Print this article

Intel Pentium 4 2.0GHz


Date
: 08/27/01 – 03:34:05 PM

Author
:

Category
: Processors

Page 1 : Introduction

Manufacturer: Intel
Price: $500

Introduction

When Intel released the Pentium 4 almost a year ago, they planned on gaining back the territory they had previously lost to other manufacturers, mainly AMD. The Pentium 4 was Intel’s new high-end processor and was targeted to people who need powerful machines to do video-editing, gaming, office applications, … in short, the power users. Unfortunately, all of that backfired a little because the benchmarks were showing that the P4, which debuted at 1.4GHz, was getting its butt kicked by the competition. The main problem with the P4 was that it lacked software support, which was crucial to unveil the true power behind the cpu.

Shifting back to the present, the P4 has established itself in the high-end market and is available at the following speeds: 1.3GHz, 1.4GHz, 1.5GHz, 1.6GHz, 1.7GHz and 1.8GHz. Today, two more flavors are released to the public, namely the 1.9GHz and 2.0GHz Pentium 4. Although both cpu’s are equal to their predecessors, with the only exception being the clockspeed, it is still a very interesting product because as of today, the first processor that breaks the 2GHz barrier is available in a store near you! At the same moment we are also looking at the last P4 based upon the ‘Willamette’ core, which means it we’ll be shifting to the 0.13 micron fabrication process soon. In a few weeks/months time, Intel will introduce their new ‘Northwood’ core, which will be using the brand new socket478 package. The package gets smaller, and of course the actual processor will get smaller to, using a 0.13 micron die resulting in less power consumption and reaching higher clock speeds. With the launch of the socket423 P4 2.0GHz, Intel has also released the socket478 P4 2.0GHz. This one does not use the ‘northwood’ core, it is still using the ‘willamette’ core so it is basicly the same processor we are reviewing today, but in a different package.



Enough talk … let’s take a look at Intel’s brand new Pentium 4 cpu running at 2GHz!

Page 2 : Specifications

Specifications – The Pentium 4

Since Intel released the 1.4GHz and the 1.5GHz Pentium 4 back in late summer of 2000, theyhold the crown for the highest clocked processor available on the market. 2With the .0GHz P4, Intel reassures themselves that they will be holding on to their position as manufacturer of the highest clocked processor. Not only that, but they are also the first to crush the 2.0GHz barrier.

As benchmarks have proven in the past, pure speed is not the most essential part of a powerful processor. Just think about the Alpha cpu’s or the G4 processor from Apple. With that in mind, let’s take a look at the Pentium 4 speclist.

- Available at speeds ranging from 1.30 to 2.00 GHz
- Featuring the new Intel NetBurst micro-architecture
- Supported by the Intel 850 chipset
- Fully compatible with existing Intel Architecture-based software
- Internet Streaming SIMD Extensions 2
- Intel MMX media enhancement technology
- Memory cacheability up to 4 GB of addressable memory space and system memory scalability up to 64 GB of physical memory
- Support for uni-processor designs
- Based upon Intels 0.18 micron manufacturing process

The most interesting feature of the Pentium 4 is the NetBurst technology. With this new micro-architecture, Intel brings a truckload of new features including Hyper Pipelined Technology, a 400 MHz System Bus, Execution Trace Cache, and Rapid Execution Engine as well as a number of enhanced features like Advanced Transfer Cache, Advanced Dynamic Execution, Enhanced Floating-point and Multi-media Unit, and Streaming SIMD Extensions 2. It is thanks to these new features that the Pentium 4 is as powerful as it is today. Now I am aware of the fact that 95% of all the terms used in the above paragraph are totally new to you, so let me give you a brief explanation:

Hyper Pipelined Technology:

