What’s red, copper, black and comes in a box made of plastic? Xtasy, what else? To be specific, it’s the VisionTek Xtasy Radeon 9800 XT. Packing 256MB of RAM, the 9800 XT is a slightly higher clocked 9800 Pro with a new and improved cooler design. Read this review to see how it compares to NVIDIA’s top dog in the form of the Leadtek 5950 Ultra.
The battle for video card supremacy recently became more interesting with the latest product refreshes from NVIDIA and ATI. The hardware I’m alluding to is of course the NVIDIA GeForce FX 5950 Ultra and the ATI Radeon 9800 XT. I recently took a closer look at a retail 5950 Ultra (Leadtek WinFast A380 Ultra TDH MyVIVO), but this time around, a retail 9800 XT gets the spotlight.
That’s right. It’s time to turn the focus on the VisionTek Xtasy 9800 XT, which is a 256MB card just like the 5950 Ultra. Also just like the 5950 Ultra, it’s basically just an overclocked version of its father. In this case, that father is a 9800 Pro, which has a VPU core clocked at 380MHz and memory running at 340MHz (680MHz DDR). The XT flexes its muscles a little harder and runs at 412MHz core and 365MHz memory (730MHz DDR). In addition to being clocked higher, the XT features a new and improved cooler design. It’s a lot more interesting design than on the basic 9800 Pro cards.
In this review, I will focus on how the card performs compared to the Leadtek 5950 and also briefly cover the new Overdrive feature available in ATI’s latest sets of Catalyst drivers. As usual though, let’s start with the features, specs and package contents.
Features, Specs and Package Contents
- AGP 8x Support
- CATALYST Driver Support
- SMARTSHADER 2.1
- SMOOTHVISION 2.1
- Hyper Z III+
- TRUFORM 2.0
- 8 Parallel Rendering Pipelines
- 4 Parallel Geometry Engines
- F-Buffer Technology, 128-bit floating point color precision, 2nd generation N-patch higher order surface, FULLSTREAM VIDEOSHADER
- Integrated TV out support up to 1024×768 resolution. De-interlacing and frame rate conversion, dual integrated 10-bit per channel palet DACs, integrated 165MHz TMDS transmitter (DVI 1.0 compliant). Unique per-pixel adaptive de-interlacing. Optimized for Pentium 4 SSE2 and AMD Athlon 3Dnow! PC 2002 and FTF compliant.
- Lifetime warranty
- Toll-free tech support
- ATI Radeon 9800 XT VPU
- 412 MHz VPU Core
- AGP Bus
- 256MB DDR Memory
- 256-bit, Quad-Channel Memory Interface
- Memory Clock: 365MHz (730MHz DDR)
- Dual 400MHz DACs
- API Support: DirectX 9.0, OpenGL 2.0
- Connectors: VGA, DVI-I, TV/S Video
- XTASY Radeon 9800 XT Graphics Accelerator
- Quick Install Guide
- 3D Demos
- ATI Catalyst Drivers
- S-Video Cable
- DVI-to-VGA adapter
- Power Cable
- Composite Video Cable
- Half-Life 2 Coupon
Xtasy 9800 XT Box – Front
Xtasy 9800 XT Box – Back
Xtasy 9800 XT
Cables … Exciting Aren’t They?
Half-Life 2 Coupon — ‘Nuff Said!
VisionTek opted to use the exact same packaging as they did with the Xtasy 9800 Pro. The box is plastic instead of the typical cardboard. It’s kind of neat but tends to be annoying when opening and closing. As is common with VisionTek’s graphics cards, the package is rather sparse. There is one major exception to this though, and that is the inclusion of a coupon to get Half-Life 2 free once it is released! You probably already know that VisionTek is not the only company to offer Half-Life 2 in its card bundle. Most, if not all, of ATI’s board partners have opted to include this coupon in their XT packages, and I’ve even seen it offered in some non-XT bundles.
