Logo   Banner   TopRight
TopUnder
Transparent
Video/DVD Performance ShootOut VideoMark 2000 Results
Transparent
Date: Dec 15, 2001
Section:Graphics/Sound
Author: HH Editor
Transparent
Video/DVD Performance ShootOut VideoMark 2000 Results - Page 1

Dueling DVD 
New Video2000 Results From The Majors In Graphics

The Online Hardware Review Media community has a solid focus on 3D Graphics performance and technology as a whole.  This goes without saying.  Too often the Digital Video Subsystem of a particular graphics card goes untested and evaluated.  Perhaps this is because there are still relatively few benchmark suites or utilities available to the masses.

MadOnion... By the way, what the heck is the deal with the name of this company?  " I am a VERY angry Onion and therefore I am going to benchmark my booty off!"  I mean really, someone in marketing was sniffing the super glue a little too much on their lunch break, to come up with this one!  :-)  But I digress...

MadOnion, known the world over for the comprehensive and technically complete 3D Graphics benchmark 3DMark 2000, has delivered yet another quality benchmark package, this time stepping up to the challenge of Digital Video Benchmarking. This new digital video benchmarking suite is called Video2000.   

The following is a quick look at the relative performance, features and quality of the digital video playback and MPEG encode/decoding of some of the top graphics accelerators in the market.  We tested  ATI's Rage Fury 128 Pro , Elsa's Erazor X2 DDR GeForce, Matrox's Millennium G400 and 3dfx's Voodoo3-3500TV.  All extremely competent 3D cards in their own respect.  Let's see how they stack up in a Video2000 Round-Up.


 

First,  let's take a quick look at what Video2000 from MadOnion tests within its benchmark suite.  This article is not intended to be a comprehensive overview of the MadOnion software itself but rather a first hand look at how each of the noted accelerators perform under these conditions.  However, we feel it is important for you to understand this testing platform so that you can make sense of the numbers produced rather than arbitrary assumptions.

The Video2000 Benchmark categorizes "performance" in three areas, performance, quality and features.

  • The performance tests evaluate a graphics card's abilities with respect to BLT Operations, Memory Transfers and Software DVD Playback.  Here is a look at what some of the BLT tests look like.  This one is for BLT downscaling.  Click the image for a closer look.

  • This is a screen shot from one of the Software DVD Decoder Performance Tests.  Click image to get the 800X600 shot.

  • The quality tests evaluate the cards ability to reproduce digital video images with respect to scaling, color space conversion, and de-interlacing or tearing in Software DVD playback.  Here is a shot of what one of these tests looks like.  Again, click this image for full view.

  • Finally, the feature test reports on the various digital video reproduction features of your accelerator like overlay color control, shrinking and stretching etc...
     

  • The overall Video2000 Video Mark score is the sum of the results of the quality, performance and feature-set testing you performed.  A higher score is better.

 

Now that you know how this test is set up and a little bit about what the numbers mean, it is time to give you the dirt on how our contenders measured up! 

 
  

 Video2000 Marks - This way! 

 

 
Transparent
Video/DVD Performance ShootOut VideoMark 2000 Results - Page 2

Dueling DVD 
New Video2000 Results From The Majors In Graphics

Hot Hardware's Test System
A top end system used with a level playing field.

Full Tower ATX Case  w/ 300W PS, Pentium3 -500E Overclocked to 750MHz., Tyan Trinity 400 S1854 Motherboard, 128MB of PC133 HSDRAM, WD 18G 7200RPM DMA66 Hard Drive, Kenwood 72X CDROM, Plextor 8432 CD-RW Drive, 56K Modem,  Sound Blaster Live Audio Card, Windows 98SE, DirectX 7.0a, 

NVidia Reference Drivers 3.68, 3dfx V3500 Drivers 1.04, 
Matrox Powerdesk Drivers 5.50,  ATI Rage Fury 128 Pro Drivers 6.31

 

Benchmarks / Comparison - The Battle On The Silver Screen 
Head To Head Match-Ups Between 3dfx, ATI, Matrox and NVidia 

It should be noted that during our test, the system was extremely stable with any of the graphics cards installed.  The tests ran without incident and we ran all tests twice to check for consistency.  The tests utilized whatever standard DVD and MPEG decoders and encoders were installed with the respective drivers.  Cinemaster DVD was the default decoder that most cards installed with the exception of the Voodoo3-3500TV, which uses one form Ligos. We also ran a test with the Xing DVD Player on the GeForce card, just to show you the difference in performance with DVD Decoders.  This is an area you need to be careful of when making a comparison between Video 2000 Video Mark scores.  Different Software Decoders will produce different results.  We are looking into this issue further as well.

 

 

This chart, generated by Video 2000, gives you a total aggregate score in graphical form much the same as in 3DMark.  However, as you can see, the bars are broken down into the relative contribution to the total score.  This gives you a good picture of how all the competitors stack up. 

Video 2000 Video Mark Total Scores 

 

 

My apologies to you folks that are still running your desktop at 800X600 resolution or lower.  The following spreadsheet generated by the Video 2000 tests, contains a lot of data.  Even with most of the individual details not shown, this image is larger than 800X600.  However, the file size is relatively small with this gif. 

Video 2000 - Video Mark Comparison Details

I would bet most of you die-hard NVidia fans were very surprised when you saw that the GeForce DDR got beaten by the ATI Rage Fury 128 Pro.  Here you can see that not only does the Rage Pro beat all the other cards in the overall Video 2000 score but it also has the lower CPU utilization.  This means you won't need a P3-733 to play DVDs with the ATI card.  Most of the other cards here, especially the G400 and with the possible exception of the V3-3500TV, should also handle well with mid range systems.  I was very surprised that the V3 had such high CPU utilization.  Perhaps this was due to software decoder performance issues with the Voodoo compatible decoder.

As you can see, each board used its own "default decoder".  This is the Software DVD decoder that is installed with the drivers for each board.  MadOnion also tests each board against its own "reference decoder".  The performance of the reference Video 2000 decoder was lower across the board with higher CPU utilization, as one might expect.  

Performance Vrs. Quality

The final score matters

In raw performance tests, the GeForce barely edged out the ATI Rage but I would speculate that in a lower end system, this would not be the case, as the GeForce CPU utilization is higher.  Matrox was the leader in raw performance but I would expect that gap to close, with a lower end CPU, due to their slightly higher CPU utilization as well.  Also of note, is the fact that the ATI and 3dfx cards do not use AGP transfers for Digital Video processing memory requirements.  Digital Video does not need the bandwidth associated with AGP transfers like 3D Graphic Acceleration does.

Finally, an important aspect of this test is the "Features" category.  Here as you will see, the raw performance leader Matrox, also has the least number of on board features.  This has an effect on performance of DVD playback which is indeed faster but not as good in the area of quality. This, you can see in the Quality Score for the G400, which is lower than both the GeForce and the Rage Fury Pro.  This is a surprising turn of events for the G400, which is known to have exceptional 3D image quality.

For all intents and purposes, the ATI card has the highest quality and the most features of this round-up along with top end performance in line with the others.  It seems, at least from these tests, that the Rage Fury 128 Pro has the best overall performance which is why its overall Video Mark is the highest.

 

 Thanks and "Props"

We would like to thank the good folks at MadOnion for their support and the creation of yet another great benchmarking product.  It makes our jobs that much easier here at HotHardware.com.

 

-Davo

 

 

To Hot Hardware!

 


Content Property of HotHardware.com