AMD Dual Athlon MP2600+ 23rd March 2003
Before we get started we need to explain the format of our reviews over the past six months. Readers will have noted that an entire review is laid out on one long page instead of 15-20 short pages. While it takes longer for a single page of this length to load we have designed the layout so that text appears first and graphics last so you can start reading before completion of loading. We decided not to split over 15-20 pages as many people find it annoying to have interruptions once they start getting into a review. It also means you can use the search option in the menu bar to go straight to parts that are of interest. We'd like our readers to let us know which method they prefer by providing feedback in our Forums. In any case, by the time you finish reading this paragraph the page should have mostly loaded.
Traditionally most consumers have dismissed the idea of using dual processors as being the domain of servers and high end workstations with the belief that they would only see benefits if they were running more than one application at a time. In recent years we have seen multi-threaded operating systems like Windows XP and Linux become common place in the home and these have the native capability of making the most of multiple CPUs. More and more applications are becoming multi-threaded to make use of more than one CPU and looking to the near future we even see games being written to take advantage of multiprocessing. Imagine playing an RPG where one processor takes care of the needs of the player and his/her immediate surroundings while another CPU keeps the rest of the game world dynamically ticking over. You would be right in thinking that is exactly what happens when you play online on a game server. What if you could run the server part on your PC without slowing down your own gaming experience regardless of how many of your friends logged onto your machine to join in the fun (obviously if hundreds of people log into your machine it will slow down).
There's another secret users of Dual CPU machines have jealously guarded over the years, to understand it we must look at the way modern operating systems run more than one process at a time on a single CPU. They use a method called time slicing where each process gets an allocation of CPU time (how much time it gets depends on priority and other factors). This is fine for system processes and applications that use little resources but if two big applications are run simultaneously that are real-time (we'll be simulating this in our tests later on) a substantial system overhead is incurred that can make the system feel as if it's crawling along. Many readers will no doubt have been watching a video clip or DVD on their PC when their anti-virus software decides it's time for an update or to run a scan and groaned as their system slowed down to a crawl. The usual response to this is to click the "remind me later" button or "not now" button and carry on watching their movie after a minute or so of inconvenience. Users of Dual CPU systems may not even notice this interruption. The downside of this is that often one CPU is sitting around not doing anything although statistically a system with a single CPU spends a lot of time with its CPU idling.
At the time this review is written there are two contenders for the top place of CPUs - the Athlon XP3000+ with its new Barton core and the Pentium4 3.06GHz with its experimental Hyper-Threading technology. This review will investigate a third option, that being a Dual CPU system using two Athlon MP2600+ processors. Both AMD's and Intel's flagship CPUs cost about twice as much as an Athlon MP2600+ which is only slightly slower in terms of PR rating and speed. This means that the systems we are testing cost about the same so the relative performance will allow readers to determine where they get the most "bang for their buck".
The CPUs examined
The MP2600+ uses the Thoroughbred core and a 133MHz FSB. There are no plans to switch this chip over to a faster bus and this is probably the last incarnation we will see the Athlon MP take. It will be replaced in the server segment by the soon to be released Opteron but from a pricing point of view will remain the most cost effective solution for low end servers and workstations for some time to come.
On the left is the old Athlon MP1800+, the only real difference between the CPUs is the die shrink resulting from the thoroughbred core. Cache size and FSB frequency remain the same.
Here's a summary of the system and the reference systems we compared it to:
The above descriptions are accurate with the following exceptions:
The memory used was DDR333 for the Athlon machines and PC1066 for the Pentium4 system. DDR400 was not used as it gave worse results due to memory timing issues.
We would have liked to include a Dual Xeon system in the comparison but we have a strict criteria that all the systems compared must be very similar in price and a Dual Xeon system is much more expensive than the Dual Athlon system.
As well as our usual tests we will be conducting some new ones towards the end of this review to simulate real world conditions were Dual CPUs could be more beneficial to see how much of a difference the extra CPU makes.
Let's start with the synthetic benchmarks. We should remind ourselves that the AMD 760MPX chipset has a double width CPU bus (2 x 266MHz) but only a single width (266MHz) memory bus.
In stark contrast to the other Athlon systems the Dual Athlon MP2600+ CPUs shoot ahead of the competition in this test usually dominated by the very wide FSB of the Pentium4 CPUs.
This is not the case with the 266MHz memory bus which is the weakest of any system we have tested so far.
The double width CPU bus really shines in this test and easily bests all rivals.
While the Whetstone scores are similar to the P4 system, the Dhrystone score is almost twice that of the other systems.
PC Mark 2002
This is a series of tests and is more comprehensive than any of the Sandra benchmarks.
The hard disk scores are not relevant as completely different drives were used. Let's compare these results with other systems.
This test does not take advantage of both CPUs. This gives us an interesting comparison as all 3 Athlons actually run at the same speed. We can see the difference going to a 166MHz FSB makes and also the difference increasing the cache size to 512KB makes.
3D Mark 2001
Before we show you the graphs we'd like you to see the score the MP2600+ gave as it first came up for us.
This is a perfectly respectable score on a Ti4400 so this system is no slouch for playing games on. Let's see how the others did.
In fact only the XP3000+ is significantly out in front here. This is particularly impressive as in the past the perceptions of Dual CPU machines have been that gaming performance has to be sacrificed for more "serious" application performance. This is clearly no the case here and our Dual CPU system shows it can handle any of today's games as well as any single CPU multimedia system.
