Introduction
Since the release of Pentium 4 a couple of weeks ago, we have seen a large number of different opinions about Intel’s new flagship processor. Due to its completely new design it shows quite peculiar behavior in what I would call ‘bread-and-butter applications’. The majority of standard benchmarks attest Pentium 4 inferior to average performance, which generated a lot of disappointment. However, we were also trying to point out that the scores with specially Pentium4-optimized software are able to show this processor in a much better light.
Confusion?
This unusual situation generated a lot of confusion, because it makes it very hard to come to a valid conclusion about Pentium 4. Should this CPU be simply dismissed as an overpriced and inadequate solution, or shall we already now call it the big winner of tomorrow’s software environment? Are reviewers that despise Pentium 4 simply trapped in the past or are the supporters of Pentium 4 just a bunch of wishful thinkers?
Another Update …
This promised and somewhat delayed Pentium 4 evaluation update will not be able to answer all of the questions raised about Pentium 4, but it will try bringing new facts to the table, so that we all get a little bit of a better idea what we can expect from Intel’s latest and greatest processor product. It will again compare Pentium 4 with AMD’s latest Athlon processor, which will soon be available at speeds of 1200 MHz and beyond, running at 133 MHz (266 MHz DDR) processor bus and aided by the highly anticipated DDR-SDRAM memory. We are basing this update purely on benchmarks, which is why I recommend reading our previous publications about Pentium 4 to learn about its new architecture and the benchmark results that we’ve been seeing so far. Please make sure you have read all of the listed articles below to be able and see the big picture:
- Intel’s New Pentium 4 Processor
- Important Pentium 4 Evaluation Update
- Painting a New Picture of Pentium 4 – Tweaked MPEG4 Encoding
- Tom’s Blurb: Pentium 4 – Another Recount?
Windows 2000
This new comparison will focus on benchmarks scored under Microsoft’s successor operating system to Windows NT, which is commonly known under the name Windows 2000. For people who work professionally with their PCs, Windows 2000 is offering the reliable platform that neither Windows 98 nor the new Windows ME are able to provide. I personally don’t run Windows 98/ME on any of my systems anymore, simply because I am tired of regular system crashes and other failures. Windows 2000 has quickly become very successful, because it finally combined the reliability of Windows NT with the ease of use and gaming support known from Windows 98. Additionally, Microsoft has finally added a few very useful features that are common in UNIX for a very long time.
The only reason that keeps end users from running Windows 2000 on their systems is mainly based on its higher price compared to Windows 98/ME and the fact that people don’t realize how much more reliable their system will become once Windows 2000 is installed. There’s also the fear that Windows 2000 could be more difficult to administer than the two ‘toy OSes’, which is not really based on actual facts.
Intel’s Strength In The Workstation Market
PCs that are used for professional applications are commonly referred to as ‘workstations’. While there is a whole lot of different software that runs on all the different workstations, a large amount of those systems is used for professional 3D-design. We were very surprised to learn that 99.9% of those professional graphic as well as other workstations are actually equipped with Intel processors. The OEMs that supply those systems are simply not interested in AMD’s Athlon, although its performance makes it a great product for exactly this market. AMD has great difficulty to penetrate this segment where the system cost is rather negligible to the customers. For workstation buyers, Athlon’s price advantage over Intel-products is of no importance, while Intel’s former good name for stability and reliability and AMD’s past failures seems to still sit very deep inside the decision makers minds.
You can imagine that this fact will ensure that the future PC-workstation processor will be Pentium 4 rather than Athlon, even though it doesn’t seem to make any sense. AMD has still a lot work to do if it wants to get into this market segment.
Benchmark Setup
Let’s now get into the Windows 2000 benchmark results, which are always preceded by the benchmark setup.