The hyper-pipelined technology of the NetBurst micro-architecture doubles the pipeline depth compared to the P6 micro-architecture used on todays Pentium III processors. One of the key pipelines, the branch prediction / recovery pipeline, is implemented in 20 stages in the NetBurst micro-architecture, compared to 10 stages in the P6 micro-architecture. This technology significantly increases the performance, frequency, and scalability of the processor. This part was crucial for Intel to be able to reach such high speeds with a 0.18 micron core. Because it is very difficult for a company (without mentioning the costs of an operation like this) to go from a 0.18 micron buidling process to a 0.13 micron process, which would enable them to reach higher core speeds, the engineers over at Intel had to come up with another way to make the cpu run at higher clockspeeds. Like I mentioned before, they doubled the number of stages in the pipeline to a total of 20, which means every ‘action’ is divided in 20 parts instead of 10 … so the processor has to do a small amount of work in each of these 20 parts, whereas before it had to work faster since there were only 10 parts. The problem with this way of working is the following: if the processor makes a mistake at the end of the pipeline, it has to start all over again which means the time gained by the faster clock speed and the higher number of sections, is lost. But don’t worry, a solution was found … and they called it ‘Rapid Execution Engine’. More on that later …

400 MHz System Bus:

The Pentium 4 processor supports Intels highest performance desktop system bus by delivering 3.2 GB of data per second into and out of the processor. This is accomplished through a physical signaling scheme of quad pumping the data transfers over a 100-MHz clocked system bus and a buffering scheme allowing for sustained 400-MHz data transfers. This compares to 1.06 GB/s delivered on the Pentium III processors 133-MHz system bus. This is the reason why we need very fast memory to be able to catch on to the cpu speed. RDRAM was born, but unfortunately it was also very expensive! Intel will soon release the i845 chipset which will give us the possibility to use PC133 memory with the P4. This is a good thing for those who have a lot of PC133 laying around, but my guess is that PC133 is too slow to keep up with the P4 power. DDR memory is the only ‘real’ alternative to RDRAM, and Intel is aware of it. That is why they will be releasing a DDR chipset for the P4 in Q102. VIA is getting ready to release their P4 DDR chipset, called P4X266. Intel and VIA are in some sort of fight because of this because Intel claims VIA has not been licensed to release a P4 chipset, and VIA claims they can since they bought S3 a while back.

Level 1 Execution Trace Cache:

In addition to the 8kB data cache, the Pentium 4 processor includes an Execution Trace Cache that stores up to 12K decoded micro-ops in the order of program execution. This increases performance by removing the decoder from the main execution loop and makes more efficient usage of the cache storage space since instructions that are branched around are not stored. The result is a means to deliver a high volume of instructions to the processors execution units and a reduction in the overall time required to recover from branches that have been mis-predicted.

Rapid Execution Engine:

Two Arithmetic Logic Units (ALUs) on the Pentium 4 processor are clocked at twice the core processor frequency. This allows basic integer instructions such as Add, Subtract, Logical AND, Logical OR, etc. to execute in half a clock cycle. For example, the Rapid Execution Engine on a 1.50 GHz Pentium 4 processor runs at 3 GHz. With our sample, the REE runs at 4GHz! Talking about
raw power ;).

256KB, Level 2 Advanced Transfer Cache:

The Level 2 Advanced Transfer Cache (ATC) is 256KB in size and delivers a much higher data throughput channel between the Level 2 cache and the processor core. The Advanced Transfer Cache consists of a 256-bit (32-byte) interface that transfers data on each core clock. As a result, the Pentium 4 processor 1.50 GHz can deliver a data transfer rate of 48 GB/s. This compares to a transfer rate of 16 GB/s on the Pentium III processor at 1 GHz. Features of the ATC include:

- Non-Blocking, full speed, on-die Level 2 cache
- 8-way set associativity
- 256-bit data bus to the level 2 cache
- Data clocked into and out of the cache every clock cycle

Advanced Dynamic Execution:

The Advanced Dynamic Execution engine is a very deep, out-of-order speculative execution engine that keeps the execution units executing instructions. The Pentium 4 processor can also view 126 instructions in flight and handle up to 48 loads and 24 stores in the pipeline. It also includes an enhanced branch prediction algorithm that has the net effect of reducing the number of branch mis-predictions by about 33% over the P6 generation processors branch prediction capability. It does this by implementing a 4KB branch target buffer that stores more detail on the history of past branches, as well as by implementing a more advanced branch prediction algorithm.

