Summary: Earlier this week NVIDIA unveiled new details on its next-generation graphics architecture, codenamed NV30. Learn all of the latest details in our preview of this technology!
Accelerating the convergence of film and real-time rendering
Although summer is still young, things are definitely beginning to heat up in the 3D graphics market. Matrox kicked things off with its re-entrance into the high-end graphics market with its Parhelia graphics card. Then, just last week ATI unveiled its RADEON 9000 DirectX 8 board and its next generation RADEON 9700 graphics card. Not wanting to be left out of the party, NVIDIA has chosen to use SIGGRAPH to unveil new details on its next generation graphics core, which is taking place this week in San Antonio. Unlike an event like Comdex, SIGGRAPH is geared solely towards the professional graphics community. This makes it a natural fit for NVIDIA to unveil new details on its graphics hardware: the content developers and engineers that use NVIDIA’s hardware on a daily basis are all right there. Moving beyond nfiniteFX
The key message NVIDIA plans to deliver with the NV3x family is the cinema quality graphics effects it brings to the table. This is something we’ve been hearing from NVIDIA for quite awhile now, and ultimately reached reality when NVIDIA demonstrated a Quadro GPU rendering scenes from Final Fantasy: The Spirits Within at SIGGRAPH 2001 last summer. As impressive as that demonstration was, NVIDIA feels it has taken an even greater step forward in bringing the world of film to the PC with its next generation graphics technology. To highlight this, NVIDIA has chosen the moniker “CineFX” to brand the graphics engine of its next generation graphics processor. While NVIDIA hasn’t disclosed the complete details of its new CineFX architecture, a few morsels of information have just been revealed.
First off, we now know that NVIDIA has taken the programmable roots of the GeForce3/GeForce4 GPUs to another level with the CineFX architecture in NV3x. For vertex processing, new instructions are introduced such as logarithmic functions and high-precision trigonometric functions. In fact, the number of instructions supported by the CineFX architecture has jumped from 128 in nfiniteFX to 65,536 in CineFX. This gives programmers the ability to create even more effects while using simpler programming techniques. Do you recall the Zoltar demo used in the original GeForce3 launch? Several shaders had to be written to create the lifelike skin and bone structure Mr. Zoltar possessed. With CineFX, one shader can be written to accomplish the same task. SIDEBAR: My current work on Doom is designed around what was made possible on the original GeForce, and reaches an optimal implementation on the NV30. My next generation of work will be designed around what is made
possible on the NV30
-John Carmack, Lead Programmer, id Software
Pixel processing
Remember how John Carmack was disappointed with the pixel processing implementation in DirectX 8 and the GeForce3 GPU? With CineFX, NVIDIA has elevated pixel processing to the same level of importance as vertex processing. Higher color precision
Another new improvement introduced with NV3x is support for 64-bit and 128-bit color (16 or 32-bit floating point for each RGB component). As you can imagine, this results in images with more vibrant color ranges. A common example is fog. Chances are, if you’ve played lots of games with extensive use of fog or smoke, artifacts were frequently visible. With a wider color range these artifacts are eliminated. SIDEBAR: Pixar rendered Toy Story 2 in 64-bit color mode.
NVIDIA’s recent announcements give us a glimpse of what NV3x will be capable of, although there are still quite a few details that have not been announced. We haven’t been given clock speeds of the graphics core or memory subsystem, nor do we know any fill rate or other numbers that can be used as a metric of performance. Instead what we are given is a look at just what is possible with NV3x, and based on what we’ve seen so far it should match up very closely with ATI’s RADEON 9700. Other details that we do know is that NV3x will be built off TSMC’s 0.13-micron manufacturing process, will fully support AGP 8X, and supports high-speed DDR-II memory. In terms of availability, all we know is that NVIDIA has timed NV3x’s release with that of Microsoft’s DirectX 9 API. This puts NV3x somewhere in the fall timeframe, which, if ATI holds up to its RADEON 9700 release schedule, gives ATI a one to three month time-to-market advantage over NVIDIA. Speaking of the RADEON 9700, a lot of preliminary performance estimates have popped up over the web. All of these comparisons have pitted a prototype RADEON 9700 board against the current performance king, NVIDIA’s GeForce4 Ti 4600. It will be interesting to see what configuration final RADEON 9700 boards will ship in, and just how great a performance advantage RADEON 9700 possess when it is launched. We wouldn’t be surprised to see RADEON 9700 ship in 30 days, retake the performance crown from NVIDIA, and NVIDIA subsequently slash prices on its entire GeForce4 line. Once NV3x is released, NVIDIA could then restore its presence in the $400 high-end market. This is just one contingency plan NVIDIA may have in the works. We’ve also heard rumors that NVIDIA is hard at work on an NV25 successor with AGP 8X graphics. Whatever ends up happening, one thing is for sure, we haven’t seen this level of competition since around this same time last year. With Matrox and 3D Labs also playing in the same field (although to a lesser extent) things are even more interesting. This gives consumers choices in the marketplace, and that is definitely a good thing. SIDEBAR: Will NV3x retake the performance crown, or do you have your money on ATI? Voice your thoughts in the comments!
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| © Copyright 2003 FS Media, Inc. |