|
Site News | |
Our Black Friday section is now online! Click here to check it
out.
|
|
| Active Discussions
|
|
[an error occurred while processing this directive]
|
|
| |
Radeon 32MB DDR vs Voodoo5 5500
This review pits ATI's Radeon 32MB against 3dfx's Voodoo5 5500. Both are at similar price points (Radeon ~ $240, Voodoo5 5500 ~ $260), and promise excellent 3D performance. Which one should be in your next box? The answer's not quite that easy. First let's review some history.
ATI has been making 3D cards since they were better known as 3D "decelerators" (games would often run faster in software mode, than with hardware "acceleration"). Over time their cards have improved, and are among the best in terms of image quality. True, a Rage Fury was no match for a Geforce in terms of fps, but if you wanted quality, ATI was the way to go. They also have one of the best built-in DVD decoders available (their only real competitor in that category is Matrox). However, they aren't known for good driver support, especially when a product is first released.
3DFX started out with a goal to provide both the fps and quality that people wanted. Since there was no DirectX when they first started, they decided to create their own graphics API called Glide. The Voodoo card was the first true 3D "accelerator" (ie it was faster running with hardware acceleration, than using software rendering), with much help from Glide. At the time the Voodoo was released, Quake was the main game in town. 3DFX could not convince Id to rewrite Quake to use Glide, so OpenGL drivers were written. With an actual game to show off Voodoo's power, developers flocked to code games, many Glide-only. A Voodoo3 based card could often beat a much newer and faster Geforce card, if it had the advantage of using Glide (and image quality was superb, even if it wasn't 32 bit).
Radeon 32MB DDR Features (taken from http://www.ati.com/na/pages/products/pc/radeon32_ddr/index.html):
- Powered by the ATI RADEONTM graphics processing unit (GPU)
- 32MB of powerful double data rate (DDR) memory
- Featuring ATI's CHARISMA ENGINETM and PIXEL TAPESTRYTM technologies
- The most robust and integrated Transformation, Clipping and Lighting (T&L)
- Accelerates all of today's top 3D games
- Industry best DVD playback
- Supports 3D resolutions (32-bit color) up to 2048x1536
- AGP Universal bus (for AGP 2X/4X systems)
Voodoo5 5500 Features
(taken from http://www.3dfx.com/prod/voodoo/v5-5500-overv.html):
- 64 MB of Graphics Memory
- 667-733 Megapixels Per Second Fill Rate
- Real-time Full-Scene HW Anti-Aliasing (FSAA)
- T-Buffer�
- FXT1� and DirectX� Texture Compression
- 8-bit Palletized Textures
- 32-bit Rendering and Textures
- 24-bit Floating Point Depth Buffer (Z or W)
- 8-bit Stencil Buffer
- DVD hardware assist: planar to packed-pixel conversion
Here's the main differences you'll notice. First, the Radeon has half the RAM of the Voodoo, however it's DDR RAM which gives it a performance boost. Second, the Radeon has Hardware T&L while the Voodoo5 has FSAA. For games that support it, Hardware T&L will make a bigger difference in image quality. However very few games support it right now (and don't take full advantage of it) though FSAA can be used for any DirectX, OpenGL, or Glide game past and present. If you upgrade every couple of years, then Hardware T&L may be more important to you (as later games will support it heavily). If you upgrade every 6 months or so, then the Voodoo5's FSAA will appeal to you as you can use it immediately. Lastly, both advertise hardware DVD decoding, though only the Radeon can do true DVD decoding.
Finally ATI has made a card for the fps gamer in the form of the Radeon. And for you non-glide users, 3DFX has again made a card that can compete with the big boys in both fps and quality (and with 32 bit color finally!). So you ask again, which one of these two should be in your next box? And why shouldn't you consider a Geforce anyways? For starters, this review is aimed towards gamers who insist on the best in both fps AND quality. If you're a pure fps kind of guy, then run buy the most expensive Geforce you can afford. But if you're like me and worry about the actual image quality, then you're probably leaning more towards a Radeon or Voodoo5 anyways. Now on to the testing....
The test system:
700@933 Coppermine (100% stable)
Abit SE6 motherboard (Intel i815e chipset)
128MB Crucial PC133 CAS3 RAM
2 30gig 7200rpm Maxtor drives (one with WinME, other with Win2K)
Sound Blaster Live!
