iXBT Labs - Computer Hardware in Detail








  1. Introduction
  2. Video cards' features
  3. Testbed configurations, benchmarks
  4. Test results: TRAOD
  5. Test results: FarCry Research
  6. Test results: FarCry Regulator
  7. Test results: FarCry Pier
  8. Test results: Half-Life2, ixbt01
  9. Test results: Half-Life2, ixbt02
  10. Test results: Half-Life2, ixbt03
  11. Test results: DOOM III
  12. Test results: 3DMark05 MARKS
  13. Test results: Chronicles of Riddick, demo 44
  14. Test results: Chronicles of Riddick, demo ducche
  15. Conclusions

Soon we are going to celebrate the anniversary of the new PCI-Express interface (PCI-E). But over more than 9 months of actual sales, PCI-E based products (both video cards and motherboards) didn't manage to conquer even half of the market (its corresponding segments, of course).

The demand intensity for new products with AGP interface already begins to decline, but still too many users want to buy exactly AGP video cards. It must be admitted that NVIDIA had a better forecast of the future situation. That's why the company designed an all-purpose HSI bridge from the very beginning. It had been initially used to manufacture video cards with old AGP chips, but equipped with a PCI-E connector. Then, vice versa, it was used to transform more advanced and interesting solutions with a GPU-embedded PCI-E interface into similar AGP products (GeForce 6200/6600/6800).

Canadian ATI was late, its experts miscalculated their forecasts, and its RIALTO bridge actually appeared only by the spring 2005, while the peak demand for the new AGP models fell on the end of 2004, when GeForce 6600GT AGP sales were simply fantastic. The Canadian company offered nothing but the outdated RADEON 9800 PRO in this market sector. More advanced RADEON X800 PRO was too expensive. Even now it comes at a high price, that's why it just leaves the market.

Let's recall the alignment of forces in the PCI-E sector:

  • RADEON X850 XT PE competes with GeForce 6800 Ultra (their prices are approximately at the 600-620 USD level)
  • RADEON X800 XT PE (leaves the market) also competes with the 6800 Ultra
  • RADEON X850 XT competes with GeForce 6800GT (the prices are about 500 USD, but in this case the 6800GT is still cheaper)
  • RADEON X800 XT (leaves the market) also competes with the 6800GT
  • RADEON X850 PRO competes with GeForce 6800GT (the situation with prices is vague so far)
  • RADEON X800 XL competes with GeForce 6800
  • RADEON X800 competes with GeForce 6600GT

And that's the situation in the AGP segment we used to have up to now:

  • RADEON X800 XT PE competes with GeForce 6800 Ultra
  • RADEON X800 XT competes with the 6800GT
  • RADEON X800 PRO competes with GeForce 6800GT
  • RADEON 9800 PRO competes with ... probably only GeForce 6600GT and is outperformed
  • RADEON 9800 competes with GeForce 6600.

Pay your attention to the gap between the X800 PRO and the 9800 PRO. It's well known that the price for the former accelerator does not go down, it froze at 400 USD, while the price for the latter is going down very slowly. New products are obviously wanted at the prices between $400 and $200.

But with the appearance of products with RIALTO bridge, actually available in stores, as well as specially released X850 cards with built-in AGP interface, the situation looks like this:

  • RADEON X850 XT PE competes with GeForce 6800 Ultra
  • RADEON X800 XT PE leaves the market
  • RADEON X850 XT competes with the 6800GT
  • RADEON X800 XT leaves the market
  • RADEON X800 PRO competes with GeForce 6800GT
  • RADEON X800 XL AGP competes with GeForce 6800
  • RADEON X800 AGP competes with GeForce 6800LE and the 6600GT
  • RADEON 9800 PRO competes with GeForce 6600
  • RADEON 9800 (9800 PRO 128bit) also competes with the 6600 or even with the 6200.

The reasons why the X850 series had different chips for different interfaces, while it was decided to equip the X800 with the same GPU and install a bridge on AGP solutions will be covered in the next article about the X850 AGP.

So, we review the X800XL AGP for the second time. However, I already wrote that such video cards are manufactured on a single plant ordered by ATI, due to their layout complexity, while partners get only finished articles from it. Only labels and stickers will help to distinguish such PowerColor card from GeCube, etc. And the bundle certainly plays its role.

