equal to a Northwood 2.0ghz CPU, which is 600mhz faster in clock speed. So its not a myth at all.

I meant the old arse Northwoods yes, but the new ones, no.

Tomshardware did that huge 68 CPU shootout (from a Pentium-133Mhz), and the higher end, better equiped Northwoods can match it clock for clock.

even if you increase the quantity, the performance:speed ratio will still remain the same. and we're talking about northwood cores (assuming A or B, but C with its higher FSB would only show gain because of memory).

with equivalent RAM, the P3 would own a P4 at the same speeds, C, EE, or not. however, technology progresses beyond what we have.

That is the most rediclous statment I have ever made.

If you took a PIII, gave it EDO RAM, clocked it down and matched it to a pentium I bet the pentium could own the PIII too. That doesn't make it better.

The P4 is built to take advantage of maximum bandwidth and L2 cache. And it has been doing so. You don't design a whole entire CPU around memory bandwidth then just say 'Oh, i'm going to compare it by killing it's memory bandwidth, then see if the more bandwidth conservative CPU owns it.'

PIII versus P4 is irrelevant. The P4 0wnzrz the P3 in almost every regard, except power consmption.

The real debate, would be Pentium-M versus P4. That is a much harder one to answer.