I tested both cards under Windows Millennium and Windows 2000 Pro. Both installed very painlessly. I simply uninstalled the previous drivers, added the card in, rebooted, installed the newer drivers, and rebooted once more. I then started the actual benchmarking. First I'd run 3DMark2000, reboot, run Video2000, reboot, then Quake3 (to ensure that running one test beforehand would not interfere in any of the benchmarks). Only essential programs were running (such as Systray and Explorer). For 3DMark2000, D3D Hardware T&L optimization was used in Windows Millennium, and Intel(R) Pentium(R) III optimization was used in Windows 2000 (none of the benchmarks would complete with Hardware T&L enabled). For Quake3, demo 001 was used. Ok, the moment you've all been waiting for, here's the results:
Radeon 32MB DDR
Windows Millennium
| 800x600 16bit | 800x600 32bit | 1024x768 16bit | 1024x768 32bit | 1280x960 16bit | 1280x960 32bit | 1280x1024 16bit | 1280x1024 32bit | 1600x1200 16bit | 1600x1200 32bit |
| 3DMark2000 | 6121 | 5809 | 4661 | 4399 | N/A | N/A | 3379 | 3086 | 2556 | 2037 |
| Game 1 fps | 88.7 | 86.9 | 64.8 | 63.3 | N/A | N/A | 45.0 | 42.6 | 33.1 | 31.3 |
| Game 2 fps | 75.5 | 71.0 | 64.6 | 61.7 | N/A | N/A | 51.9 | 46.5 | 40.4 | 26.8 |
| Quake3 | 102.1 | 100.5 | 75.4 | 72.1 | 48.0 | 46.4 | N/A | N/A | 34.2 | 32.6 |
Video2000:
2340 Video Marks
Quality 1079, Performance 781, Features 480
Windows 2000 Professional
| 800x600 16bit | 800x600 32bit | 1024x768 16bit | 1024x768 32bit | 1280x960 16bit | 1280x960 32bit | 1280x1024 16bit | 1280x1024 32bit | 1600x1200 16bit | 1600x1200 32bit |
| 3DMark2000 | 4832 | 4701 | 4510 | 4449 | N/A | N/A | 3601 | 3331 | 2753 | 749 |
| Game 1 fps | 63.9 | 62.3 | 62.5 | 61.6 | N/A | N/A | 47.8 | 45.8 | 35.4 | 11.1 |
| Game 2 fps | 60.1 | 59.5 | 59.9 | 59.2 | N/A | N/A | 35.5 | 51.4 | 45.4 | 10.2 |
| Quake3 | 83.0 | 77.6 | 69.7 | 61.2 | 47.2 | N/A* | N/A | N/A | 33.8 | 28.9 |
*Quake3 would run at 1280x960 32bit, however graphics were so corrupted that a number was not discernable. Graphics anomalies were also present at 1600x1200 32bit, however a number was obtainable.
Video2000:
2408 Video marks
Quality: 1066, Performance: 810, Features: 532
Voodoo5 5500
Windows Millennium
|
800x600 16bit | 800x600 32bit | 1024x768 16bit | 1024x768 32bit | 1280x960 16bit | 1280x960 32bit | 1600x1200 16bit | 1600x1200 32bit | |
3DMark2000 | 4682 | 4390 | 4473 | 3516 | 3744 | 2512 | 1723 | N/A |
|---|
| Game 1 fps | 60.6 | 59.7 | 60.1 | 50.1 | 52.7 | 34.6 | 22.1 | N/A |
| Game 2 fps | 57.7 | 56.9 | 57.5 | 51.1 | 53.6 | 38.6 | 27.0 | N/A |
| Quake3 | 96.9 | 90.1 | 91.2 | 65.6 | 65.2 | 41.1 | 39.9 | 25.2 |
Video2000:
2283 Video marks
Quality: 840, Performance: 891, Features: 552
Windows 2000 Professional
| 800x600 16bit | 800x600 32bit | 1024x768 16bit | 1024x768 32bit | 1280x960 16bit | 1280x960 32bit | 1600x1200 16bit | 1600x1200 32bit | |
3DMark2000 | 4435 | 4240 | 4179 | 3544 | 3531 | 2584 | 2948 | 2015 |
|---|
| Game 1 fps | 56.1 | 55.9 | 56.8 | 50.3 | 49.7 | 36.3 | 40.8 | 27.7 |
| Game 2 fps | 55.0 | 54.7 | 55.4 | 50.6 | 50.9 | 39.6 | 44.8 | 30.8 |
| Quake 3 | 84.1 | 69.1 | 69.4 | 53.1 | 50.3 | 37.1 | 33.9 | 22.5 |
Video2000:
2169 Video marks
Quality: 820, Performance: 797, Features: 552
Both faired very well in WinME. You can tell the Radeon was designed with 32bit color in mind, as scores in 16bit and 32bit in the same resolution were very close to each other. With the Voodoo5 however, a 32bit score was similar to a 16bit score of the next resolution up. So it becomes a question of, would you rather run at 800x600 32bit or at 1024x768 16bit? Win2K was a different story however. ATI is still struggling with their drivers. At higher resolutions it appears to actually beat the Voodoo5, however the benchmarks had a lot of "jerking" in them which I believe affected the scores (ie a demo would run around 20fps, jump suddenly to above 70 and then drop right back down). Anomalies were present in Quake3 at higher resolutions in 32bit mode. 3DMark2000 would lock up if I switched from a 16bit test to 32bit (I ended up having to do all 16bit tests, rebooting, then running the 32bit tests). Their new beta drivers fix some of these problems, however since they're still in beta I can't benchmark those.
My conclusion? If you play most or all of your games in Windows98 or Millennium and don't have any Glide-only games, the Radeon is definitely for you. However if you want to use Win2K for gaming (it is, after all, much more stable than Win9x), or have some Glide-only games, I'd stick with the Voodoo5. It has much better Win2K support, and you can still play all your Glide games. You wouldn't be disappointed with the Radeon's performance however, and if they get better driver support, it will give the Voodoo5 a run for its money. If ATI's beta drivers are any indication (I've used the 3054 drivers in WinME and the 3035 drivers in Win2K) their next official drivers will make this card a strong contender in both WinME and Win2K.
Rating:
Radeon 32MB DDR: 9 in WinME, 5 in Win2K
Voodoo5 5500: 8 in WinME, 8 in Win2K
Questions? Ask in the
forum or
email me.
For the Privacy Policy, click
here. | |
| |
|
