iXBT Labs - Computer Hardware in Detail

Platform

Video

Multimedia

Mobile

Other

Gainward Powepack FX Ultra/1600 GS CoolFX
(NVIDIA GeForce FX 5900 Ultra) Video Card Review






CONTENTS

  1. Video card's features
  2. Testbed configuration, test tools, 2D quality 
  3. Test results: Quake3 ARENA
  4. Test results: Serious Sam: The Second Encounter
  5. Test results: Return to Castle Wolfenstein (speed and quality)
  6. Test results: Code Creatures DEMO
  7. Test results: Unreal Tournament 2003 (speed and quality)
  8. Test results: Unreal II: The Awakening (speed and quality)
  9. Test results: RightMark 3D
  10. Conclusion

I can't express how tired I am reviewing the increasing number of video cards that glut the market. Earlier High-End products were released more seldom, we didn't have to hurry when testing cards and had time for careful examination of every solution including its modifications... What's now? 

Over a year ago ATI Technologies launched its RADEON 9700 PRO which was very popular and sated the market with the powerful DirectX 9.tool. Further incarnations like RADEON 9800 PRO, RADEON 9800 weren't that popular as they were just overclocked versions. The R360 which is soon to come will be just an accelerated version of the RADEON 9800 PRO too.

NVIDIA failed to take the lead in the DX9 niche. Its NV30 was too expensive and had nothing to beat the RADEON 9700 PRO with. Moreover, its 128bit bus worsened its look. The new revision named NV35 lifted the speed but could it grasp the palm of supremacy? 

Not at all. Although it had higher clock speeds, the rendering architecture was closer to 2000-2002 rather than to 2003: 4 pipelines and 8 texture units. 8 pipelines could be formed in case of the Z buffer and when shadows are processed, but it didn't help much. However, it's not what matters most of all. Its higher clock speed could have helped it, and the guys at nVIDIA are able to take the maximum from hardware. 

The worst problem of the NV35 is a low shader speed. I mentioned it earlier that the smaller number of ALUs (which are slower at that) is a real headache of the company. The latest tests based on the shaders 1.1 and 2.0  prove it.

Theoretical materials and reviews of video cards which concern functional properties of the GPU NVIDIA GeForce FX

But we could shut our eyes to this problem because the NV35 is efficient enough for today's and upcoming games. But it's too expensive. I think the marketers at NVIDIA should work harder and make the management to cut prices.

NVIDIA's latest products are much discussed over the Net and even blamed for the downsides mentioned, but we know that their lab is currently working on the mysterious Detonator 50.XX where the performance must be considerably lifted. Today we will discuss it.

But the center of our attention today will be a video card outfitted with a water-cooling system. I don't think many would want a card priced at 800 euros. That is why the Gainward Powepack FX Ultra/1600 GS CoolFX will be reviewed just because it's interesting to take a look at such a unique product as this is the first video card coming with a water cooler.

Gainward is a widely known company. Unfortunately, it lost a bit of our local market because of too high prices set for its top products.

So, we have a big box in front of us filled up with various stuff. I invited the founder of www.3dmark.ru - Ivan Pritula, who helped me assemble it.

Card

Gainward Powepack FX Ultra/1600 GS CoolFX



 
AGP x8/x4/x2 interface, 256 MB DDR SDRAM in 16 chips on both PCB sides. 

Gainward Powepack FX Ultra/1600 GS CoolFX
Hynix 2.2ns memory chips; it corresponds to 454 (908) MHz, but the memory work at 425 (850) MHz, the GPU runs at 450 MHz. 256bit memory interface. The proprietary utility lifts the clock speeds up to 500/900 MHz (according to the Golden Sample law).



 
Comparison with the reference design, front view
Gainward Powepack FX Ultra/1600 GS CoolFX Reference card NVIDIA GeForce FX 5900 Ultra 256MB










 
Comparison with the reference design, back view
Gainward Powepack FX Ultra/1600 GS CoolFX Reference card NVIDIA GeForce FX 5900 Ultra 256MB










 

The card is based on the reference design. The chip has a special protection frame. Here is the chip:




The card supports VIVO via the Philips 7108:




Now let's examine the water cooling system but before I want you to look at the whole accessory pack:




Surely, no water cooler can do without a pump:


 

Here is the water container:




These two parts form a single unit:




You should cut off of the plastic pipe a part you need in your case:




Water is cooled with the following heatsink:




and a big 120mm fan which comes with a protective grill:




There's also special anticorrosive liquid which must be mixed with distilled water, and fasteners for pipes and other parts:


 

The cooler itself looks like a closed container which has circulating cooling liquid inside and covers the core and memory chips. Above you can see two sockets for drawing in and out water.


 

So, let's start building it up. First of all, Ivan assembled the heatsink with the fan:




Then we looked for the place to get up the system and fasten its parts:




Here is how it was growing up:


 

After that we measured off and cut off the pipes:


 

The system is ready. It works, the pump drives water, and the heatsink is a just little warm. But the test results will be given a bit later. 

