Pentium is Intel's #1 (It's All About the Pentiums)

[ta152h]

[citation][nom]pepperman[/nom]The first gen of PIII's (Katmai core) were the exact same as the PIIs (except with the added SSE implementation), however during the die shrink to the Coppermine core, several improvements were made, including full-speed on-die L2 cache, a reworking of the instruction pipelines, among others. These (in addition to the faster bus speed-133MHz vs. 100 MHz) were enough to increase the clock-per-clock performance in relation to the PIIs, and allowed them to remain competitive in the early days of the PIVs.[/citation]

Now you're comparing the Coppermine to the Katmai. The original Pentium III was identical except for SSE.

The Katmai also could use the 133 MHz bus - it was not just for the Coppermine.

The core of the Coppermine was identical - the improvements were to the L2 cache. They didn't change the ALU pipelines at all. The improvement was pretty dramatic though.

The Coppermine was pretty competitive to the Pentium 4 when it came out, but keep in mind, the Pentium 4 had plenty of room to grow on that process, whereas the Coppermine maxed out at 1.1 GHz (the 1.13 GHz failed badly because the L2 cache could not keep up) on .18. The Pentium 4 hit 2.0 GHz on the same process. I have a Tualatin running at 1.6 GHz, with no issues, but officially it stopped at 1.4 GHz. The .13 based Northwood finished up at 3.4 GHz. Clearly, the Pentium III was not a competitive technology, until enhanced with the Pentium M line, and later the Core 2 line. The clock speeds gave it no chance.

 

[radiowars]

Pentiums are nice.
Actually, they were nice. Now it's all about the i's.

 

[pepperman]

[citation][nom]ta152h[/nom]Now you're comparing the Coppermine to the Katmai. The original Pentium III was identical except for SSE. [/citation]

Yes, but I was also comparing P3 to P2, and perhaps more accurately than you originally did.

[citation] The core of the Coppermine was identical - the improvements were to the L2 cache. They didn't change the ALU pipelines at all. [/citation]

"Intel also re-worked the chip internally, and finally fixed the well known instruction pipeline stalls." Though this didn't necessarily improve performance, the instruction pipelines were reworked.

[citation]
The improvement was pretty dramatic though.The Coppermine was pretty competitive to the Pentium 4 when it came out, but keep in mind, the Pentium 4 had plenty of room to grow on that process, whereas the Coppermine maxed out at 1.1 GHz (the 1.13 GHz failed badly because the L2 cache could not keep up) on .18. The Pentium 4 hit 2.0 GHz on the same process. I have a Tualatin running at 1.6 GHz, with no issues, but officially it stopped at 1.4 GHz. The .13 based Northwood finished up at 3.4 GHz. Clearly, the Pentium III was not a competitive technology, until enhanced with the Pentium M line, and later the Core 2 line. The clock speeds gave it no chance.[/citation]

The Williamette core P4s didn't even soundly beat the 1 GHz P3 until the P4 reached 1.6 GHz. In addition to the low performance of the early P4 in relation to the P3, the P4 also carried a price premium; not just for the cpu, but also the RDRAM. This DID allow the P3 to remain competitive, until the SDRAM support was added to the P4's chipsets.

Intel discontinued the P3's because it was older technology, and their capitol was invested in the P4. The fact that they took a step backwards with the early P4s is still a fact. We also see that the Core and Core 2 architectures are based on the P3 lineage, which alludes that perhaps Netburst as a whole was a step backwards.

 

[ta152h]