We still used the same Pentium 4 platform as in the earlier tests, the P4T motherboard from Asus. Due to the high clock potential of Pentium 4 we decided to include numbers generated by Pentium 4 at the overclocked speeds of 1.6 and 1.73 GHz once more. Pentium III ran on the Asus CUSL2 i815 motherboard, as i815 is currently the fastest chipset available for this processor.
Athlon was mainly tested in its new version at 133/266 MHz system bus and with DDR-memory, running in an AMD760 platform from Gigabyte. Gigabyte’s GA-7DX motherboard has lately been a hot topic in the news, because it is the motherboard that is supposed to provide the platform of Micron’s latest Athlon/DDR platform. Micron has put systems with AMD760 and 133 MHz FSB on shipment hold, due to a problem between AMD’s 760 chipset and Athlon processors with 133 MHz bus. While AMD is working on a fix for this problem, it is not supplying any 133 MHz chipsets to motherboard makers or motherboard makers are unable to get AMD760-boards at 133 MHz FSB through their quality testing. This is why it is close to impossible to get your hands on any AMD760 motherboard right now. AMD’s problem with 760 is a bit nebulous, but it reminds you of Intel’s situation with i820 a year ago.
Surprisingly enough we did not encounter any problems with Gigabyte’s GA-7DX motherboard. It proved to be a better as well as more reliable performer than AMD’s own 760-reference board. Different to AMD’s reference platform the GA-7DX from Gigabyte is using the 686B southbridge from VIA. This makes it very easy to find an IDE-driver for Windows 2000 that enables ATA-100.
You might be surprised to find Athlon at clock speeds of 1400 and 1466 MHz in the comparisons. I included them to give Athlon the chance to show its performance in the same clock speed area that Pentium 4 is using. You may wonder how I got Athlon running at those high clock speeds, but you will have to wait a few days for the answer, as I will dedicate a special article to this very special achievement and the hardware involved.
Neither AMD’s 760-reference board, nor Gigabyte’s GA-7DX provide any means to alter the multiplier of Athlon processors. Therefore I had to do some modifications to the GA-7DX to be able and overclock Athlon to those 1400 (133 MHz x 10.5) and 1466 MHz (133 MHz x 11). In this process I learned how easy those modifications actually are and I will soon supply you with an article that explains how to easily change each SocketA-motherboard to an overclocker’s board.
Benchmark Setup, Continued
Hardware Setup | |
I850 Socket423 Pentium 4 Platform | ASUS P4T, BIOS 1001-performance BIOS |
Rambus Memory | 2 128 MB Samsung PC800 RDRAM RIMMS |
SDRAM Socket A platform for AMD Athlon and Duron Processors | ASUS A7V, BIOS 1004D final |
SDRAM Socket 370 platform for Intel Pentium III and Celeron processors | ASUS CUSL2, BIOS 1004.003 |
SDRAM Memory | 128 MB Wichmann Workx PC133 SDRAM CL2, setting 2-2-2-5/7 |
DDR Socket A platform for AMD Athlon processors at 133 MHz Front Side Bus | Gigabyte GA-7DX Rev.2.3, BIOS Rev. |
DDR Memory | 256 MB Infineon , 8-8-5-2-2-2-3 |
Overclocking Aid | Special Super Cooling Solution |
Hard Drive for Windows 2000 Tests | IBM DPTA-305020 ATA66 IDE, 20 GB, NTFS |
Graphics card | NVIDIA Geforce 2 GTS 64 MB Reference Card Core Clock 200MHz, Memory Clock 333 MHz, Driver 6.47 |
Network Card | Netgear FA310TX REV-D2 |
Software Setup | |
Windows Version | Windows 2000 Professional, Service Pack 1 |
Windows Resolution | 1024x768x16x85 |
Quake 3 Arena | Retail VersionSetting Normal, 640x480x16 bit color, no sound |
DirectX Version | 8.0 |
FlasK Settings | Video Codec: DivX 3.11 alpha, Fast-Motion, keyframe every 10 seconds, compression 100, data rate 910 kbpsAudio Codec: audio not processedVideo Resolution: 720×480, 29.97 fps, interlacedResizing: Nearest Neighbor |