Unreal Tournament 2003
Now for some real world benchmarks starting with UT2003 Flyby.
There seems to be a limitation with the memory controller on the 760MPX chipset that restricts performance at lower resolutions which is to be expected on this older chipset. We are still waiting to hear VIA's excuse though (the release of the KT400A is probably an admission of not getting it right first time - again). Most gamers prefer to play at 1024x768 and above and at these higher resolutions the Dual CPU machine beats all of the P4 Systems.
The same results are repeated here with an even bigger lead at the highest resolutions.
Let's turn to an area where we know that fast CPUs will make a difference - Audio and Video encoding. This is becoming more and more popular and is very computationally intensive with long processing times (relatively speaking that is, this field is not for those that complain about how long their Outlook Express takes to load).
For consistency we will use Jet Li's The One as our test matter. It is not interlaced and contains a mixture of action types and is not too long. There will be three tests all using Divx 5.02. Audio will be encoded separately. I will try and keep my commentary to a minimum as all configuration information is shown in the images below.
and here are the results
This test shows the SSE2 optimization of Xmpeg and is by no means a real world indication. Among the Divx community Xmpeg is being used less and less in favour of VirtualDub.
AviSynth and VirtualDub
No serious Divx encoder uses Xmpeg alone and it's just used by the media for benchmarking purposes so let's get serious. We ripped our source material to hard disk and created a DVD2AVI project file using forced film (it was 99% film). Loading this into Gordian Knot we first saved an .avs file with no changes at all (720x480) and no filters of any sort. This was loaded into VirtualDub with the following CODEC parameters :
After encoding we got these results:
Raw Divx encoding is largely a function of bandwidth and here we can see the double width CPU bus of the Dual Athlon system keeping it neck and neck with the Pentium4 system. The other systems struggle to keep up.
This is all good stuff but how about a real-world test? To simulate a realistic test we added a neutral bicubic resize filter in the .avs file and used the following CODEC parameters (including two popular Pro settings) which are designed to total 700MB (when the audio is muxed in):
Which resulted in the following.
The addition of computationally heavy filters put less emphasis on the memory bandwidth and more on raw CPU speed, changing the situation dramatically. The Dual MP2600+ is well in the lead due to the extra processing required for the filters and Pro settings making this the best Divx encoding machine we have reviewed so far.
What about audio? We took the AC3 track from the above sample material and used HeadAC3he to convert it into Vorbis format so our final muxed file could have Ogg containment. There isn't space here to go into the advantages of Ogg Vorbis over MP3 and AVI so let's just say that Vorbis sounds about the same as MP3 for half the file size or twice as good for the same file size (that is subjective though).
Since it is more meaningful to show throughput than time taken (which depends on the length of the source) we display the results thus:
Another good result in a test that uses only one CPU (this will change soon as the application's author provides multithreading capability). Only the XP3000+ is faster for audio encoding.
Multi-Processing Test 1
For this test we run Windows Movie Maker and Windows Media Player at the same time.
The workload is multitasked where Window Movie Maker encodes a DV-AVI video to a MPEG-4 Windows Media VIdeo 8 format while Windows Media Player plays a 57 second WMV 8 file. The input video is a 208MB DV-AVI. The encode rate is 768kbps, 320x240, 30fps. Prior to starting the video conversion and benchmarking WMP is started and set to high priority in Task Manager.
As in tests of this kind the shortest bar is the winner and here we can clearly see that the Dual Athlon system is clearly superior. Analysis shows that Hyper-Threading gives the P4 3.06GHz a boost but is nowhere near as good as two actual CPUs.
Multi-Processing Test 2
For this test we ran MusicMatch Jukebox 7.2 and the Novalogic Commanche 4 demo at the same time.
The workload for MusicMatch Jukebox 7.2 encodes into mp3 format a .wav file which is about 600MB. The .wav file contains about 60 minutes of music, has a 1411 kbps bit rate, 16-bit audio sample size and 44KHz audio sample rate. This .wav file is converted into an .mp3 file. To test Commanche 4, the built-in benchmark function was used, set at a screen resolution of 640x480 with 32-bit color depth.
Again the Dual Athlon MP2600+ wins by a substantial margin showing it is best for content creation.
Things have changed so much over the last 2-3 years that a Dual CPU system has become a compelling buy. Prices of Dual Athlon motherboards have come down to the level of single CPU boards (Dual Xeon boards are still relatively much more expensive). A major contributing factor to this review has been the pricing structure both AMD and Intel apply to new CPUs. The cost of an Athlon XP3000+ and Pentium4 3.06GHz is about twice that of an Athlon MP2600+ allowing two of these CPUs to be purchased for the price of either the aforementioned top end CPUs.
In virtually all computationally intensive benchmarks the Dual CPU system came out ahead and even in the gaming and multimedia benchmarks it was amongst the top scorers. Gone are the days when Dual CPU systems were regarded as specialist items used in servers and expensive workstations. Their benefits are now within the reach of the general public and thanks to operating systems like Windows XP application developers are finding it easier than ever to incorporate support for Dual CPUs without undue effort. Big name forthcoming games like Doom3 are set to support multiprocessing and our tests have shown that a Dual CPU system combined with a good graphics card will make a formidable gaming system.
Whether the requirement is for a home workstation for professional use, a budget server or indeed a powerful gaming machine it seems that current CPU pricing models are a gift in building that elusive all-round system which will give the most performance per pound (£) or "bang for buck" as our American cousins like to say.
All trademarks are the property of their respective owners.