This article has been combined with the other two articles for one long resource, please see:

http://www.sudhian.com/index.php?/articles/show/3dfx_voodoo_5_6000_review/ 

This is Part II of our three-part series on 3dfx and potential Voodoo5 6000 performance.  Part I discussed the history of the company and can be found here, Part II will examine the V5 6K in-depth, and Part III will examine the Rampage technology that never was, as well as other explanations for how and why 3dfx died.

 Why Review the V5 6K?

 Some of you are undoubtedly asking why a major hardware site would take the time (substantial) and the cash (even more substantial) to review a three year-old product from a dead company.  It?s a fair question, and the answer sets the stage for the review itself and what we are and aren’t trying to accomplish.

 There’s no question that the Voodo5 6000, whatever else it may be, is outdated.  Its once-massive 128 meg of memory is now standard, its barely-DX7 compliant design is far surpassed, and its T-buffer technologies (save for FSAA) are useless.  It’s an anachronism, an out-performed piece of history, and there’s no chance of it reaching the performance levels of a GeForce4, to say nothing of a GeForce FX or a Radeon 9800.

 
The Voodoo5 6000 in its original incarnation--note the 2x2 card design.

 All of the above is true, yet stopping here would only tell half the story.  The Voodoo5 6000 may be an outdated relic today, but looking at the card one still gets a sense of just how high 3dfx was attempting to raise the bar.  The Voodoo5 6000 offers a fill rate of 1300 Mpixels per second, a number that puts it on par with GeForce4-era technology.  Its memory bandwidth of 10.4 Gb/second outstrips the GeForce3 and Radeon 8500?equaling, in fact, the GeForce4 Ti 4600. 

 Lest you get your hopes up, the numbers don’t tell the whole story here, and the V5 6000 is not GeForce4 competitive--that card employs far more advanced and efficient memory algorithms that existed when the Voodoo5 design was built, to say nothing of the card’s DX8 features.  The Napalm iteration of VSA-100 technology employed no memory bandwidth saving technology (though it was hoped Gigapixel might one day provide a software-based solution), and is not a match for such a rival.  Against the video cards of its time period, however, the situation may be different?and it is for this reason that we’ve chosen to benchmark 3dfx’s magnum opus against the video cards of its day, from the GeForce DDR to the GeForce 3, from Radeon to Radeon 7500.  (The GeForce2 MX, with performance very close to that of the GeForce SDR, stands in for its older cousin).

Why review the Voodo5 6000?  Because it?s the greatest "what-if" in video card history, and the question of what kind of performance it offered and what might’ve happened had it shipped has never been answered.  It?s the last symbol of a pioneer’s trailblazing in 3D accelerator history, the Titanic of its era, a product whose potential was never realized and who died before it got a chance to spread its wings.  On a more personal note it also cost me $850 out of my own pocket, which means I’m going to write about it and you’re going to read it.  

So join us as we take a trip back in time to the year 2000 (pausing along the way to grab our testbed and some new drivers) and have a look at how the last scion of 3dfx would’ve performed?and the impact it would’ve had.

next > next > next > next > next > next > next > next > next > next > next > next >

---

• Testbed and Drivers (0 of 12)
• Jedi Knight, Serious Sam (0 of 12)
• Quake 3 (0 of 12)
• D3D Performance Analysis (0 of 12)
• Expendable (0 of 12)
• Evolva (0 of 12)
• 3DMark 2001 (0 of 12)
• 3DMark 2000 (0 of 12)
• Benchmarking Methodology (0 of 12)
• Our card. (0 of 12)
• V5 6K Design Stages (0 of 12)
• Analyzing the Numbers (0 of 12)