Webmark2001 | MSIE 5.01Timer Resolution 10 msJava enabled, 5.0JVM 5.0.0.3310JavaScript 1.3Adobe Acrobat Reader 4.05Cult3D 5.0.1.43Flash 5.0.30.0Microsoft Agent 2.00.0.3422RealPlayer 6.0.9.357Quicktime 4.1.1NetMeeting 3.01Windows Media Player 6.4.09.1109Windows Media Services 4.1.00.3918 |
FlasK MPEG Once More
I’d like to start by picking up the topic of the last P4-Update article. You remember that Intel supplied us with a special version of the free video compression software FlasK MPEG of Alberto Vigatб that included optimizations for Pentium 4. Intel had done those optimizations to FlasK in only one night and mainly with compiler-work. With the optimized code Pentium 4 was able to leave Athlon behind, once the software was using the new SSE2-enhancements of Pentium 4. However, even the newly compiled x87-IDCT performed better on Pentium 4 than on Athlon, which was very surprising. Usually nothing can touch Athlon’s floating-point performance, especially if the common x87 FPU is used.
Of course the comparison between Pentium 4 and Athlon, which is using software that is only optimized by Intel is not able to provide a fair picture of the two processors. That’s why I had asked AMD to provide me with a FlasK version that is optimized for Athlon. Alexander Goodrich and Sean Stanek, who are both software engineers that are involved with AMD-optimizations, had already started the whole issue shortly after the publication of my last P4-article. The code that I finally ended up using comes from the two, but it has been looked over by AMD. I was told that AMD would try to provide me with an even more optimized version this week, but that Alex’s and Sean’s FlasK-optimizations were already too good to be improved by AMD’s engineers in a short time. I will publish results with AMD’s final FlasK-version as it becomes available, but for the time being we should thank Alex and Sean for their great achievement of quickly providing an Athlon-optimized version of FlasK that AMD was not able to enhance any further within the last week.
Looking at the new picture shows that P4 1.5 GHz is still in the lead when compared to Athlon 1200/133 plus DDR-memory. While the performance of Athlon 1200 and P4 1500 is close to identical once the optimized x87-executable is used, Pentium 4 using its new SSE2-enhancements is way faster than Athlon using 3DNow!. This is rather surprising, since most of us would probably have expected Athlon pulling away from Pentium 4.
Comparing Pentium 4 and Athlon on a clock-for-clock basis (Pentium 1.5 GHz vs. Athlon 1.466 GHz) shows however, that Athlon runs x87 FPU optimized code still faster than Pentium 4. What this comparison also shows is that with FlasK SSE2 is clearly superior to 3DNow!. Please be reminded that this example only focuses on FlasK. There might well be other software that would run a lot better on Athlon than on Pentium 4 once the code is optimized for each.
Bottom line is that Pentium 4 is indeed able to deliver excellent performance once software has been optimized for it. FlasK is an example showing that Athlon cannot always beat Pentium 4, even when the software has special Athlon-optimizations. My estimate of Athlon’s inability to beat Pentium 4 in this test is that Athlon runs into bandwidth limitations with this streaming benchmark. In case of MPEG4-encoding Pentium 4’s fast quad-pumped 100 MHz bus plus the dual-Rambus channel memory access of i850 seems superior to Athlon’s dual-pumped 133 MHz bus and the 133 MHz DDR-SDRAM memory solution of AMD760. That’s why even the best optimizations can’t give Athlon enough of a boost to overtake Pentium 4.
BAPCo Sysmark 2000 Under Windows 2000 Professional
Sysmark 2000 from BAPco is slowly ageing a bit, but it still represents a good benchmark for current office application performance. We know that Athlon leaves Pentium 4 far behind in Sysmark2000 under Windows 98 and so it is not very likely that it will look too different under Windows 2000.