The Card and Installation
I’m glad to see that ATI got a little more creative with the 9800 XT’s cooling solution. It looks cooler (no pun intended) than any other reference ATI design I’ve ever seen. On the practical side, it is all copper-based and is of course big enough to cover all the memory chips (on both sides) and the VPU. ATI also added a bigger fan to help cool the card. The other side is cooled by a copper plate that sits directly against the chips.
Although the performance and cooling have been beefed up for the 9800 XT, the physical size is still considerably smaller than the 5950 Ultra. The 9800 XT is not as long or as thick (referring to the thickness due to the HSF) as the 5950U. The 9800 XT is heavier than the 9800 Pro 256MB though because of the new copper cooling solution. Let’s take a closer look at the card, especially its cooling solution.
Closer Look at the Cooler
Copper! And the Power Connector
L to R: VGA, S-Video, DVI
There Are 13 of Them…Blades That Is
Copper Plate on the Back of the Card
As expected, I didn’t run into any problems during installation of the Xtasy 9800 XT. Because the card is not very long and only takes up one slot, size should not be an issue in any typical system. High-end ATI cards always seem at least a little easier to handle than equivalent NVIDIA cards when working inside a case.
I removed the previously installed NVIDIA drivers and then removed the 5950 Ultra after shutting down. Next, I plugged the 9800 XT into my MSI motherboard’s AGP slot and fed it a molex power connection from my power supply (don’t forget to connect that power cable!). The board fit easily with plenty of room around it. I closed everything up and booted up to load the drivers.
I downloaded the latest driver package – Catalyst 3.9 – from ATI.com and installed it. Once again, I had no problems at all.
Performance / Benchmarking
To test the Xtasy 9800 XT, I used a range of benchmarks and compared it to its main competitor – the 5950 Ultra, specifically the Leadtek A380 Ultra TDH MyVIVO. Because the 9800 XT has slightly bumped up clock speeds compared to the 9800 Pro, we know that the XT will perform at least marginally better than its predecessor. To see just how much better the XT performs than the Pro, check out this page of Bjorn’s 9800 XT article from a couple months ago.
For testing, I used the following benchmarks and games: Futuremark’s 3DMark2001 SE – Build 330, 3DMark2003 – Build 330, AquaMark3, Codecreatures Benchmark Pro, Unreal Tournament 2003, Gun Metal Benchmark 2, X2: The Threat rolling demo, and Papyrus’ NASCAR 2003. In all tests, performance is set to “Quality” to create the best image quality. When possible to set anti-aliasing and anisotropic filtering in the benchmark, the drivers were set to “Application Preference.” Otherwise, the levels were changed manually. Before getting to the scores, let’s take a look at my test system and settings.
|Motherboard:||MSI K8T Neo-FIS2R (review)|
|CPU:||AMD Athlon 64 3200+|
|Memory:||2 x 256MB XMS3500 Corsair DDR|
|OS:||Windows XP SP1 with all updates installed|
|Videocard 1:||Leadtek WinFast A380 GFFX 5950 Ultra with ForceWare 52.16|
|Videocard 2:||VisionTek Xtasy 9800 XT 256MB with Catalyst 3.9|
Okay, the race is about to begin, so place your bets. Considering that the Radeon has become the high-end leader, it makes some sense to place your money on the XT, but don’t underestimate that 5950. It’s definitely not going to be a landslide victory one way or the other. Let’s just look at the benchmark results, shall we?
3DMark01SE (Build 330)
I’m starting this benchmark party off with 3DMark2001SE build 330. I had expectations that the 9800 XT would finally push me over the 20,000 mark. I can still remember being excited about surpassing the 10k mark! Sheesh…
Oh yes… I can now shoot for 25,000 3DMarks. The Athlon64 system really helps push the 9800 XT to its potential. Although the 9800 XT takes the first benchmark setting easily, it falls behind the 5950 Ultra in the last two settings. I was surprised to see such a strong lead dwindle and lead to being significantly behind. This is just the first benchmark though, so let’s move on.