[citation][nom]pepperman[/nom]Yes, but I was also comparing P3 to P2, and perhaps more accurately than you originally did.[citation] The core of the Coppermine was identical - the improvements were to the L2 cache. They didn't change the ALU pipelines at all. [/citation]"Intel also re-worked the chip internally, and finally fixed the well known instruction pipeline stalls." Though this didn't necessarily improve performance, the instruction pipelines were reworked.[citation] The improvement was pretty dramatic though.The Coppermine was pretty competitive to the Pentium 4 when it came out, but keep in mind, the Pentium 4 had plenty of room to grow on that process, whereas the Coppermine maxed out at 1.1 GHz (the 1.13 GHz failed badly because the L2 cache could not keep up) on .18. The Pentium 4 hit 2.0 GHz on the same process. I have a Tualatin running at 1.6 GHz, with no issues, but officially it stopped at 1.4 GHz. The .13 based Northwood finished up at 3.4 GHz. Clearly, the Pentium III was not a competitive technology, until enhanced with the Pentium M line, and later the Core 2 line. The clock speeds gave it no chance.[/citation]The Williamette core P4s didn't even soundly beat the 1 GHz P3 until the P4 reached 1.6 GHz. In addition to the low performance of the early P4 in relation to the P3, the P4 also carried a price premium; not just for the cpu, but also the RDRAM. This DID allow the P3 to remain competitive, until the SDRAM support was added to the P4's chipsets.Intel discontinued the P3's because it was older technology, and their capitol was invested in the P4. The fact that they took a step backwards with the early P4s is still a fact. We also see that the Core and Core 2 architectures are based on the P3 lineage, which alludes that perhaps Netburst as a whole was a step backwards.[/citation]

Where is your citation from?

You seem to be confused. The 1.5 GHz Pentium 4, which was initially released, beat the 1 GHz Pentium III easily. You can find those benchmarks pretty easily on this website. Also, the Pentium 4 had SSE2, which would show later benefits as x87 became unnecessary because of it.

SDRAM was not competitive for the Pentium 4. It was extremely poorly suited and crippled the performance of the processor. RDRAM was the memory of choice, although DDR worked OK. Oddly, by the time they moved to DDR, the cost of RDRAM had come down so much it was competitive.

If you're saying the Pentium III was competitive in the market place, that's fine. It was not competitive in performance. The Pentium 4 design was much faster on the same fabrication process. Remember, Pentium III maxed at 1.1 GHz, Pentium 4 at 2.0 GHz at .18, and 1.4 GHz, and 3.4 GHz on .13. The Pentium 4 design was much faster.

Intel did not discontinue the Pentium III line. Where do you get your constant misinformation from? They used it for the Pentium M, which became the Core line.

Netburst was a superior technology taken out of the context of power and heat issues. When it met with these realities, it sucked. The later cores could probably hit 10 GHz with proper cooling on 45nm, but would never make a product you could sell. They weren't limited by the speed the transistors could switch at, but by the massive heat and energy delivery issues. When they were being designed, these had not been hindrances to clock speed before, and I guess Intel somehow figured they'd find a way around it. Strange, but true.

I'm not criticizing the Pentium III architecture. It was very good, but the original Pentium III had serious clock speed issues, and bandwidth issues that made it, from the perspective of performance, completely out of the league of the Athlon and Pentium 4. When they added additional transistors to the different stages for greater clock speed, and increased the memory bandwidth, the performance of the processor became very competitive, and many were wondering for years before the Core 2, why Intel would not sell the Pentium M in a desktop form.

I would still love to see the Pentium 4 on 45nm, which is much lower power than 65nm. I think for single core/single thread performance, with exotic cooling, it would beat other x86 processors pretty handily, especially if you add in enhancements they normally do for each node change.

I think a Celeron was benched over 8 GHz on 65nm. 45nm would go much higher. It sure would be interesting to see, but we never will.

 

[hemelskonijn]

Though your tech talk seems correct for the most part it still seems like you are unable to read in context.

My original post was not about the tech it was about the names they gave the tech of several generations. If pentium is 5th then the pentium 2 should have been named sexium or whatever.

Using google before you post what you believe to be the truth is confirming your knowledge so at least you wont sound stupid close minded and or ignorant.

Also you seem to be the only one that actually liked netburst.

 

[pepperman]

"Intel also re-worked the chip internally, and finally fixed the well known instruction pipeline stalls." Citation (though may not be word for word): http://www.search.com/reference/Pentium_III

You seem to be confused. The 1.5 GHz Pentium 4, which was initially released, beat the 1 GHz Pentium III easily. You can find those benchmarks pretty easily on this website. Also, the Pentium 4 had SSE2, which would show later benefits as x87 became unnecessary because of it.

There's a difference between beating in some apps and beating soundly (which I clearly stated in my previous post). True, the P4 @ 1.5 beat the P3 @ 1.0 in games (likely due to the SSE2 instructions), but look at any other benchmark based on computations and you'll see the P3 beating the P4.

SDRAM helped bring the P4 into the mainstream; I agree, it wasn't a good match (due to the high bandwidth requirements of the P4).

