by Anton Shilov , Alexey Stepin
02/13/2004 | 03:48 PM
The original Unreal game released in May 1998 changed the way people treated computer games in general and first person shooters in particular. For the very first time a game was truly beautiful from esthetic standpoint – there were trees, birds, animals, caverns, ruins and never-before-seen water along with loads of other things that made the crowd to love the Unreal.
<%BANNER[article]%>A year later Epic Games introduced the Unreal Tournament – a bloody first person combat that really put you into action, a kind of action you would expect from a “tournament” – not so smart as in Counter Strike and not so fast and furious as in Quake III Arena. The community appreciated the Unreal Tournament not only for excellent graphics, perfect weapon and comfortable game play, but also for exciting maps and game-types.
Three years later, in 2002, Epic rolled-out the new Unreal Tournament 2003 with innovative 3D engine, fresh levels, new worlds and incredibly complex and realistic graphics. Now it is early 2004 and the developers are readying the UT2004 – a seriously improved version of the 2003 with just more realism and allure. There is a paradox – while the worlds of Unreal become more and more fabulous, the game really gets more and more realistic.
Hey, are they preparing us for Mars invasion or something? :)
Given rather short timeframe between the UT2003 and UT2004, do not expect a lot of technology innovations in the latter. The Unreal Tournament 2004 does not bring much more shader effects and definitely does not pack a new lighting model – the engine is still the same as in the Unreal Tournament 2003, but with some modest improvements, mainly – special effects, detailed geometry and complexity of buildings. Constructions and levels just got bigger and more imposing. Now you really understand that calm grandeur of the tournament!
Briefly speaking from technology and hardware standpoint, in case your computer played the UT2003 well, it is nearly totally likely to swallow the UT2004 too, if not, you will surely not get satisfied with on the whole more demanding game at all.
The main updates in the Unreal Tournament 2004 concern maps, bots, new game modes as well as availability of vehicles along with aircrafts. General weapons were not changed substantially, but there are some new items available in a new game mode called Onslaught.
In fact, Onslaught is the main new adventure of the whole title! This is where you will get numerous new weapons as well as will be able to ride vehicles and even fly! The idea of the game is to gain control over a number of special points called nodes. Once your team controls all such points, you will be able to destroy your enemy’s base. The levels of the Onslaught are really massive – you just need vehicles and aircraft to move from point to point. While FPS players may not be really accustomed to see such kind of stuff in a combat game, believe me, driving such vehicles is a lot of fun. This is not Carmageddon or Need for Speed, but still very exciting.
Assault game-type is something we all missed in the Unreal Tournament 2003 and that is available in 2004. Yet another environment that strongly requires team-work, as Onslaught, became even more interesting compared to the original Unreal Tournament.
You will not find anything new in the game-play of the UT2004 compared to the previously released Unreal Tournament titles, this is what you have to understand so not to get disappointed. If you liked the UT2003, you are going to love the UT2004, as the things generally got better.
In the death match modes you have to be rather quick, still, not that quick as in the Q3A, but still have to fight your way through, not to hide in the corners.
In Capture the Flag and Bombing Run modes you main ability is to run. Not that you do not need to shoot your opponents at all, but in the majority of the cases it is more crucial to escape rather than to “kill ‘em all”.
In the Assault mode sniping and team-work become two great advantages you need to have; as in the original UT there are both “fight through” and “intelligent” elements in the Assault.
Onslaught, as the name suggests, demands your ability to take the control nodes by storm, to protect them from invasion and to be able to organize another attacks with your teammates involved. Everything should be done in a rather quick manner even with bots, I hardly imagine the skill required to play Onslaught via the Internet.
On the whole, this is just a big fun to play the Unreal Tournament 2004. There is no thrill you will experience in games like the Half-Life 2, there is no so lot of adrenaline you get from the Quake III Arena and there are no dozens of cool-headed decisions you have to make in games like CS of Battlefield 1942.
In case you downloaded the Unreal Tournament 2004 demo, you probably can experience the game yourself. In case not, check out the list of UT2004 Demo downloads over here. As a hardware web-site, X-bit labs delivers what you probably came here for – tests of modern hardware using up-to-date software.
For our graphics cards comparison we used the same AMD64 rig we use for other reviews of graphics cards:
What was great about our graphics cards’ test is the absence of any glitches and artefacts in the game. Epic as well as graphics chips makers did a nice job with polishing the game and the drivers. Well, the engine is the same as in the Unreal Tournament 2003 – why should it have problems?
Since the Unreal Tournament 2004 Demo contains no pre-recorded demos and because of our policy to use our own demos and benchmarks when possible, we recorded a number of sequences in all levels available in UT2004 demo. Given that the demo has less-detailed textures compared to the full version as well as because of further optimizations of the game engine and drivers, the actual title that will ship at a later date may perform a bit differently. Quite naturally, benchmark results obtained reflect performance only in the particular environments, though, this allows us to draw some trends.
As a rule we measured speed in “pure mode” 1024x768, 1280x1024 and 1600x1200 resolutions without anisotropic filtering and antialiasing enabled as well as in the same resolutions, but with 8x Quality anisotropic filtering and 4x FSAA on. Speed numbers for S3 Graphics and XGI hardware are only available for “pure” modes because in “eye-candy” cases drivers from those companies did not enable FSAA properly.
All game settings were at the highest possible levels. Textures were at “Normal” simply because the demo does not include high-quality textures because of the size constraints.
We used the following levels for benchmarking purposes:
We will begin with Onslaught environment, as it is basically the core of the whole UT2004 game. Graphics and effects on the Torlan level do not seem to be very complex for graphics cards, there is hardly too lot of rather tough geometry in the level and there are just a few shader effects. But the map itself is just pretty large – is puts the graphics processors under very heavy load.