Enhanced Floating-Point and Multimedia Unit:

The Pentium 4 processor expands the floating-point registers to a full 128-bit and adds an additional register for data movement which improves performance on both floating-point and multimedia applications.

Internet Streaming SIMD Extensions 2 (SSE2):

With the introduction of SSE2, the NetBurst micro-architecture now extends the SIMD (Single Instruction Multiple Data) capabilities that MMX technology and SSE technology delivered by adding 144 new instructions. These instructions include 128-bit SIMD integer arithmetic and 128-bit SIMD double-precision floating-point operations. These new instructions reduce the overall number of instructions required to execute a particular program task and as a result can contribute to an overall performance increase. They accelerate a broad range of applications, including video, speech, and image, photo processing, encryption, financial, engineering and scientific applications. SSE2 is one of the most important features of the Pentium 4, as these extra instructions will give the processor a huge performance increase. The sad part is that the SSE2 instructions need software support. This the main reason why the P4 gets kicked by the AMD processors: current software doesn’t offer SSE2 support … AMD’s future cpu’s will also have SSE2 support, which means that the standard will be set and soon all software will be SSE2 minded.

Features Used for Test and Performance / Thermal Monitoring:

- Built-in Self Test (BIST) provides single stuck-at fault coverage of the microcode and large logic arrays, as well as testing of the instruction cache, data cache, Translation Lookaside Buffers (TLBs), and ROMs.
- IEEE 1149.1 Standard Test Access Port and Boundary Scan mechanism enables testing of the Pentium 4 processor and system connections through a standard interface.
- Internal performance counters can be used for performance monitoring and event counting.
- Includes a new Thermal Monitor feature that allows motherboards to be cost effectively designed to expected application power usages rather than theoretical maximums.

Page 3 : 850 Chipset

Specifications – The i850 Chipset

A good processor is worth nothing if it is missing a good chipset to guide all of its raw power in a good way. Intel’s history shows they know how to build a good chipset (just think of the Intel BX chipset that outperformed chipsets built over a year later), but they also know how to build a not so good chipset I’m sure you remember the i820 fiasco). Fortunate for all of us, the chipset that goes along with the P4, the i850, is actually quite good. The only downside on the i850 is that it needs RDRAM (RamBus) to operate, and that kind of memory is still very expensive. To tell the truth, RDRAM has been a chain on Intel’s legs, because the audience is staying away from the P4 because they don’t want to pay the premium for the RDRAM. Like I said earlier, the i845 chipset is on its way and will offer the possiblity to use normal PC133 memory with a Pentium 4. But the real solution lays with DDR, but because of a deal Intel made with RamBus, manufacturer of RDRAM, Intel can not sell a DDR chipset until 2002. VIA is aware of this as well, and they jumped the boat by manufacturing the P4X266 chipset, that does support DDR memory along with the Intel Pentium 4. This ended up in a catfight between the lawyers of both Intel and VIA … Intel claims VIA can not build and sell this chipset whereas VIA spreads the word they can … I’m sure we’ll hear more from this as soon as a decision has been made.



At the present moment, the only available chipset available is the i850 one from Intel, so lets see what it can do for you:

- 400 MHz system bus delivers a high bandwidth connection between the Intel Pentium 4 processor and the platform. To ensure maximum performance, the system bus is balanced with the dual RDRAM channels at 3.2 GB/s, providing 3x the bandwidth of platforms based on Intel Pentium III processors.
- Intel Hub Architecture with optional P64H provides increased I/O bus bandwidth to allow better concurrency for media-rich applications and multitasking
- Dual RDRAM memory channels deliver 3.2 GB/s of memory bandwidth to the processor. High memory bandwidth, coupled with an efficient protocol, deliver balanced platform support and provide the memory bandwidth necessary to extract full performance from the Intel Pentium 4 processor.
- The AGP4X interface allows graphics controllers to access main memory at over 1 GB/s, twice that of previous AGP platforms.
- With RDRAM memory and the new Intel Pentium 4 processor, AGP4X delivers the next level of 3D graphics performance.
- Two USB controllers double the bandwidth available for USB peripherals to 24Mbps over four ports. This results in a significant increase over previous integrated 1-4 port hubs at 12Mbps.
- The latest AC97 audio delivers six channels of audio for enhanced sound quality and full surround sound capability for live broadcast and other Digital Dashboard programming.
- LAN Connect Interface (LCI) provides flexible network solutions such as home phone line, 10/100 Mbps Ethernet, and 10/100 Mbps Ethernet with LAN manageability.
- Dual Ultra ATA-100 controllers support the fastest IDE interface for transfers to storage devices. Additional performance is gained with Intels Storage Driver over standard ATA drivers.
- Communication and Network Riser (CNR) offers flexibility in system configuration with a baseline feature set that can be upgraded with an audio card, modem card, or network card.

Again we see the 400MHz system bus emerge, supplying the P4 with a memory bandwidth of 3.2GB per second. Shifting to PC133 memory will kill this advantage and with that, take down a lot of the P4 power. Although all of that is based on a theory, I think the benchmarks will prove me right in the near future … Just like most modern chipsets, the i850 offers support for AGP 4x, USB and onboard sound based upon the AC97. Of course onboard LAN is available as well. Last but not least, the i850 chipset carries support for the UDMA100 standard.



Lets get this show on the road …

Page 4 : Installation

Installation

When you look at a P4, you don’t get that scary feeling you get when holding an AMD processor that is about to be installed …

We all know how fragile Athlons and Durons are, especially when we install heatsinks onto them. How many of you can say that they never crushed, killed and destroyed one of these prescious babies? I sure can’t …



The Pentium III featured a pretty robust core already, and with the Pentium 4 Intel added a metallic heatspreader on top of the core to help dispate the heat. This makes the core almost impossible to destroy, and the only way to actually kill a P4 with your bare hands would be by clipping of the pins at the bottom of the cpu. Not too bad huh!

I already mentioned that Intel has installed a heatspreader onto the core, to improve heat transfer. This would make you believe that the P4 running at 2GHz emits a lot of heat (remember that an Athlon at 1.4GHz produces around 80W) but nothing is less true! Actually, a 2GHz Pentium 4 is rated at around 68W … which means the heatsink will only get warm to the touch. O used the Intel heatsink to do all tests, and let me tell you these heatsinks are big. The good part is that this lets them use a whisper fan that is almost totally silent! I loved it!

Installing a P4 isn’t the hardest thing in the world … pull the lever, slide in the cpu, close the lever. Installing the heatsink can be a bit harder, but I don’t expect the average computer user to have problems with it. Intel uses a new heatsink mouting system, which makes sure that the heatsink is levelled perfectly onto the core. What they do is preinstall the retention mechanism onto the motherboard. The only thing you have to do is put the heatsink onto the core, close two levers and you’re all done! How sweet can it be? No more hassles with screwdrivers (with the risk that you slip and kill the mobo).

The new socket478 package uses an identical heatsink mouting mechanism, but you won’t be able to use heatsinks for the socket423 on socket478 motherboards.

Page 5 : Overclocking

Overclocking

Of course we couldn’t write a review about this powerchip without having a closer look at the overclocking potential behind it.

However, note that if you are looking for the ideal overclocking processor, Intel is not (yet anyway) way to go. AMD makes it a lot easier to overclock as they leave the multiplier unlocked whereas Intel hardlocks it. This leaves you with the front side bus only … and overclocking by the use of the FSB means you are also overclocking the memory and other devices.

In order to get the most out of this cpu, we increased the core voltage to the highest setting available on our testboard, the EPoX 4T2A which we reviewed recently. After that we gradually increased the cpu speed.