The P3 was competitive in performance when the P4s were first released; there's a difference between clock speed and IPCs that you don't seem to understand: the P3 was competitive because it performed better at lower clock speeds.

In terms of the 130 nm P3 maxing @ 1.4GHz, I'm sure if Intel cared they would have gotten it higher (look at Pentium M), so that particular process isn't a valid comparison. I do agree that P4 scaled better on the 180 nm process, however.

The P3 is discontinued, and has been for nearly 7 years. The cpus that are based on it are not called P3s. Are you sure you understand what how a cpu naming scheme works? Is the original Pentium line still in production because Core i7 can trace its roots to it?

 

[lotri]

Lol at the vid. Regarding Pentiums, I'm sure Core2 processors will pass them up soon enough. Not only are they (mostly) dual-core, they also are cheaper than the recently released Core-i3-i5 with only a slight drop in performance in most applications. Not to mention that people keep their computers for an insanely long time.

I think I had my old Pentium III for at least 10 years before it got replaced by another Pentium (way before Core2 came out). That computer's still chugging away and should last a while (low usage = longevity).

 

[neiroatopelcc]

[citation][nom]Article[/nom]With the steady rise and replacement of the Core brand, however, Pentium will likely be on the way out starting in 2011[/citation]
I don't think so. Like with the core 2 family intel will provide the newer core based pentium along the road (one already being shipped). It's natural that pentiums are their #1 as most any cheap office or school computer [running intel] comes with a cheap pentium dualcore these days. They're not good, but it's got the right brand and it's cheap. It would be foolish for intel to discontinue the pentium brand.

 

[saran008]

[citation][nom]aford10[/nom]This time next year, the results will be very different.[/citation]
+1^ aford

 

[saran008]

[citation][nom]kelemvor4[/nom]I'm surprised. I assumed they had stopped manufacturing Pentium (and even core1) processors already. Obviously I was mistaken![/citation]

Hmm me too!

 

[neiroatopelcc]

By Pentium they don't only mean netburst and older architectures, but the branding pentium being used on processors. Like xeon that's a branding that intel is now slapping onto anything they intend for a specific marked regardles of the technology powering it.

 

[anamaniac]

There has been some interesting history n the comments of the article.
However, this still makes me curious if we had continued netburst properly (RDRAM, not SDRAM etc.) and were it would be today.
=)

 

[poorgeek]

Pentium CPUs are legendary. I'm using my "state of the past art" Pentium 4 Northwood to type this comment. I actually have two P4 systems right now, but only one is active. The inactive system uses a P4 2.4GHz non-HT Northwood on an Asus P4PE board with the Intel 845PE chipset. The P4 CPU is a 400Mhz FSB chip. My "advanced" Geforce 3 still worked with that system. I was going to install Windows 7 on it, but decided to retire that computer.

The P4 2.8GHz I'm using right now is on an Asus P4C800-E with the Intel 875P chipset. This P4 system is using 800MHz FSB with 2GB DDR1 RAM and a Geforce 7600GS card. I am using Windows 7 32-bit Home Premium on two WD raptor drives in RAID 0. Except for video playback and encoding/decoding, this single core P4 runs faster than my 2GHz Merom core 2 dual on a Win XP Pro SP3 notebook computer.

Previous Intel CPU systems that I have owned included a "state of the art" 8088 CPU at 4.77MHz with 8087 math co-processor and advanced 640KB RAM, an Intel Pentium 120 MHz system with 16MB RAM (I have only the CPU now), and a Pentium III 733 Coppermine with a Geforce 2 card.

I recently built a new computer that includes an Intel i5-750 at 3.6GHz and 4GB DDR3 RAM. The most memorable CPU, however, is still the good old Pentium. I hope the P4 will remain a little longer.

 

[ta152h]

[citation][nom]hemelskonijn[/nom]Though your tech talk seems correct for the most part it still seems like you are unable to read in context.My original post was not about the tech it was about the names they gave the tech of several generations. If pentium is 5th then the pentium 2 should have been named sexium or whatever.Using google before you post what you believe to be the truth is confirming your knowledge so at least you wont sound stupid close minded and or ignorant.Also you seem to be the only one that actually liked netburst.[/citation]

I think the fact you're a moron is interfering with your understanding. The Pentium Pro was the first of the Pentium II/III generation, not the Pentium II. Also, they built up name equity with Pentium, and probably didn't think using a different name made a lot of sense - it might not have been recognizable.