The card ships in a gigantic retail box:
 

Gainward Powepack FX Ultra/1600 GS CoolFX 
The box is white and has urethane foam inside. Here you can also find drivers and utilities, software for VIVO, DVI-to-d-Sub adapters, VIVO splitter and a strange cup for a 20pin power supply connector of a mainboard (we couldn't understand what it was for).






 

Testbed and drivers

Testbed: 

  • Pentium 4 3200 MHz based computer:
    • Intel Pentium 4 3200 MHz CPU;
    • DFI LANParty Pro875 (i875P) mainboard; 
    • 1024 MB DDR SDRAM; 
    • Seagate Barracuda IV 40GB HDD; 
    • Windows XP SP1; DirectX 9.0a;
    • ViewSonic P810 (21") and ViewSonic P817 (21") monitors.
    • NVIDIA drivers v51.75.

VSync off, S3TC off in applications. 

The cards used for comparison:

  • ATI RADEON 9800 PRO (380/350 (700) MHz, 256 MB DDR-II, driver 6.378).

Test results

Before we start examining 2D quality, I should say there are no complete techniques for objective 2D quality estimation because: 

  1. 2D quality much depends on certain samples for almost all modern 3D accelerators; 
  2. Besides videocards, 2D quality depends on monitors and cables; 
  3. Moreover, certain monitors might not work properly with certain video cards. 

With the ViewSonic P817 monitor and BNC Bargo cable the card showed excellent quality at the following resolutions and clock speeds:

Gainward Powepack FX Ultra/1600 GS CoolFX  1600x1200x85Hz, 1280x1024x120Hz, 1024x768x160Hz
 

Test results: performance comparison

Conventional signs: ANISO 8xP - Anisotropic 8x Performance (earlier it was called Balanced), ANISO 8xQ - Anisotropic 8x Quality, ANISO 16xQ - Anisotropic 16x Quality. 

Test applications: 

  • Return to Castle Wolfenstein (MultiPlayer) (id Software/Activision) - OpenGL, multitexturing, ixbt0703-demo, test settings - maximum, S3TC OFF, the configurations can be downloaded from here

  •  
  • Serious Sam: The Second Encounter v.1.05 (Croteam/GodGames) - OpenGL, multitexturing, ixbt0703 demo, test settings: quality, S3TC OFF

  •  
  • Quake3 Arena v.1.17 (id Software/Activision) - OpenGL, multitexturing, ixbt0703 demo, test settings - maximum: detailing level - High, texture detailing level - #4, S3TC OFF, smoothness of curves is much increased through variables r_subdivisions "1" and r_lodCurveError "30000" (at default r_lodCurveError is 250 !), the configurations can be downloaded from here

  •  
  • Unreal Tournament 2003 v.2225 (Digital Extreme/Epic Games) - Direct3D, Vertex Shaders, Hardware T&L, Dot3, cube texturing, default quality

  •  
  • Code Creatures Benchmark Pro (CodeCult) - the game that demonstrates card's operation in DirectX 8.1, Shaders, HW T&L.

  •  
  • Unreal II: The Awakening (Legend Ent./Epic Games) - Direct3D, Vertex Shaders, Hardware T&L, Dot3, cube texturing, default quality

  •  
  • RightMark 3D v.0.4 (one of the test scenes) - DirectX 8.1, Dot3, cube texturing, shadow buffers, vertex and pixel shaders (1.1, 1.4).

If you want to get the demo benchmarks that we use let me know by e-mail. 

In the review of the RADEON 9800 PRO UE I drew your attention to how the trilinear filtering was simplified in some tests. First of all, it was in the UT2003. Today we will carry out the tests with the new drivers from ATI and NVIDIA.

The card reached 530/1000 MHz with the water cooling only! That is why the card was also tested at these clock speeds (in comparison with the overclocked RADEON 9800 PRO).

In the usual modes the card worked at the default frequency of 450/850 MHz. 

Quake3 Arena

 













The new driver helped the 5900 Ultra strengthen its leading position. No complaints about the quality (though it can be skilfully hidden).

Serious Sam: The Second Encounter

 













Just the same. 

Return to Castle Wolfenstein (Multiplayer)

 













The figures are good, the 5900Ultra is ahead, but here is what we noticed:
 

RADEON 9800




FX 5900




A lot of objects are light-struck though the leaves and sky have the color appropriate for this level.

Code Creatures

 













The 5900Ultra wins in this test thanks to the optimizations. NVIDIA's latest response is that optimizations are possible everywhere be it a game or a benchmark if a given application is popular. They affirm that if quality is not affected, there is nothing to worry about. Well, the quality does not get worse because of the optimizations in this benchmark, that is why we just establish a fact that the 5900 Ultra takes the lead here.

Unreal Tournament 2003

 













The speeds of the competitors are approximately equal. But what about quality? As you know UT2003 allows turning on/off the trilinear filtering and changing the anisotropy degree (in UT2003.ini). Let's enable just the trilinear filtering in the High Quality mode for both cards (ATI/NVIDIA). MIP levels are colored with the variable "FirstColoredMIP 1".
 