After trying, testing and torturing, we managed to get our cpu rockstable at 2.2GHz but we ran it at 2.1GHz in all tests because the heatsink was getting a bit too hot at 2.2GHz, causing instability after running prime95 for over 2 hours. With a bigger heatsink, 2.2GHz would have been totally stable for sure!

Our sample booted at 2.3GHz as well, but while loading Windows 2000, it rebooted itself everytime …

So we ended up with a 10% overclock … not bad but not world shaking either … Lets take a look at some benchmarks to see how the P4 will perform in real life situations shall we?

Page 6 : Benchmarking

Benchmarking

In order to get a good view on the real power of Intel’s new ‘king’, we have thrown the cpu in front of the lions and measured how well it held up against the competition. Our benchmarks are testing the cpu on several categories: raw cpu power, 3D gaming power, rendering, integer calculations, floating point calculations, memory tests and so on and so forth.

Before we look at the numbers, I want to point out once more that the P4 is not yet able to show its full power right now because a lot of the force comes from software support. That support is starting to arrive, but it is far from ideal. With this said, lets take a look at the figures shall we?

We compared the 2.0GHz to most of his smaller brothers, but we also included AMD’s most powerful processor in the tests to see how it would hold up against the P4.

The test setups:

P4 System

- Intel P4 2.0GHz socket423 processor
- EPoX 4T2A motherboard / Intel 850 chipset
- 2x64MB PC800 RDRAM memory
- VisionTek GeForce3 video card
- Western Digital 20GB 7200rpm UDMA100 hdd
- Accton 100mbit NIC
- Enermax 431W psu
- Windows 2000 + SP2 + Detonator 12.90 drivers + latest Intel 850 drivers

Athlon System

- AMD Athlon 1.4GHz socketA processor
- EPoX 8K7A+ motherboard / AMD761 northbridge & VIA 686B southbridge
- 1x128MB Crucial PC2100 DDR memory
- VisionTek GeForce3 video card
- Western Digital 20GB 7200rpm UDMA100 hdd
- Accton 100mbit NIC
- Enermax 431W psu
- Windows 2000 + SP2 + Detonator 12.90 drivers + VIA 4.32 drivers + latest AGP Miniport drivers

Both systems are using the lastest state-of-the-art components to make sure we get a realistic view on today’s power cpu’s. It would be out of the questions to test a 2GHz cpu with a Riva TNT or something like that heh ;). Each system has completed a prime95 torture test round of over 48 hours so both systems were rockstable as well.

Now, as Caesar would have said: ‘Let the games begin …’

Sisoft Sandra Benchmark Suite

Over the years, Sisoft has established themselves as one of the most used benchmark suites. This is quite understandable since they offer a benchmark for each computer component out there. We only used their processor and memory benchmarks today, since we are concentrating on the processors and not the other components.

CPU Benchmark

CPU Multimedia Benchmark

Memory Benchmark

If we look at the CPU benchmarks, it is obvious that the P4 is kicking serious butt. This cpu is pulling the sweetest results we have seen so far, but note that the Athlon is performing very well too. The 1.4GHz Athlon positions itself right above the P4 1.5GHz.

The multimedia benchmark shows a pattern that can be compared to the above … Again the 2.0GHz P4 pulls away at high speed, but AMD is right on its tail with the 1.4GHz Athlon. Nevertheless, the fastest cpu currently available is the Pentium 4. It has become clear already that AMD and Intel use different ways to power their processors, as the Intel is depending more on the raw processor speed whereas AMD clearly uses another way to achieve these outstanding results.

Now how will the battle continue in the gaming environment? Lets take a look …

Page 7 : 3D Mark & Quake 3

3D Mark 2001 & Quake3

MadOnion’s 3D Mark 2001, followed by ID Soft’s Quake3 are the best benchmarks currently available to measure the gaming performance of a system. We let both systems run all 3D Mark 2001 tests and after those were done, we used both the Demo001 benchmark and the Torture demo in Quake3 to see how both cpu’s would hold up in a real life game, because although 3D Mark 2001 is sweet, it is still a benchmark. Of course these benchmarks also depend a lot on the video card, which is why we used the fastest GPU currently available: the GeForce3.

