3dfx quietly stole the limelight once again at this year’s fall Comdex in Las Vegas. From the wax museum in the Venetian, 3dfx announced the first implementation of the Voodoo Scalable Architecture with the VSA-100. We all expected a fill rate monster out of 3dfx and the products they announced based on the VSA-100 managed to fulfill every last one of our expectations.

The VSA-100, as you can probably already gather, is the basis for quite a few products. Much like the Voodoo3 was the core behind the Voodoo3 graphics cards, the VSA-100 is the heart and soul of 3dfx’s newly announced Voodoo4 and Voodoo5 products. Let’s first start with the VSA-100 specifications:

The VSA-100 will operate at somewhere between 333 and 367 megapixels/s, putting the actual clock speed at somewhere in the 166MHz to 183MHz range while rendering two pixels per clock. If you recall, this is essentially a Voodoo3 3000 or a 3500.

The VSA-100 adds support for 32-bit color rendering, 32-bit textures, 32/24-bit Z & W, and an 8-bit stencil buffer. Furthermore, the VSA-100 can also render two single-textured pixels per clock or one dual-textured pixel per clock. Support for 2K x 2K textures has now been implemented into the VSA-100, thus the VSA-100 offers essentially everything the Voodoo3 lacked and was criticized for tremendously.

The chip is an AGP 4X part, with support for AGP 2X, AGP 1X and PCI operating modes. In spite of this the VSA-100 does not support AGP texturing. 3dfx still feels that AGP texturing is not truly beneficial and thus there is no reason to pursue support for it with their products. The chip itself is composed of 14 million transistors, a little more than half the count of the GeForce, and is manufactured on an enhanced 0.25-micron, 6-layer metal process. The "enhanced" 0.25-micron process just means that it takes advantage of shorter gate lengths, which allow for faster switching thus allowing for higher frequencies and greater yields at those frequencies. According to 3dfx, the 0.18-micron process is not a mature one and thus they felt that they would achieve higher yields on a more mature 0.25-micron process. For reference purposes, the only true 0.18-micron graphics chip available is the mobile Savage MX from S3, with the Savage 2000 being a hybrid 0.18/0.22-micron solution.

The VSA-100 supports all T-Buffer effects, Full Screen Antialiasing, FXT1/DXTC texture compression and all of the other features 3dfx has been talking about for the past few months. For more information on those technologies read our coverage of them here.

The VSA-100 supports anywhere from 4MB to 64MB of memory per chip, whose clock is synchronized with the core clock, just like the Voodoo3. The memory bus is 128-bits wide and will offer anywhere between 2.7GB/s and 2.9GB/s worth of memory bandwidth depending on the final clock speed of the chip. The excellent 350MHz RAMDAC of the Voodoo3 is carried over to the VSA-100.

From the above description the VSA-100 doesn’t appear to be much more than a Voodoo3 with support for a few new visual features and 32-bit color rendering support, but the chip’s support for up to 32-way SLI scalability (hence the name Voodoo Scalable Architecture) is what truly defines it and sets it apart from the Voodoo3.

3dfx was the first company to pioneer the linking of multiple chipsets for the consumer market with the Voodoo2, using a technique known as Scan Line Interleaving, or SLI for short. On the Voodoo2, two boards could be linked together, theoretically doubling the performance. Each card rendered half the image by rendering alternate scan lines of the image, hence the SLI name. One company, Quantum 3D, went as far as to produce a single board Voodoo2 SLI configuration. In fact, for high-end development systems and arcade machines, they offered a setup with four of these boards linked together for an 8-way Voodoo2 SLI configuration. However, since the Voodoo2, 3dfx has not released another product that supported SLI, until now…

The VSA-100 architecture allows for the long awaited return of SLI to 3dfx products. There is support for up to 32-way SLI, which Quantum 3D will likely produce for extremely high-end applications. Each chip has its own 128-bit wide interface to memory, meaning that, unlike NVIDIA’s GeForce, the VSA-100 will not be limited by memory bandwidth. Although the memory is setup to be completely shared, texture data has to be repeated for each VSA-100 chip. This is typical of all current multi-chip solutions, such as 3dfx’s own Voodoo2 and ATI’s Rage Fury MAXX. However, the VSA-100 does have the advantage of FXT1/DXTC texture compression in hardware, which will minimize the amount of wasted space.