RADEON 9800, CATALYST 3.6






RADEON 9800, CATALYST 3.7






FX 5900, Detonator 45.23






FX 5900, Detonator 51.75






Although the drivers have the Quality preset the trilinear filtering is crippled. The version 51.75 didn't help. But I must say that I haven't notice the difference between the simple and full methods in the game, the border between the MIP levels doesn't catch my eye.

Now I'm forcing 16x and 8x anisotropic degrees at the Quality preset in the drivers (the game itself has anisotropy disabled).
 

RADEON 9800 Quality CATALYST 3.6


 

RADEON 9800 Quality CATALYST 3.7


 

FX 5900 Detonator 45.23


 

FX 5900 Detonator 51.75


 

Surprise again! Well, we know from the previous test that NVIDIA lightens the trilinear filtering, but ATI also demonstrates it and last time they explained it with an error in the drivers. Why isn't it corrected now? But if NVIDIA can do that, why wouldn't ATI do the same. Poor users...

Now let's enable anisotropy in the game by correcting ut2003.ini and disable it in the drivers. 
 

RADEON 9800 CATALYST 3.6


 

RADEON 9800 CATALYST 3.7


 

FX 5900 Detonator 45.23


 

FX 5900 Detonator 51.75


 

Finally the RADEON 9800 shows the normal trilinear filtering! ATI says it's because of the drivers' control panel when in the Quality mode the trilinear filtering is executed only for the texture stage 0, all the rest go with the bilinear filtering. It works for all games which "know" the independent anisotropy. But such an unpleasant picture appears because of the peculiarities of multitexturing in the UT2003. We were promised to have this flaw corrected, but it still exists.

As to the FX 5900, it doesn't have the normal trilinear filtering. It looks even worse than in the 45.23.

Well, everyone including the game developers are to be blamed. 

  • ATI: the full-featured anisotropy works with the trilinear filtering only when it's set in the game itself, which must be done manually in ut2003.ini. It's unpleasant even when accounting for how ATI explains it. 
  • NVIDIA: optimizations... What to do with the quality? 
  • Game developers: if you include the anisotropy management into the game why to hide it from users? Why wouldn't you bring it into the game options panel?

The discussion will be developed in the next test as it's based on the same engine. 

Unreal II: The Awakening

 













The FX 5900 Ultra still works slower though the gap is not that big as before. I wonder if optimizations are used here as well... the speed has grown by 51.75.

Unfortunately, MIP levels in the Unreal II are all colored into black starting from the second level. However, it perfectly shows the border between the MIP levels. Let's force anisotropy in the Quality mode where the normal trilinear filtering must work (ANISO Forced) and also enable it in Unreal2.ini (ANISO by Application) switching the forcing off:
 

RADEON 9800 CATALYST 3.6, ANISO Forced!


 

RADEON 9800 CATALYST 3.7, ANISO Forced!


 

RADEON 9800 CATALYST 3.7, ANISO by Application!


 

FX 5900 Detonator 45.23, ANISO Forced!


 

FX 5900 Detonator 51.75, ANISO Forced!


 

FX 5900 Detonator 51.75, ANISO by Application!


 

The trilinear filtering works though I think it looks lighter in case of the FX 5900. NVIDIA shows no difference between the anisotropy forcing and its enabling in the game while ATI has a little one.

With the v51.75 the FX 5900 loses fog in some scenes, which didn't take place in case of v45.23. 

RightMark 3D

 













The RADEON 9800 grasps the victory thanks to the higher shader speed. 

Conclusion

Well, judging by the overclocking scores the water cooling is really effective. The system was equipped with NO additional fans. The overclocking helps a lot to the GeForce FX 5900 Ultra. But is it worth such an unreasonably high price? Not sure. But nobody talk Matrox out of selling its Parhelia at high prices. I think there'll be few Gainward Powepack FX Ultra/1600 GS CoolFX cards released. Primarily for enthusiasts.

The driver version 51.75 was examined today only in the traditional tests (let's call them old), where shader technologies are applied only in two tests. There're some problems with quality, but it's not the final version yet. I hope NVIDIA will solve the problem. As to the optimizations, this issue is still discussable. Can NVIDIA fool those users who paid a lot for a top product by doing nothing to lift the speed? Surely, they can't. I think that by lowering quality levels in unimportant scenes the experts do increase the speed. There're no other wayouts. Another problem is the price.. it certainly should be cut down.

In general, the Gainward Powepack FX Ultra/1600 GS CoolFX can offer you:

  1. high 3D speed, almost the best, and definitely the best in the overclocking mode; 
  2. finished modern technological solution with the DX9 support; 
  3. bulky and complicated cooling system, which is not easy to assemble and arrange inside or near the PC case; 
  4. a bit noisy pump (there are rubber fixers for the pump).

In our 3Digest you can find full comparison characteristics for video cards of this and other classes. 
 
 
 

Andrey Vorobiev (anvakams@ixbt.com)
Ivan Pritula (iamdevil@3dmark.ru
 

Write a comment below. No registration needed!


Article navigation:



blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook


Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.