Video card

Interface: AGP 2x/4x/8x

Frequencies (chip/memory — physical (memory — effective): 400/495 (990) MHz (nominal - 400/495 (990) MHz)

Memory bus width: 256bit

Number of vertex pipelines: 6

Number of pixel pipelines: 16

Dimensions: 225 x 100 x 16 mm (the last figure is the maximum thickness of a video card).

PCB color: red.

Output connectors: DVI, d-Sub, S-Video.

VIVO: available (RAGE Theater)

TV-out: integrated into GPU.

The video card has 256 MB of GDDR3 SDRAM allocated in eight chips on the front and back sides of the PCB.

Samsung (GDDR3) memory chips. 2.0ns memory access time, which corresponds to 500 (1000) MHz.

Comparison with the reference design, front view
Reference card ATI RADEON X800 XL PCI-E

Comparison with the reference design, back view
Reference card ATI RADEON X800 XL PCI-E

There is even no need in pointing out the key difference from the PCI-E counterpart – it's the bridge:

Note two peculiarities: firstly, the bridge does not get heated much (up to 70-80 degrees, while the operating temperature of HSI may reach 100-110 degrees); secondly, the PCB is longer due to the bridge. Compare it with the widespread X850 PCB (the same PCB is used in the X800 XL PCI-E):

You can see that the card is longer even on the photos above. It should be noted that the PCB dimensions are even larger than in GeForce 6800 Ultra, which used to have the largest PCB among single-core video cards up to now. You should take it into account, if you plan on buying this card and you have a small PC case.

In order to avoid damaging the open RIALTO die, a soft heat-conducting material is glued to the chip.

The card is equipped with VIVO, where VideoIn is based on RAGE Theater.

Let's review the cooling system.


It's a closed-type heatsink with an off-center fan, which drives the air along the heatsink over GPU. The air is sucked in from inside the PC case and is blown out back to it, but it does not diminish the efficiency much, because the heating is not that intense (0.11-micron process technology, and the frequencies are not too high). The heatsink on the core is made of copper alloy. Memory chips on the front side use the same cooler, chips on the back side are cooled with a special plate.

A fan inside the cooler is equipped with a tachometer, so the rotational speed changes from 2500 to 5000 rpm depending on the heat. The cards produce no noise at startup.

You can watch a video and estimate the startup noise of the system unit at this link (1.1MB, AVI DivX 5.1). You can also compare it to the background noise without the video card (740KB, AVI DivX 5.1).

Let's have a look at the X800 XL core.

What we see is the same R430, which was for some reason called X800 PRO from the very beginning, though it's a sterling 16-pipe chip. Probably the company wanted to confuse users again (even before the final decision on X800 XL, X800 titles) and called such cards X800 PRO (to have them confused with the 12-pipe AGP counterpart).


The bundle contains: User's guide, CD with drivers, software bundle (Cyberlink PowerDVD and the CS:CZ game), DVI-to-d-Sub, HDTV, VIVO adapters.



A stylish cover with a box of white thick cardboard inside. The entire bundle is arranged into cardboard sections, so nothing dangles free inside the box.

Installation and Drivers

Testbed configurations:

  • Athlon 64 3400+ based computer:
    • CPU: AMD Athlon 64 3400+ (L2=1024K)
    • Motherboard: ASUS K8V SE Deluxe based on VIA K8T800
    • RAM: 1 GB DDR SDRAM PC3200
    • HDD: Seagate Barracuda 7200.7 80GB SATA

  • Athlon 64 (939Socket) based computer
    • CPU: AMD Athlon 3800+ (L2=512K)
    • Motherboard: ASUS A8N SLI Deluxe based on NVIDIA nForce4 SLI
    • RAM: 1 GB DDR SDRAM 400MHz
    • HDD: WD Caviar SE WD1600JD 160GB SATA

  • Operating system: Windows XP SP2 DirectX 9.0c
  • Monitors: ViewSonic P810 (21") and Mitsubishi Diamond Pro 2070sb (21").
  • ATI drivers 6.517 (CATALYST 5.3); NVIDIA drivers 71.84.

VSync is disabled.

