To that end, we have rounded up a healthy contingent of the latest high-end graphics cards for a shootout in DOOM 3cards that are relatively recent and available for about $299 and up. The list includes no less than four flavors of GeForce 6800, a bevy of Radeons, and even a GeForce FX. We’ve tested them all in both the single-player game and in deathmatch to see how they handle the game of the moment. And we’ve tested both performance and image quality exclusively with DOOM 3’s “High Quality” graphics settings, the best logical choice for gamers with decent graphics hardware.
So this is a serious comparison of DOOM 3 performance. Even our test system means businessan Athlon 64 3800+ with a gig of dual-channel DDR400 memory running at uber-tight timings.
Will NVIDIA’s prowess in OpenGL games translate into across-the-board dominance in DOOM 3, or will ATI’s new Catalyst 4.9 beta drivers propel the Radeon X800 back into contention? What is the best high-end graphics card value for DOOM 3? Read on to find out.
About DOOM 3’s High Quality mode
Before we go on, I should say a few words about DOOM 3’s High Quality mode, which we used for all our testing. The graphics geeks among us will want to know that HQ mode uses texture compression for diffuse and specular components, but not for normal maps, which tend to develop artifacting problems with common texture compression methods. This mode also incorporates 8X anisotropic filtering, so it’s always active in all of the tests you’ll see on the following pages. If you want to know more about HQ mode, check out the July 26 .plan update from id Software’s Robert Duffy. He explains more about what each mode is and why. DOOM 3 does have an “Ultra quality” setting that uses no texture compression at all, but Duffy says that setting is really most appropriate for future graphics cards with 512MB of memory onboard. All in all, HQ mode is what you’ll want to use to play DOOM 3 on any current high-end graphics card with 256MB of memory.
The demos
We recorded a couple of custom demos for benchmarking with DOOM 3. The first one, with the amazingly exciting name “trdemo2,” is a bit of action from the single-player game. If you watch it, you’ll see me getting beaten on by assorted demons and zombies. The lighting, effects, and bad guys that show up in this demo seem reasonably representative of the first third of the game or so (I’m not done yet!).
The second demo comes from a deathmatch game on the map “The Edge 2.” This map offers a rather different sort of environment than our first demo, with wider open spaces, different shader effects, and some very detailed “rocky” walls that I suspect use normal maps. You’ll see what I mean in the screenshots.
To record this demo, I set up a quick multiplayer server and started running around in the map. To my surprise, some bots showed up when the game started, ready to play. I’d just clicked through the default menus, and I hadn’t seen anything about bots, but there they were. I started recording and proceeded to get the stuffing beaten out of me by some bot named “Harold O.” Harold O. had some really good moves. After a little humiliation, I quit recording, and shortly after that, the “bot” disconnected.
At this point I realized that Harold O. was a real person, and that my server was announcing itself on the Internet. I didn’t know whether to be more humiliated, or less. Ah, well.
Now, let’s move on to the results.
Our testing methods
Both the ATI and NVIDIA cards were left at their driver default settings for image quality, with the exception that we turned off vertical refresh sync on all cards.
Our test system was configured like so:
Processor | Athlon 64 3800+ 2.4GHz |
System bus | HT 16-bit/800MHz downstream HT 16-bit/800MHz upstream |
Motherboard | Asus A8V |
BIOS revision | 1006 |
North bridge | K8T800 Pro |
South bridge | VT8237 |
Chipset drivers | 4-in-1 v.4.51 ATA 5.1.2600.220 |
Memory size | 1GB (2 DIMMs) |
Memory type | Kingston HyperX DDR SDRAM at 400MHz |
CAS latency | 2 |
Cycle time | 5 |
RAS to CAS delay | 2 |
RAS precharge | 2 |
Hard drive | Seagate Barracuda V ATA/100 120GB |
Audio | Integrated |
Graphics | Radeon 9800 XT 256MB AGP Radeon X800 Pro 256MB AGP Radeon X800 XT 256MB AGP GeForce FX 5950 Ultra 256MB AGP GeForce 6800 128MB AGP GeForce 6800 GT 256MB AGP GeForce 6800 Ultra 256MB AGP GeForce 6800 Ultra “Overclocked” 256MB AGP |
OS | Microsoft Windows XP Professional |
OS updates | Service Pack 2 RC2, DirectX 9.0c |
We used NVIDIA’s ForceWare 61.77 drivers for all of the GeForce cards, and we used ATI’s CATALYST 4.9 beta drivers, intended for DOOM 3, with all the Radeon cards.
The test systems’ Windows desktop was set at 1152×864 in 32-bit color at an 85Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests.
We used custom-recorded demos for testing. You can download both trdemo2 and trdm1.
If you have questions about our methods, hit our forums to talk with us about them.
Single player
For you bar chart people, I’ve done four separate bar charts showing the cards’ performance at different resolutions, from 640×480 up to 1600×1200. After that, I have all the info consolidated on to a single line graph with a data table, where each card’s scaling across different resolutions is more apparent.
There’s an anomaly at 640×480, and I’m not sure what to make of it. The GeForce 6800 is much faster than the 6800 GT or 6800 Ultra cards. I used the same settings for all the cards, and I re-ran the tests multiple times to confirm the results. I’m not quite sure why the 6800 is so much faster than its siblings. It’s possible the difference is related to the fact the 6800 has only 128MB of memoryperhaps the game was dynamically enabling some form of texture compression or the like. However, all of the seemingly relevant variables related to texture compression were set to “0”, so I am at a loss. Things do snap back to normal at 1024×768 and above.