It is true, Athlon is beating Pentium 4, but the difference is not quite as huge as it was under Windows 98. The overclocked Athlon at 1400/133 and 1466/133 shows how high Sysmark2000 scores can actually go. It leaves everything else in the dust. Pentium 4 can’t beat Athlon 1200/133 plus DDR even when it is overclocked to 1728 MHz. It is obvious that office applications are not able to take any advantage of Pentium 4’s architecture right now. It is questionable to me how those applications could benefit from Pentium 4 optimization, which is why I doubt that the above picture will change a lot in the future.
Quake 3 Arena Under Windows 2000
Pentium 4 was able to beat Athlon quite well in Quake 3 under Windows 98. This is mainly due to the fact that Quake 3 is extremely dependent on memory bandwidth, which you can easily see when you benchmark your own system with Q3A and make minor changes to your memory timings. Let’s see if there’s any difference under Windows 2000.
Surprisingly enough, Athlon scores a lot better under Windows 2000, while Pentium 4 stays where it was at under Windows 98. Athlon is still not able to beat Pentium 4, even when it is overclocked to 1466/133 MHz. You can see that there is only a tiny difference in frame rate between Athlon 1200/133, 1400/133 and 1466/133. I consider this behavior to be the result of a clear bandwidth problem. I doubt that even an Athlon at 1.6 GHz on AMD760 could reach the results of Pentium 4 at 1.5 GHz. Athlon is suffering from either memory or bus bandwidth restrictions and Pentium 4 can take advantage of its high data bandwidths once more.
BAPCo Webmark 2001
BAPCo supplied us with their latest new benchmark by the name of ‘Webmark2001‘. This benchmark is supposed to test the ‘Internet-performance’ of a system via a simulation of a web-session on a client system.
It is quite a bit of work to set-up the testing environment, because the benchmark requires that the to-be-tested client system gets hooked up to a dedicated web-server via a network connection.
The workload of WebMark includes a combination of Internet technologies such as:
- Adobe Acrobat Reader
- Apple Quicktime
- Cycore Cult3D
- Java
- JavaScript
- Macromedia Flash
- Microsoft Agent, ASP, Windows Media Player, VML
- RealNetworks RealAudio and RealVideo
- SSL
- XML
You start the benchmark from within Microsoft’s Internet Explorer and then it connects to different sites simulated by the web server mentioned above.
BAPCo Webmark 2001, Continued
WebMark2001 incorporates the following website scenarios:
- Business-to-Business Website Scenarios:
- eHouseBuilder.com: A web marketplace for building supply materials and interactive tools for house builders.
- electronics-designer.com: The online arm of an electronics design company that allows other engineers to acquire detailed information about the integration of electronics-designer.com’s products.
- eMedInsure.com: A medical insurance company that receives insurance claims from hospitals and then places the medical data online for fast approval of those claims.
- eCommodity-Traders.com: An online brokerage house that assists the trader in trading with other agencies by providing trading services and information utilities like charts and prediction models.
- eHouseBuilder.com: A web marketplace for building supply materials and interactive tools for house builders.
- SuperEtailer.com: A shopping site that offers consumers the power of shopping online combined with the same interaction they get from visiting a store.
- My-Foyer.com: An online news and data center that provides a quick summary of the user’s favorite topics, as well as links to interactive and media-rich sites which provide the user with more detailed information.
- Auto-Concepts.com: A corporate intranet website at an automobile company that enables online creation, sharing, and publishing of documents using standard office applications.
The benchmark runs for about an hour and finally presents its result on a web page as well. It looks very fancy and seems to have been a huge development project. However, I still have to get a real feel for it to understand what is actually happening in this new benchmark.
You also have the option to do a so-called ‘technology run’ with this benchmark, which supplies you with results like Flash Animation Performance, XML Performance, Java Performance, Encoding Performance and even Download Performance. However, so far the benchmark is too young as that I would rely on this technology run for processor performance evaluation yet. I will work with it for a while and see if it will make it into our standard benchmarking suite.
