Hands-On: NVIDIA GeForce4 Press Event, February 5, 2002

February 5, 2002
NVIDIA GeForce4 Press Event

At Macworld Tokyo 2001, Steve Jobs unveiled the NVIDIA GeForce3, well before the rest of other PC manufacturers and even NVIDIA itself. At Macworld San Francisco last month, Steve Jobs unveiled the NVIDIA GeForce4, once again before other PC manufacturers and NVIDIA, who officially launched the GeForce4 family tonight. When asked how Apple beat everyone else again, Jen-Hsun Huang, President & CEO of NVIDIA laughed, pointed to Jon Rubinstein and said that Apple has very talented engineers and awesome time to market.

There’s no doubt that Apple and NVIDIA have a great relationship as evidenced by Jon Rubinstein and Phil Schiller’s presence at tonight’s GeForce4’s coming out party. Apple has the world’s fastest personal computer and NVIDIA has the world’s fastest graphics processor. (Apple announced the availability of the GeForce4 Ti, the GeForce4 MX’s big brother, as a build-to-order option today.) Apple and NVIDIA also have a strong balance sheet.

Mr. Huang started the press conference by announcing that NVIDIA has shipped over 100 million processors and was the fastest to reach $1 billion in revenues making them the world’s largest semiconductor company that doesn’t fabricate their own chips. According to an industry report, they are the “most respected fabless company”.

Mr. Huang then began to review some of NVIDIA’s accomplishments last year. In front of a screen grab of the movie “Final Fantasy” that had been rendered in real-time, he reviewed the GeForce3 and its breakthroughs such as the industry’s first programmable shader (the part that draws textures). NVIDIA also powers Microsoft’s Xbox game console whose advanced technology was developed in just twelve months. He then invited Tony Tomasi, Senior Director of NVIDIA’s GPU (Graphics Processor Unit) Business to introduce the GeForce4 family.

The GeForce4 family consists of the GeForce4 Ti for the enthusiast, the GeForce4 MX for the mainstream and the GeForce4 Go for mobile PCs. (NVIDIA retains the GeForce2 for the value conscious although it will take a third party to bring such a product to the Mac.) The GeForce4 delivers two to four times the performance of previous GeForce generations. It introduces Accuview Antialiasing (makes jaggies smooth) without a performance hit and nView Display Technology that finally appears to give PC users running WindowsXP a Mac-like multiple monitors experience.

Mr. Tomasi started off with the GeForce4 MX. He showed a slide of the GeForce4 MX chip schematic to highlight the Accuview Antialiasing, nView Display Technology and other advancements over previous generations. He then showed a graph to demonstrate how GPU speed is rapidly outpacing Double Data Rate (DDR) memory speed. NVIDIA began to solve this problem with the GeForce3 when they introduced their Lightspeed Memory Architecture (LMA). The GeForce4 has the second-generation LMA II plus a 128-bit DDR interface to double available memory bandwidth and increase efficiency. It also improves upon NVIDIA’s “Z-occlusion culling” which basically means the GeForce4 doesn’t waste bandwidth drawing pixels that you don’t see; waste typical with older graphics technology. The result is that the GeForce4 is 50% more efficient than the GeForce3 and 100% more efficient than the GeForce2 in these areas.

Mr. Tomasi moved on to a demo using Windows XP and Windows Media Player on a 2GHz system to demonstrate the aforementioned multidisplay technology. The GeForce4 can drive 2 displays, even television (not necessarily standard). In perhaps another first for Windows XP users, Mr. Tomasi showed off the GeForce4’s ability to do transparency that reminded me of Apple’s Quartz in Mac OS X.

With performance fully covered, Mr. Tomasi moved on to the second most sought after feature by NVIDIA’s customers: anti-aliasing (AA). The GeForce4’s Accuview AA is “so good you just leave it on”. In fact, the GeForce4 is faster with AA on than the GeForce3 with AA off! Mr. Tomasi then went into a series of demos.

The first, entitled “Bugs”, was of futuristic ants that highlighted the GeForce4’s second-generation “Z-occlusion culling”. The second demo was “Chameleon” from last year’s GeForce3 launch but looking even better on the GeForce4. Thirdly, was a humorous demo of a lifelike tornado destroying a trailer park, complete with flying cows. This demonstrated the ability to offload all of the graphics processing to the GPU to leave the CPU free to calculate the destruction of the trailer park. Lastly, Mr. Tomasi showed a marketing presentation for the GeForce4 using an unnamed streaming 3D web plug-in as an example of why everyone needs a GPU. Then, as if the GeForce4 MX was not awesome enough, it was time to behold the “most powerful GPU in the universe”.