The Radeon X800 cards start out well with strong scores at 640×480, when they’re not limited by fill rate (or pixel-pushing power). As we move up to higher resolutions, though, the Radeon X800 cards’ performance drops much quicker than the GeForce 6800s. Amazingly, the $299 GeForce 6800 is within five frames per second of the $499 Radeon X800 XT Platinum Edition once we reach 1600×1200. The $399 Radeon X800 Pro trails by miles.
These numbers suggest that ATI’s usual methods for boosting effective fill rate, such as occlusion detection and Z data compression, may not be working effectively in DOOM 3. We’ve seem similar scaling problems in older OpenGL games on the X800. One wonders whether ATI’s OpenGL drivers don’t fail to make use of the Radeon X800’s full capabilities here.
The older cards, the Radeon 9800 XT and GeForce FX 5950, are neck and neck in DOOM 3. I half expected the GeForce FX to do relatively well in DOOM 3, but it hasn’t turned out that way.
Single player with antialiasing
With 4X antialiasing enabled, the Radeon X800 cards do relatively better. Still, at 1600×1200, the GeForce 6800 GT outruns the Radeon X800 XT PE, and the Radeon X800 Pro just falls behind the GeForce 6800. That means NVIDIA’s cards offer equivalent performance for about $100 less than ATI’s.
In the battle of the geezers, the GeForce FX 5950 opens up a slight lead over the Radeon 9800 XT when AA is turned on.
Deathmatch
Deathmatch is pretty much the same story as the single-player game. Overall, the scores are a little bit lower, but the race between the various cards is about the same.
Deathmatch with antialiasing
Yet more of the same with antialiasing enabled. The Radeon X800 Pro looks mighty weak compared to its direct competitor, the GeForce 6800 GT.
Image quality
We usually like to take our screenshots from a game with a minimum of image processing, but DOOM 3 is so… very… dark… and the game’s gamma and brightness settings don’t seem to affect the output of its screen dump routine. As a result, I’ve done a slight gamma correction to the images that follow, consistent across all of the images.
The pictures below are cropped from a 1024×768 source screen in the game. What you see on this page and the next are low-compression JPEGs. You can click each image to open a window with a lossless PNG version of the image.
These shots were all taken at High Quality, just as we tested, with 4X antialiasing enabled. 8X anisotropic filtering is active in High Quality mode, as you can tell by the clarity of textures in the images below. This first shot should show texture filtering quality, especially on the floor.
I’ll admit I’m hard pressed to detect meaningful differences between the images with the naked eye. That’s a good thing, I suppose. All the cards produce decent output.
Image quality
This next image is a little more complex. You can see the amazing rock walls of this room. As I said, I suspect normal maps (which are just fancy bump maps) are in use on the walls. Note that I’ve included the 128MB GeForce 6800 separately from the 256MB GeForce 6800 GT, so we can see if there are any image quality differences between the cards as a result of the 6800’s smaller amount of on-board memory.
Again, all the images look excellent, and I’m hard pressed to see weaknesses in any of the cards’ output. Perhaps you all can find something, though. That’s why we take the screenshots.
Well, we’ve seen some striking performance differences and practically no substantial variances in image quality between the various cards we’ve tested in DOOM 3. In HQ mode, DOOM 3 looks spectacular on any of these cards.
If DOOM 3 performance figures heavily into your next video card purchase, you’ll want to think seriously about getting one of NVIDIA’s GeForce 6800 cards. The $299 vanilla GeForce 6800, with 12 pixel pipes, delivers playable frame rates in HQ mode at 1600×1200 in the single-player game, or at 1280×1024 with 4X antialiasing. In deathmatch, you may want to knock the resolution down a notch, but the GeForce 6800 is still a very good performer for the money.
The GeForce 6800 GT, meanwhile, is just about perfect for DOOM 3. It blows away the Radeon X800 Pro at the same price, and it’s fast enough that I fail to see the point of paying the premium for the GeForce 6800 Ultra. If you must have the fastest possible card for DOOM 3, though, the GeForce 6800 Ultra (especially the “overclocked in the box” flavor) is it.
If you’re interested in playing DOOM 3 but you anticipate playing lots of other titles as well, especially those based on DirectX 9, the Radeon X800 series is still a very solid choice. The Radeon X800 XT Platinum Edition is fast by any measure, even if it does fall behind a cheaper NVIDIA card in DOOM 3. Owners of the X800 XT PE won’t likely be troubled by getting 70 FPS at 1280×1024 in DOOM 3 deathmatch, even if they aren’t tops in bragging rights. Radeon X800 Pro owners, however, will have to settle for 1024×768not great for a $399 card.
The two older cards in our comparo essentially tied, with the GeForce FX 5950 Ultra taking a slight lead over the Radeon 9800 XT most of the time. Neither card is good for much over 1024×768 in HQ mode, though, and I still can’t recommend dropping upwards of $300 for a graphics card based on the GeForce FX series of GPUs. The 5950 may perform well enough in DOOM 3, but it has pixel shader performance and image quality issues in other titles. The Radeon 9800 XT is a more well-rounded alternative with almost the same DOOM 3 performance. The 9800 XT faces stiff competition all around, however, from the vanilla GeForce 6800. We’ll have more to say about that soon.