3DMark2003 (Build 330)
The constant controversy surrounding this benchmark doesn’t prevent it from being a useful tool. Although the validity of synthetic benchmarks seems to continually be in question for various reasons, I don’t think they’ve completely lost their appeal. Many readers find it easy to go and download a free benchmark, install it, run it and then compare their results to their friends’ results. A lot of games just aren’t that easy to benchmark, so it’s a matter of convenience. With that said, let’s take a look at the results.
The Xtasy 9800 XT puts in a very convincing victory here and leaves no room for doubt at the finish line.
AquaMark3 is a good looking benchmark based on the same engine used in AquaNox 2. I ran the default benchmark and then a custom benchmark at 1280×1024 with 4xAA / 8xAF.
Once again, the 9800 XT wins over the Leadtek 5950 Ultra by a good margin.
Unreal Tournament 2003 – Inferno
For the UT2K3 tests, I used the benchmarking utility from benscustomcases.com. Within the utility, I used the HardOCP standard benchmarks at high quality. I recorded the average FPS for Inferno and Antalus.
The 9800 XT looks like it might not be able to keep up with the Leadtek 5950U at first, but then once AA and AF are cranked up, the XT decides to strut its stuff. It surpasses the 5950U by a considerable margin at both 1280×960 4xAA/8xAF and 1600×1200 4xAA/8xAF. It’s very interesting to see these two cards can start out completely opposite of how they finish in these benchmark sets.
Unreal Tournament 2003 – Antalus
While the VisionTek 9800 XT dominated the Inferno map, it does not put up a repeat performance on the Antalus map. The Leadtek 5950 once again beats the 9800 XT decisively at 1024×768 NoAA / NoAF, and it also beats the XT at the other two settings but not by as much as the XT beat the 5950 on the Inferno map. What does this tell us? Basically, each card performs best on different maps. Both perform quite well though and will allow you to crank up the quality settings and resolution.
Gun Metal Benchmark 2 – Benchmark 1
Gun Metal was originally an Xbox game, but it has been ported as a DirectX 9 game to the PC. The Gun Metal Benchmark 2 allows you to choose between two different benchmarks, creatively named Benchmark 1 and Benchmark 2. I set AA to 4x within the benchmark’s options and left AF to “Application Preference” in each card’s respective control panel. I ran Benchmark 1, and you can see the results below.
It is typical for nVIDIA cards to beat ATI cards in this benchmark. Some would guess this is due to the fact that the game was originally developed for the Xbox, which happens to use an NVIDIA graphics solution. That’s a reasonable assumption. Regardless, the 9800 XT just can’t match the performance of the 5950U at any of the settings.
X2: The Threat – Rolling Demo
X2: The Threat is another DirectX 9 game, but it is much newer than Gun Metal. It was actually just recently released. The non-playable, rolling demo shows you what the game will look like as it runs through many different types of in-game scenes. Rather than show the results of each scene, I have included the overall average FPS. This demo allows you to toggle bumpmaps, shadows and AA. I left bumpmaps and shadows on for all benchmarks but changed AA as indicated in the chart.
Just like in Gun Metal, the Leadtek 5950U dominates this benchmark at each setting. I was surprised by how much the 9800 XT trails the 5950U here. Keep in mind this is just the overall average FPS score. It doesn’t mean the 5950U scores the highest in ever scene rendered. The 9800 XT actually beats the 5950U in several scenes by a fair margin, but the average tells us that the 5950U puts up some higher scores where it counts.
Papyrus Nascar 2003 Demo
Nascar 2003 from Papyrus can really test your system, which is why we started using it as a benchmark. In order to record FPS, we use Fraps 2.0. The demo race takes place on the Talladega track, and I recorded performance for two minutes. All quality settings were set to maximum. Here are the results.
At first, it looks like it’s going to be close between the two cards, but then the 9800 XT pulls out in front by a considerable margin at both 1280 4xAA/8xAF and 1600 4xAA/8xAF.
The Xtasy 9800 XT is of course a top of the line card when it comes to performance. It beats the 5950 Ultra in many benchmarks, but it doesn’t win them all. I briefly tested the TV-out performance on my 27″ TV, and the XT provided excellent visual quality for my HTPC setup. I also used the card in a small form factor system to make sure it could handle the cramped space well, and it did. The card appeared to be keeping cool enough while I played several games.