Mr. Tomasi unveiled the GeForce4 Ti that has twice the performance of the GeForce3 and four times the performance of the next closest non-NVIDIA competitor (which I speculate is ATI’s Radeon 8500). Like Apple’s hardware, the GeForce4 AGP board is impressive to look at. Demos for the GeForce4 Ti included “Squid” which was a highly detailed squid moving over a seaweed bed and through a kelp forest. He showed the near-solid wireframe, the skeleton of 3D graphics, and pointed out that the GPU had left the CPU free to control the fish schools with artificial intelligence algorithms. All of this was made possible by the GeForce4’s second-generation nfiniteFX Engine, dual vertex shaders, advanced pixel shaders and other previously mentioned technologies. Mr. Tomasi mentioned that NVIDIA took the technology from the Xbox and advanced that significantly.

In the final series of demos, we were shown a computer generated nature scene that was almost close to Pixar’s “A Bug’s Life”. It included a tidepool that demonstrated improved bump mapping, a subtle feature that brings computer-generated images closer to indistinguishable realism. “Grace”, a demo of a dancer on water with bubbles coming out of her hands showed off the GeForce4’s ability to do many calculations and leave the CPU free to create a realistic cloth texture, a feat not trivial before. For the grand finale, “Wolfman”, a demo that elicited holy explicatives and applause from the audience, has upstaged “Zoltair” from last year’s GeForce3 launch for the awesome amount of polygons used to create the werewolf-like creature. It also sported realistic fur, another impressive feat for real-time rendering without Pixar-class hardware. On a side note, Mr. Tomasi pointed out that the group at NVIDIA that put together the demos that push bleeding edge technology have a pretty cool job.

After the demos, Mr. Tomasi showed a collage of games that take advantage of the GeForce4. He then showed a slide of NVIDIA’s partners that included Apple. As if to take a page from Apple, there was a video of NVIDIA’s partners saying how great NVIDIA and the GeForce4 is. Finally, Mr. Tomasi provided some interesting perspective on the GeForce4 power:

    - 1 trillion operations per second
    - 10x SGI Infinite Reality geometry processing power ($500 million each a while back)
    - Can render more than 100 Jurassic Park dinosaurs in real time
    - 8 GeForce4’s can render more than all of the 3dfx Voodoo1’s ever shipped (NVIDA acquired 3dfx)
    - More than the sum of all fp power on earth in 1985
    - Which is over 384 million IBM PC AT’s
    - Which would have cost $2 trillion in 1985
    - Which was more than the GNP of the US that year

Mr. Huang joined Mr. Tomasi on stage for questions. Some interesting tidbits gleamed from the Q&A session are that the GeForce4 is more advanced than the Xbox. This leaves the possibility open that if Microsoft ever ships Halo for the Mac, it can look better on the Mac, especially since it can be played at higher resolution than what is viewable on a television. When asked what NVIDIA has in store next, Mr. Huang commented that the GeForce5 would not be called the GeForce5 and he wished that he could share some details about what they have for future generations because they’re even more mind blowing. An NVIDIA employee on the stage noted that they’re striving to make reality indistinguishable on screen and that it will be at least another ten years before we start to get close. Mr. Huang stated that NVIDIA is advancing their technology as fast as possible and reiterated their rapid release cycle from last year that they fulfilled by releasing the GeForce4 tonight.

After the press conference, I had the chance to talk with some Apple representatives working on Apple’s graphics hardware. They were all very enthusiastic about their relationship with NVIDIA and agreed that there is room for third party companies to make upgrades for AGP 4x Macs. (Apple’s own GeForce4 Ti upgrade will be out this Spring although I was unable to get anyone to throw out a date.) Now that Apple has the best GPU, they acknowledged that they’re working on solutions for offloading audio processing from the CPU, which is an option sorely lacking in Mac OS X since the only consumer sound card is Creative’s outdated SoundBlaster Live.

At the reception, Apple’s kiosk was busy and I also ran into several people very excited about the Mac. Apple definitely had the best looking hardware; especially displays compared to the stock IBM LCD’s that the other companies were using. Perhaps if a future PowerBook uses the GeForce4 Go, I’ll be content to stop upgrading my desktop PowerMac completely.

Brian Nakamoto
Macs Only!'s West Coast Reporter

. . [Go back to Macs Only!'s Daily News]

[Back to top of page]



Click On Images Above!

Apple, Apple Logos, Macintosh, and Mac OS Logos are registered trademarks of Apple Computer, Inc. All other trademarks mentioned belong to their respective owners.

[Back to top of page]

Saw something? Send a tip or a correction

The archive ran on reader tips. What did you see, where, and do you want the credit? Something wrong on this page? Say so and it gets fixed.

Read by the editor. You get a copy by email. Never published without your say.

More in Reviews · This month in the archive