MadOnion 3D Mark 2001

Quake 3

Intel is still the king when it comes to Quake 3. Back in the days when the PIII was Intel’s powerhouse, they had an advantage over AMD when it came to games like Quake 3. Today, that tradition hasn’t changed and the P4 kicks the Athlon hard.

3D Mark is another story … because the Athlon gains distance on the P4 and even surpasses it at high resolutions! Not bad for a chip that is clocked 600MHz lower heh ;).

It’s not over yet peeps, we still have some tricks in our sleeves … just wait and see :).

Page 8 : ZDnet & RC5

ZDnet & RC5

To end all the benchmarking in our bunker, we ran some cpu intensive applications on both systems. We decided to use ZDnet CPU Mark and also the ZDnet FPU Mark because they give us a good view on the pure cpu force. Although both benchmarks have been around for quite some time, they still proove to be very useful in today’s test environments!

As always, we also threw some RC5 action into the game, since a lot of you are in a distributed computing team (like the O RC5 team for example). When it comes to programs like RC5 (or SETI), the processor has to take a package and decode it as fast as he can. Easy? Sounds like it yes … but there is more to it.

ZDnet CPU Mark

ZDnet FPU Mark

Distributed.net RC5 Client

In case you wonder why the Pentium 4 is scoring so bad in RC5, let me enlighten you. RC5 has several different ‘core selections’ which are used to make optimal use of the processor. Unfortunately, the RC5 client does not know the P4 core yet, so it tries to pick another one that resembles it. This means you’ll get a huge performance drop, but this should be fixed as soon as there is a new client available.

The FPU tests also indicated that the 1.4GHz Athlon can beat the P4 at 2GHz, but it is no secret that AMD has impleneted a fantastic FPU unit into their processors, and it is the FPU part that handles these kind of calculations …

Page 9 : Conclusion

Conclusion

This concludes our look at Intel’s new Pentium 4 processor, clocked to an amazing 2.0GHz or 2.000MHz (sounds impressive huh).

The Pentium 4 is still very young, and the technology behind it is just starting to get used in today applications. This is the main reason why this processor has had a bad start. The second reason why most people decide against buying a P4 is because they have to buy the ultra expensive RDRAM as well in order to get a working system. With Intel releasing the 845 ‘Brookdale’ chipset anyday, the P4 will be able to work along with PC133 memory. Although this will damper the performace, it will make the cpu available to a wider audience. VIA’s P4X266 chipset that brings DDR memory to the P4, will also give the Pentium 4 a huge boost because DDR has been accepted as the new standard. Intel is aware of this as they are releasing a DDR chipset in Q102 as well …
Last but not least, Intel will be implementing a huge price cut in the P4 market, which means this powerful processor will become very affordable!

Would I buy the P4 2.0GHz for my personal system? Yes, but I would suggest that you go for the socket478 version and not for the socket423 version because Intel’s future cpu’s will be using the socket478 package as well, leaving you with better upgrade options as time goes by. With the announced price cuts, you can get yourself the most powerful cpu currently available, and with that also the only one that breaks the 2.0GHz barrier.

Something sure was proven in this article: raw cpu speed is not the only thing that makes a processor fast or slow anymore …

Like the benchmarks show, the Athlon 1.4GHz is just as fast as Intel’s P4 2.0GHz. Sometimes the P4 takes the advantage, sometimes the AMD takes the lead … so it comes down to what you, the end user needs. Both processors have good and bad parts, it just depends on your personal opinion which one wins the battle.

Good

- 2.0GHz!
- Low heat emission.
- Q3 power!

Bad

- Future technology so performance is not at full force yet.
- More expensive than the competition
- Socket423 version will be outdated very soon, so go for the socket478 one!.
- At the moment, RDRAM is necesarry.