Without anisotropic filtering and FSAA in a mode we call “pure” RADEON products from ATI Technologies are faster compared to their competitors based on NVIDIA GeForce FX processors. However, the gap does not appear to be really tangible. S3 Graphics’ DeltaChrome S8 delivers nearly enough speed, at least, it is sometimes more than twice faster compared to XGI Volari Duo V8 Ultra.
The situation with speed in FSAA 4x + Anisotropic Filtering 8x – we call this mode “eye-candy” – is nearly the same as in “pure” – performance of comparable RADEON and GeForce FX products is merely the same.
Something we were not surprised to see at all is the RADEON “dead man” 9500 PRO not only walking, but alive and kicking!
The Convoy resembles popular train assault map from the original UT. This time we do not have moving train, but a number of vehicles moving along a canyon. Given that there are not a lot of tough graphics elements, but dozens of physics calculations, the level is more CPU dependent than other demos.
When talking about graphics cards’ speed, we should exclude 1024x768 resolution, as it shows CPU addiction along with quality of drivers. In 1280x1024 and 1600x1200 resolution we see ATI leading among the high-end graphics cards and very close performance of mainstream products by ATI and NVIDIA. Notice very positive S3 Graphics results and rather pale XGI’s.
With eye-candy enabled, ATI’s RADEON 9800 series reiterates its performance leaderships, while ATI’s 9600 series falls behind the GeForce FX 5700 Ultra that is the clear winner in mainstream.
Bridge of Fate level includes a lot of columns, caverns, doors, walls, etc. This all results in pretty high overdraw as well as increased complexity of geometry details. Actually, based on the obtained results we may say that modern graphics cards are more than capable of successfully handling such load.
Powered by ATI RADEON products appear to be slightly faster than rivalling offerings based in NVIDIA GeForce FX GPUs. DeltaChrome S8 is on par with the GeForce FX 5600 Ultra, while XGI’s dual-chip monster is in the very bottom of the charts both directly and indirectly.
Solutions with ATI chips onboard are a bit faster compared to NVIDIA-powered again. There is a breakthrough for the GeForce FX 5700 Ultra in 1600x1200 resolution, but it does not improve the whole picture. The margin between equivalent ATI RADEON and NVIDIA GeForce FX products is not too significant, but what is faster is faster.
Colossus level is rather large and grandeur. It also has loads of geometry details and high intensity of overdraw. But all graphics cards, except XGI Volari V8 Ultra, still managed to work well.
All the comments said about performance in the Unreal Tournament 2004 are effective here. Putting aside single-digit differences between the results of graphics cards built upon chips from different vendors, we have right to say that they are more or less close.
But there are some pretty unexpected things about results of NVIDIA’s graphics processors here. Pay attention how the GeForce FX 5900 falls behind the GeForce FX 5950 Ultra by about 25% in 1280x1024 resolution, a figure we did not see in previous test scenes. We cannot attribute this gap to extra 128MB of memory the 5950 Ultra has over 5900, as we do not see any serious improvements the RADEON 9800 XT has over the 128MB RADEON 9800 PRO. All we can suggest is that massive improvement comes from clock-speed increase of the GeForce FX 5950 Ultra. Maybe additional 128MB play a role, but not a very important one.
We were surprised to see S3’s DeltaChrome S8 beating the RADEON 9600 PRO by even a small single-digit number.
The results in “eye-candy” mode are in line with expectations: ATI’s high-end VPUs are ahead competition, ATI’s mainstream VPUs are a bit slower than comparable rivals. The GeForce 5900 is seriously behind the RADEON 9800 PRO and even more behind the GeForce FX 5950 Ultra.
The only death match map available in UT2004 Demo also shows CPU dependence. Nevertheless, those graphics cards that are slow in general are seriously behind the winners.
ATI RADEON 9800 XT and ATI RADEON 9600 XT clearly win the performance crown in their classes. The history with GeForce FX 5900 considerable performance defeat to the FX 5950 Ultra is repeating here – the card is roughly 20% slower in 1280x1024 resolution.
S3 DeltaChrome S8 manages to outperform the GeForce FX 5600 Ultra, unlike XGI’s dual-chip monster Duo V8 Ultra that is seriously behind the rest.
ATI RADEON 9800 XT and RADEON 9800 PRO win the high-end battle, while the ATI RADEON 9600 XT this time cannot keep the mainstream fortress that is taken by storm by NVIDIA’s GeForce FX 5700 Ultra. Moreover, even the RADEON 9600 PRO falls behind the GeForce FX 5600 Ultra for some reason in this demo.
Now let us try to put the UT2004 performance facts straight for every graphics card we tested.
This is not a news that ATI offers excellent technology for high-end graphics cards. But what about mainstream solutions?
We did not concentrate on image quality, uncovering possible cheats and carefully investigating every single aspect of performance. We can prove that there are no major issues with graphics quality on ATI, NVIDIA and S3 hardware, though, our aim was not to detect such problems, but to examine the possible “out of the box” experience.
 |  | |
| ATI RADEON 9800 XT | NVIDIA GeForce FX 5950 Ultra |
Here is how the Unreal Tournament 2004 looks like on ATI RADEON 9800 XT (on the left) and NVIDIA GeForce FX 5950 Ultra (on the right) graphics cards.
The Unreal Tournament 2004 is nearly here. Today we have got a demo version with less-detailed textures to measure performance of graphics cards we have at hands here. We discovered that even with “normal” textures quite some mainstream graphics cards cannot give us enough speed. The results with highest quality textures will be even lower, therefore, if you want to play the UT2004 at full-speed with excellent quality – consider buying a high-end graphics card.