The SLI on the VSA-100 is actually a little bit different from the setup used on the Voodoo2 and is in fact quite a bit improved. The first change is that the number of scan lines that each chip renders can be adjusted from 1 to 128 lines depending on a number of factors, including number of chips, resolution, fillrate required, etc. This is something 3dfx will have to tweak for optimal performance. They have yet to determine whether this will be something fixed in the driver or if it will dynamically adjust depending on the load or the application involved. Since it is definitely software programmable, chances are that it will be a variable that end users will be able to tweak through the registry or any number of tweaking utilities.

The other big change is the SLI interface. On the Voodoo2, the interface was a multi-board, analog solution that required very strict timing control and identical boards. Furthermore, the analog signal was prone to degradation as it passed over a short cable inside the PC. VSA-100 features a 30-bit digital interface between the chips, which are all on one board this time. There are 28-bit data bits, a data valid bit, and a clock.

One advantage of the Voodoo2’s multi board SLI setup is that you could buy one Voodoo2 board to get started and add a second one at a later date to nearly double performance. Of course, this time that’s not possible as all chips must be on one board.

Not only did 3dfx announce the VSA-100 architecture, but they also introduced the board level products that will be using the technology. The product line from 3dfx includes boards from 1 - 4 chips, with Quantum 3D once again filling in the high end with 8 - 32 chip solutions.

Both the Voodoo4 and Voodoo5 lines were announced Monday and both are based on VSA-100 chips. The distinguishing factor is that the Voodoo4 series is based on a single VSA-100 chip, while the Voodoo5 series refers to a board with multiple chips.

• Single 3dfx VSA-100
• 32MB memory
• 2 pixels per clock rendered
• 333-367 megapixels/s
• $179 US

The Voodoo4 4500 is targeted at "mainstream consumers" and is thus the more cost effective single VSA-100 product. The Voodoo4 4500 will be available in both PCI and AGP versions at $179. Once again, note that T-buffer effects are not enabled on single chip products. Think of the Voodoo4 4500 as the Voodoo3 3000 with 32-bit rendering, large texture support, and 32MB of memory. Expect performance similar to the Voodoo3 3000, but with greatly enhanced image quality thanks to these new features.

• Dual 3dfx VSA-100 SLI
• 32MB memory
• 4 pixels per clock rendered
• 667-733 megapixels/s
• Real-time full-scene anti-aliasing
• T-Buffer digital cinematic effects
• $229 US

The entry level for the Voodoo5 line, the 5000 PCI, is actually just $50 more than the Voodoo4 4500. You get quite a lot for that $50 though, including double the fillrate and T-Buffer effects thanks to a second VSA-100 chip. However, the 32MB of memory is slightly less effective here since texture data will be duplicated in memory thanks to the dual chip configuration. Also note that the Voodoo5 5000 is PCI only at this point. Performance is theoretically double the Voodoo4 4500 without full scene anti-aliasing enabled, or approximately the same as the 4500 with it enabled.

• Dual 3dfx VSA-100 SLI
• 64MB memory
• 4 pixels per clock rendered
• 667-733 megapixels/s
• Real-time full-scene anti-aliasing
• T-Buffer digital cinematic effects
• $299 US

The first AGP card in the Voodoo5 line up is the 5500, which is much like the 5000 PCI with an additional 32MB of memory and an AGP interface. The increased bus transfer rate and onboard RAM serve to enhance performance as game complexity increases.

All this will cost you $50 more than the 5000 PCI, primarily to pay for the additional RAM. If RAM prices drop, expect the cost difference between the boards to also drop.

• Quad 3dfx VSA-100 SLI
• 128MB memory
• 8 pixels per clock rendered
• 1.33 - 1.47 gigapixels/s
• Real-time full-scene anti-aliasing
• T-Buffer digital cinematic effects
• $599 US

The Voodoo5 6000 is definitely the mother of all graphics cards with easily the highest fillrate of anything available at its launch. With 128MB of RAM, texture space should not be a problem as this card will have more RAM than many systems have. 3dfx is shooting for 85 fps at 1024x768x32 in Quake 3 with full scene anti-aliasing enabled - not too shabby.

The price is quite high at $599 and is clearly targeted at the hardcore gamer. We know some people will buy it because quite a few people paid about $600 for a Voodoo2 SLI setup when it was released. The 6000 AGP will feature an external 100W power supply that hooks up to the board via a connector on the cards back plate.

Although the chips are identical, all Voodoo4 series products will feature just a heatsink, while the Voodoo5 series gets a heatsink and fan. The heatsink is sufficient, but this just gives Voodoo5 users a little more piece of mind as 3dfx received quite a few complaints about the Voodoo3 overheating due to the lack of a fan. It turns out that the majority of these problems were traced back to the motherboard providing insufficient power to the graphics card.