That's how look temperature conditions of this video card (maximum temperature readings) operating in a closed PC case without additional cooling with minimal air flow (a fan at the rear of the case):

As you can see, everything is within the mark, no overheating problems. We used RivaTuner for monitoring purposes (written by A. Nikolaychuk AKA Unwinder).

Test results: performance comparison

We used the following test applications:

  • Tomb Raider: Angel of Darkness v.49 (Core Design/Eldos Software) — DirectX 9.0, Paris5_4 demo. The tests were conducted with the quality set to maximum, only Depth of Fields PS20 was disabled.

  • Half-Life2 (Valve/Sierra) — DirectX 9.0, demo (ixbt01, ixbt02, ixbt03 The tests were carried out with maximum quality, option -dxlevel 90, presets for video card types are removed from dxsupport.cfg.

  • FarCry 1.3 (Crytek/UbiSoft), DirectX 9.0, multitexturing, 3 demos from Research, Pier, Regulator levels (-DEVMODE startup option), Very High test settings.

  • DOOM III (id Software/Activision) — OpenGL, multitexturing, test settings — High Quality (ANIS8x), demo ixbt1 (33MB!). We have a sample batch file to start the game automatically with increased speed and reduced jerking (precaching) d3auto.rar. (DO NOT BE AFRAID of the black screen after the first menu, that's how it should be! It will last 5-10 seconds and then the demo should start)

  • 3DMark05 (FutureMark) — DirectX 9.0, multitexturing, test settings — trilinear,

  • The Chronicles Of Riddick: Escape From Butcher Bay (Starbreeze/Vivendi) — OpenGL, multitexturing, test settings — maximum texture quality, Shader 2.0, demo 44 and demo ducche.

    I wish to thank Rinat Dosayev (AKA 4uckall) and Alexei Ostrovski (AKA Ducce), who have created a demo for this game. I also want to thank Alexei Berillo AKA Somebody Else for his help

TR:AoD, Paris5_4 DEMO

Test results: TRAOD

This test goes by itself, because its load on shader units is heavy, even though it's the oldest among the currently used tests. That's why the difference between the X800 XL AGP and PCI-E is within 4-5% even considering the difference between the platforms. Our conclusion – both cards are practically identical in this test.

In other respects, we can see that the X800 XL AGP is the leader by the total score, having outscored even the more expensive 6800GT. It probably has to do with the fact that the pure peak scene fill rate is a tad higher in the X800 XL. As is well known, ATI products are better at calculations, while NVIDIA specializes in texturing, general pipeline performance, and shading. This game requires a great calculating capacity from a GPU. That's why ATI products have a little advantage.

FarCry, Research

Test results: FarCry Research

The following three tests belong to the same game, but the scenes are quite different, so they give a complex idea of accelerator's performance in this popular game (it's actually the first game with this new graphics level).

Taking into account that a processor plays a very important role in this game without additional load of AntiAliasing and Anisotropic Filtering (AA+AF), many powerful video cards are limited by the system resources and thus may demonstrate seeming parity, though nothing depends on a video card already and the final result is unpredictable. But if the cards are loaded with AA+AF, the X800 XL AGP is victorious.

FarCry, Regulator

Test results: FarCry Regulator

Even though the scene is quite different, the CPU being critical still has a strong effect (lots of physics, AI). If you run the game without AA+AF, almost all the results are equal, only the 1600x1200 mode demonstrates some differences. But AA+AF again tip the scale to the ATI products (I also mean video cards from ATI's partners). Note that now the leader of the race is the X800 XL PCI-E.

And even the more expensive 6800GT is again outperformed. It's not surprising as ATI drivers have been actively fine-tuned to this game for a long time and probably the Canadian programmers have been more successful than their colleagues from California. Some time ago ATI accelerators used to be outperformed by their competitors in this game.

FarCry, Pier

Test results: FarCry Pier

This scene is characterized by vast open space, where the number of polygons may reach 200000. That's why the vertex unit is heavily loaded and the scene is certainly very critical to a CPU. Thus, the performance without AA+AF is again limited by the system resources. That's why all the cards demonstrated approximately equal results.

The CPU is so critical to the performance that even activated AA+AF sometimes result only in insignificant performance drop at 1024x768.