BAPCo Webmark 2001, Continued
Here are the overall results:
Pentium 4 beats Athlon by quite a long shot. Only in a clock-for-clock comparison of Pentium 4 1.5 GHz and Athlon 1.466 GHz Athlon can reach the same scores as Pentium 4. Let’s get into a bit more detail:
In business-to-business, Athlon is still way behind Pentium 4. Even the overclocked Athlon at 1466 MHz cannot reach Pentium 4 scores.
BAPCo Webmark 2001, Continued
For some reason Athlon beats Pentium 4 in the business-to-consumer part of Webmark 2001.
However, in the intranet-business simulation ‘Auto Concepts’ Athlon is extremely far behind Pentium 4.
I can already hear some of you say that BAPCo is an Intel-paid operation, which is why this benchmark favors Pentium 4. This is wrong however. BAPCo is a non-profit organization with a whole lot of members. It is true that Intel and Dell are the two members that might be able to make the Webmark2001-results look a bit sketchy, but BAPCo expressed to me that those two companies are not trying to influence BAPCo (who is indeed using office space in Intel’s Santa Claran headquarter). One thing is a matter of fact though. AMD is for some strange reason NOT a member of BAPCo.
I personally don’t quite think that this benchmark was ‘tuned’ to make Pentium 4 look good. However, it seems pretty obvious that the software used in this benchmark is not particularly optimized for Athlon at all. This software is what we use all day on the web however, regardless if we despise the missing Athlon-optimizations or not. Adobe Acrobat, Quicktime, Internet Explorer, Macromedia Flash, … are absolutely common applications. Therefore I would say that we have to respect the high scores of Pentium 4, while keeping in mind that Athlon is clearly put at some kind of unfair disadvantage by common Internet software developers. It is interesting to see this, because so far nobody was ever aware of this fact. I don’t know if we can blame the software makers of the above-mentioned titles, or if AMD has failed to work with those companies closely enough. It would certainly be advisable for AMD to join BAPCo.
Bottom Line
This time I presented you four different benchmarks and in three of them Pentium 4 came out victorious. Basing a whole article on those four benchmarks would automatically have to lead to a favorable review of Pentium 4.
This is not my only Pentium 4 article however and those are not the only benchmarks we ran. Therefore the outcome still remains the same. Pentium 4 has got clear flaws, but it’s not a piece of crap either. It performs well in 3D-games, it is able to do very fast video-encoding and it seems to give you some kind of edge using the Internet, boosting Intel’s amusing ‘NetBurst’-hype.
Pentium 4 is still much more expensive than AMD’s Athlon, it only runs with the expensive RDRAM-memory and it requires a new kind of power supply and case, thus blowing up system costs even more. In office applications and several other areas Athlon is able to beat Pentium 4 and that at significantly lower costs. Then there is the issue that by this time next year a new model will replace the current Pentium 4, introducing a different socket (Socket478), which destroys a reasonable upgrade path.
There is enough that speaks against Pentium 4 for people who are reasonable. Once again I have to express my feel about Pentium 4 as some kind of life-style symbol rather than a must-have. However, AMD has started to run into trouble as well. While I am being bold enough supplying you with benchmark data of Athlon 1200/133 on an AMD760-platform, you are utterly unable to get your hands on one. It is currently unknown how long it will take until AMD760 motherboards will become available and the alternative ALi Magik1 platforms have so far shown significant lower performance than AMD760.
If you have dedicated areas in which you want your computer to perform particularly well, if those areas should be 3D gaming, video encoding or other bandwidth intensive software and if you should not shy away from high system costs and the missing upgrade path, you should indeed consider Pentium 4. However, if you want a balanced system with excellent performance at a good price I heavily suggest Athlon, even though you might have to wait until the high-end AMD760/DDR platforms will become available.