Of course gaming with the 9800 XT was great. I cranked up the visual effects and resolution in several newer games, like Max Payne 2 and Jedi Academy, and there was no stuttering during many hours of playing. I mostly tested the card at 1280×1024 4xAA/8xAF because my resolution is limited by my 17″ LCD monitor.
Overdrive & Overclocking
The Catalyst drivers and XT card offer an automatic overclocking feature called Overdrive. The first step I took for overclocking was to just try out Overdrive to see if it’s really a viable option for added performance. Unfortunately, Overdrive really didn’t do much in my testing. In the benchmarks I ran, very little was gained. One FPS here and 100 3DMarks there really isn’t worth the effort of going into the driver control panel and clicking the Overdrive option, is it? Hardcore enthusiasts will definitely still want to manually overclock, but newbies still might like the feeling of brainless overclocking, even if it’s not actually doing a whole lot. Overdrive is supposed to monitor the chip’s speed and temperature and push it to its limit whenever the demand is present (i.e. while playing a game with intense graphics). If the card gets too hot, it will slow clock speeds down. If it’s chugging along and not breaking a sweat, Overdrive should kick in and push the card harder. ATI seems to have a good idea here, but I can’t see very many people checking that box until the driver automatically pushes the card even more.
During my manual overclocking, I used the included PowerStrip tweaking utility (included on the driver CD). I was able to push the core clock to 430MHz and the memory to 380 (760MHz), which is a decent overclock. I’ve seen overclocks for the 9800 XTs range from 425 – 480MHz core and 370 – 410MHz (740 – 820MHz) memory, so your mileage will vary. Once I get this card in a permanent home, I think I’ll try to push it a little more. At 430 / 760, performance picked up a bit but was only really noticeable in the more intense benchmarks, such as 3DMark2003 or when the settings were 1280×1024 4xAA/8xAF and 1600×1200 4xAA/8xAF.
Feaures & Bundle
Even though the Xtasy 9800 XT is just a reference card and doesn’t feature a unique PCB color or custom cooling unit, it still is a good looking card with some promising features. The reference cooling unit is copper-based and remains relatively quiet during operation. I didn’t notice it over my case fans, but keep in mind I don’t have a super quiet system. The TV-out option and quality should make most HTPC enthusiasts happy.
Undoubtedly, one of the favorite extras in the XT packages is Half-Life 2. This game alone sets bundles like the one included with the Xtasy 9800 XT apart from the competition’s offerings. As a big fan of big but practical bundles, I am ecstatic to see companies making the extra effort to deliver bundle pieces that really do add value to the complete package. Considering that HL2 will probably retail at around $45-50 like most new games, one can’t deny the value added to this VisionTek package. The bundle also includes all the cables you will need and the PowerStrip tweaking utility. Overall though, this bundle is pretty boring and sparse with the exception of HL2.
- Nice and quiet cooling solution
- Great performance
- Lifetime warranty
- Coupon for Half-Life 2 included!
- Not as big as 5950 Ultra – easier to physically install in some cases
- Expensive like all high-end cards
- Overdrive feature needs work
- Bundle pretty basic (besides HL2 of course)
If you don’t want to or don’t care about turning on as much visual quality as possible, why buy a 9800 XT? Or any top card for that matter? Save your money and buy a mid-level card. This is obvious to many of you, but I think there are a lot of people who need reminding. Neither the 5950 Ultra nor the 9800 XT can promise to deliver the highest performance every time in every game, so which card do you buy if you go the expensive, high-end route? It’s not an easy question to answer. You should look at benchmarks of the games you play on as many review sites as you can and gather as much impartial information as you can. In short, do your research first. That said, I can recommend the Xtasy 9800 XT to those of you who choose the Radeon path. It’s a top performer with a decent bundle. Top performance costs top dollar though unfortunately, so you can expect to pay around $480-500 for this card (CompUSA.com, TigerDirect.com). If big bundles are your thing, then you might be able to find a better bundle with a 9800 XT for around the same price.