Everything is the same in other respects, the X800 XL is victorious. Note that the X800 PRO almost always brings up the rear, so it's high time to dump this card.

Half-Life2: ixbt01 demo

Test results: Half-Life2, ixbt01

The next series of tests is also within a single game, but the scenes are different again, demonstrating versatile operations of accelerators. Note that this game is also CPU-critical, so a special attention should be paid to modes with AA+AF, because there is no point in buying such expensive accelerators (above $300) and not using anti-aliasing and anisotropy. Regardless of the fact that the X800 XL AGP still outperforms the X800 PRO and greatly outscores the 6800, the 6800GT has taken over the leadership anyway. It's more expensive though. That's why the X800 XL is generally more advantageous.

Half-Life2: ixbt02 demo

Test results: Half-Life2, ixbt02

Even though this scene differs much from the previous, the results are somewhat similar. Anyway, the X800XL is again victorious in this test, and even the more expensive 6800GT is outperformed. Though it should be stressed again that the difference is significant only at 1600x1200 with AA+AF. There is even no point in comparing cards without this load - the performance is limited by the system resources anyway (CPU).

Half-Life2: ixbt03 demo

Test results: Half-Life2, ixbt03

It's a similar situation again. As all the three tests at various scenes demonstrate similar relative results, we can say that the X800 XL is generally a HL2 leader in its segment.


Test results: DOOM III

Like TR:AoD, this game stands by itself. As is well known, it's developed so that NVIDIA products are in the lead due to optimal operations with a stencil buffer as well as with shading technologies in general (though part of shading in DOOM III is up to a CPU). Besides, the pipeline optimization (a famous feature of NVIDIA products) and more flexible and optimal operations with texture caches show themselves to advantage here.

So it's not at all surprising that the 6800GT has become a leader and the X800 XL AGP has been outperformed. Though it looks not very bad against the 6800, considering the heavy modes with AA+AF. And the new product from GeCube is really excellent compared to the X800 PRO (which is more expensive)!

3DMark05: MARKS

Test results: 3DMark05 MARKS

Motley 3DMark results cannot be generalized for all driver versions and test versions themselves, because, as is well known, the best programmers from ATI and NVIDIA are busy with optimizing drivers for this package. That's why each driver or test version may demonstrate a different alignment of forces. What concerns this very test, the X800 XL AGP has lost to the 6800GT, but it has greatly outperformed the 6800 as well as the X800 PRO.

Chronicles of Riddick, demo 44

Test results: Chronicles of Riddick, demo 44

And finally, here's the last series of tests with two scenes from the same game, characterizing it in different ways. This game depends less on a CPU than the previous ones; the load on shader units is heavy, but not dominating as in TR:AoD. A great deal of complex texturing, that's why a fine-tuned pipeline is critical. The same applies to the flexible and fine-tuned operations with caches. What concerns this test, the 6800GT is again victorious (it's not surprising, I have already mentioned the texturing optimization requirement). It should be noted though that the platform difference introduced some oscillations of performance delta between the X800 XL AGP and the X800 XL PCI-E.

Chronicles of Riddick, demo ducche

Test results: Chronicles of Riddick, demo ducche

We can see that X800 XL AGP is heavily outscored by nearly all competitors. What is it? An accident that depends on the test DEMO? Or this game has serious reefs for ATI products? Time will show. But I guess that a more fine-tuned operation of NVIDIA products with texturing, caching, and an optimally configured pipeline will not allow ATI products to shoot forward.


GeCube RADEON X800XL AGP 256MB is an excellent product for its price segment (if there is no artificial rise in prices!). These cards are expected to have a very high popularity (considering that the AGP sector is still very large). What concerns this sample, its operating stability is up to the mark, its quality is very high, 2D picture at 1600x1200@85Hz is sharp.

You can find more detailed comparisons of various video cards in our 3Digest.

Theoretical materials and reviews of video cards, which concern functional properties of the GPU ATI RADEON X800 (R420)/X850 (R480)/X700 (RV410) and NVIDIA GeForce 6800 (NV40/45)/6600 (NV43)

Andrey Vorobiev (anvakams@ixbt.com)

May 13, 2005.

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.