 

[ta152h]

[citation][nom]pepperman[/nom]"Intel also re-worked the chip internally, and finally fixed the well known instruction pipeline stalls." Citation (though may not be word for word): http://www.search.com/reference/Pentium_IIIYou seem to be confused. The 1.5 GHz Pentium 4, which was initially released, beat the 1 GHz Pentium III easily. You can find those benchmarks pretty easily on this website. Also, the Pentium 4 had SSE2, which would show later benefits as x87 became unnecessary because of it.There's a difference between beating in some apps and beating soundly (which I clearly stated in my previous post). True, the P4 @ 1.5 beat the P3 @ 1.0 in games (likely due to the SSE2 instructions), but look at any other benchmark based on computations and you'll see the P3 beating the P4.SDRAM helped bring the P4 into the mainstream; I agree, it wasn't a good match (due to the high bandwidth requirements of the P4).The P3 was competitive in performance when the P4s were first released; there's a difference between clock speed and IPCs that you don't seem to understand: the P3 was competitive because it performed better at lower clock speeds.In terms of the 130 nm P3 maxing @ 1.4GHz, I'm sure if Intel cared they would have gotten it higher (look at Pentium M), so that particular process isn't a valid comparison. I do agree that P4 scaled better on the 180 nm process, however.The P3 is discontinued, and has been for nearly 7 years. The cpus that are based on it are not called P3s. Are you sure you understand what how a cpu naming scheme works? Is the original Pentium line still in production because Core i7 can trace its roots to it?[/citation]

OK, I guess you're not very technical, and that's the problem. Your simple way of looking at things only goes by the name they use?

Discontinued would mean the technology was not upgraded or used again. I mean, that's what most people would use it for. Because any other way would mean EVERY technology is discontinued. The Pentium 4 tech was discontinued, the Pentium III line was not.

The IPC for the Pentium III did not make up for the fact it was clocked so much lower. In many instances, it ran slower clock for clock, and only ran slightly faster overall. The big problem was the memory.

Your remarks about SSE2 are uninformed. No games using SSE2 were available when it was released. It beat the Pentium III in games badly because of memory bandwidth. The Pentium III was seriously crippled in this regard.

The Pentium M was an enhanced Pentium III, and they added transistors to increase clock speed in the generations succeeding the original. The Coppermine could NOT get past 1.1 GHz, and the Tualatin got to around 1.6 GHz (I've tried five different, Celeron and Pentium III-S, and they get stuck around 1.6). That's pretty much the max. Pentium 4 blew past them, 2.0, and 3.4 were released, both could be overclocked higher.

And no, I was never fan of Netburst, and actually never even owned one or would. It still doesn't mean I like disinformation though. I also thought the design was very interesting and advanced, but obviously failed where the rubber met the road.

 

[ta152h]

[citation][nom]anamaniac[/nom]There has been some interesting history n the comments of the article.However, this still makes me curious if we had continued netburst properly (RDRAM, not SDRAM etc.) and were it would be today.=)[/citation]

I think the move to 45nm would have been huge for Netburst, because power is what held it back, and 45nm was much better for power use. They'd have to do some changes though. The damn decoder was way underpowered, and if the instructions were not in the trace cache, it basically acted like a scalar processor. This happened a lot too, somewhere around 30-40% of the time are the numbers I saw. They either needed to increase the I-Cache or at least add a simple decoder.

I honestly don't believe many people would want it. But, I'd love to see it just from a purely academic perspective. For maybe dual cores, you'd have a killer, but you'd never be able to go to quads and be competitive because of the power and heat.