AnandTech has noted this issue in the past as well. With a multiple chips drawing 13-16W each, the problem is only exacerbated. Even AGP Pro’s enhanced power handling ability is not sufficient. Fortunately, 3dfx has come up with a fairly elegant solution. All dual chip boards will feature a power connector, which you can see if you enlarge the picture of the 5500 AGP, that connects to a standard 4-pin drive connector.

The quad chip board, the 6000 AGP, will use an external 100W power supply that 3dfx has named "Voodoo Volts." Power will still be a problem for single chip boards on any motherboards that had trouble with a Voodoo3 or TNT2 Ultra. Just to be clear on this issue – this is the fault of the motherboard manufacturers not meeting the AGP power guidelines and not the fault of the video card manufacturers.

Notice that 3dfx has not forgotten the PCI market, which is actually quite large for the add in board market where 3dfx is the leader. So many new PC’s are coming with i810 chipsets or other integrated graphics solutions that have no AGP slot on the motherboard. The CPU’s on these systems may be quite powerful, but the graphics solution lacking somewhat. A large number of these consumers will eventually want to upgrade to much more powerful graphics solutions and 3dfx is looking to capture a large portion of this market as NVIDIA and Diamond/S3 are largely ignoring PCI products. The popularity of PCI add in boards is evidenced by the PCI Voodoo3 sales, which are almost the same as the AGP version.

You may have noticed in the shot of the 6000 AGP an Intel chip onboard. This is an AGP to AGP bridge that is necessary to prevent loading down of the AGP bus. 3dfx found that they could work around the loading issue with just two chips, but not with four. Hence the lack of said chip on the Voodoo5 5000/5500.

Also note that 3dfx has played around with the T-Buffer and determined it would not be an effective solution on a single chip board. The reasoning is that for the effects to look good, four samples are necessary, but a single chip can only do two at once. Furthermore, a single chip simply does not have the raw power, in terms of fillrate, to handle any of the T-Buffer effects. When probed, 3dfx did say that it may be possible to enable such effects through a registry hack, but there would be no reason to do so as the result would not look that good and would run quite slow. Thus, do not play on using T-Buffer on a Voodoo4 4500 card.

3dfx did show a demonstration of T-Buffer effects using Quake3. Id has given 3dfx the source code in order to implement such effects and a patch is planned for release when boards are available. The effects they had implemented so far were motion blur on objects and players as well as a "drunken sailor" effect when a player received damage. The latter effect was similar to blurry vision. The final effects that will be implemented in the patch have yet to be determined, but the demos clearly show the potential of T-Buffer effects in situations other than cinematic sequences.

Click to enlarge

Click to enlarge

Full scene anti-aliasing was of course shown as well, but as we’ve noted before, FSAA will not require any game patches or special programming. We were told, however, that FSAA will effectively half the fillrate when enabled. With so much raw power, it shouldn’t matter too much.

The current stance on Transform and Lighting from 3dfx is that it is one of the important steps forward in 3D gaming, but it is simply not necessary now or even 6 months from now. With relatively few games offering truly compelling features that require T&L, 3dfx may well be correct here. Although they wouldn't come righ out and say it, they strongly suggested that we could expect the next 3dfx part to have T&L support.

In case you’ve forgotten, 3dfx told us in July that products featuring T-Buffer would be available for the holiday season. Back in September, rumors started spreading around the web, and AnandTech confirmed, that the release was delayed until the first quarter of 2000. It turns out that 3dfx was having trouble with the back end design of the chip, literally the last stage of design before it is sent off to the fabs.

What it all boils down to is that 3dfx does not actually have working silicon back at this point in time. All they could tell us is that they are "eagerly awaiting" their first samples. At Comdex, they were still showing T-Buffer demos using Quantum 3D’s 8-way Voodoo2 SLI setup.

The delay means that by the time VSA-100 is available, NVIDIA will be preparing their spring refresh, known internally as NV15. This GeForce follow up will most likely move to a 0.18-micron process that should allow for higher clock speeds. Whether or not this will enable NVIDIA to match 3dfx’s monster fill rate is impossible to know right now, but it does not seem likely. Of course, if NVIDIA is successful in evangelizing transform and lighting, 3dfx could be in big trouble.

AnandTech will of course have full details, including the usual plethora of benchmarks, on the Voodoo4 and 5 as soon as samples are ready from 3dfx.