 

[ta152h]

[citation][nom]poorgeek[/nom]Pentium CPUs are legendary. I'm using my "state of the past art" Pentium 4 Northwood to type this comment. I actually have two P4 systems right now, but only one is active. The inactive system uses a P4 2.4GHz non-HT Northwood on an Asus P4PE board with the Intel 845PE chipset. The P4 CPU is a 400Mhz FSB chip. My "advanced" Geforce 3 still worked with that system. I was going to install Windows 7 on it, but decided to retire that computer.The P4 2.8GHz I'm using right now is on an Asus P4C800-E with the Intel 875P chipset. This P4 system is using 800MHz FSB with 2GB DDR1 RAM and a Geforce 7600GS card. I am using Windows 7 32-bit Home Premium on two WD raptor drives in RAID 0. Except for video playback and encoding/decoding, this single core P4 runs faster than my 2GHz Merom core 2 dual on a Win XP Pro SP3 notebook computer.Previous Intel CPU systems that I have owned included a "state of the art" 8088 CPU at 4.77MHz with 8087 math co-processor and advanced 640KB RAM, an Intel Pentium 120 MHz system with 16MB RAM (I have only the CPU now), and a Pentium III 733 Coppermine with a Geforce 2 card.I recently built a new computer that includes an Intel i5-750 at 3.6GHz and 4GB DDR3 RAM. The most memorable CPU, however, is still the good old Pentium. I hope the P4 will remain a little longer.[/citation]

Actually, the 8088 was never the state of the art, and 4.77 was a clock speed IBM chose so they didn't need to use an additional crystal to synchonize with a TV. The 8086 was the best Intel processor at the time, but IBM chose a stripper model to save costs. The 68000 was a much more powerful processor from Motorola also available at the time, and was used extensively by IBM for add-on cards (like the S/370), but they chose the 8088 and have cursed us with this instruction set since. In their defense, they only thought they'd sell 10,000 machines total, so didn't put a lot of thought to the future into it.

 

[poorgeek]

The 8088 CPU was state of the art back in the early 1980's and, no, I don't think Motorola's 6800 series was any better at that time. The 4.77MHz had nothing to do with a TV. The 8088 and 8087 processors were designed by Intel. IBM used them in their PCs because of cost.

The P4 is a great design. I hope the current Core processors are as reliable as the P4.

 

[pepperman]

[citation][nom]ta152h[/nom]OK, I guess you're not very technical, and that's the problem. Your simple way of looking at things only goes by the name they use? Discontinued would mean the technology was not upgraded or used again. I mean, that's what most people would use it for. Because any other way would mean EVERY technology is discontinued. The Pentium 4 tech was discontinued, the Pentium III line was not. The IPC for the Pentium III did not make up for the fact it was clocked so much lower. In many instances, it ran slower clock for clock, and only ran slightly faster overall. The big problem was the memory. Your remarks about SSE2 are uninformed. No games using SSE2 were available when it was released. It beat the Pentium III in games badly because of memory bandwidth. The Pentium III was seriously crippled in this regard.The Pentium M was an enhanced Pentium III, and they added transistors to increase clock speed in the generations succeeding the original. The Coppermine could NOT get past 1.1 GHz, and the Tualatin got to around 1.6 GHz (I've tried five different, Celeron and Pentium III-S, and they get stuck around 1.6). That's pretty much the max. Pentium 4 blew past them, 2.0, and 3.4 were released, both could be overclocked higher. And no, I was never fan of Netburst, and actually never even owned one or would. It still doesn't mean I like disinformation though. I also thought the design was very interesting and advanced, but obviously failed where the rubber met the road.[/citation]

So according to your definition nothing is ever discontinued if it is adapted into something else that lives on. So the Intel 4004 (the first microprocessor) is not discontinued, right? And in that sense, Hyperthreading was ported to both Atom and the i7, so the P4s aren't discontinued either. Interesting logic...

The P3 @ 1.0 beat the P4 @ 1.5 in nearly every office and\or productivity software when the P4 was released (since 1.0 is less that 1.5 and the 1.0 beat the 1.5, I'd say the 1.0 performed more IPCs). Just look at Tom's initial review of the P4.

If you don't like "disinformation" why do you practice it?

 

[anamaniac]

People hating on me just because I want to know were netburst could possibly be today? =(
My i7 also runs hotter than my Pentium D did (even with a far superior cooler)...

 

[jfem]

According to the report, Intel projects that Pentium-branded processors will account for roughly 42 to 43 percent of Intel’s desktop chips volume in 2010.

This is surprising.

 

[tuannguyen]

Test

 

[Mousemonkey]

Test

Failed

 

[pepperman]

[citation][nom]tuannguyen[/nom]Test[/citation]

Test to lock the thread?