Our own Tuncer Deniz participated in a conference call with Tony Tamasi, director of product management for NVIDIA, yesterday afternoon. The subject in question? The Geforce 3 chipset, of course, set to appear on the Mac platform this March as a BTO option and as an add-in for G4 Macs in April. Questions were submitted by many Mac webmasters/authors, but we have condensed the extremely long conversation down to highlights for you. It is split in two parts, due to limitations of our news display system.

We're sure you can understand that longtime Mac users are a tad suspicious about such a powerful, high-end chipset debuting on the Mac platform -- after all, Macintosh systems have lagged behind in 3D hardware for nearly their entire history. Thus the question was asked, will the Mac version of this card be identical to the PC versions that will be released? Here is Tamasi's reply:

Tomasi: Yes, the clock speeds are exactly the same vs the clock speeds for
the PC product which was exactly the same case for the GeForce 2 MX product
we delivered for Apple. We don’t deliver a sub-standard, below speed version
for the Macintosh. It’s exactly the same product: same speeds, same memory,
same everything.

Tomasi: One of the great things that we think about the Macintosh in terms of
graphics and their strategy is that they’ve embraced OpenGL. OpenGL is
really the industry’s only completely cross-platform graphics API. And in
fact developers that have embraced Open GL typically develop cross-platform.
A great example of that is Doom III with id Software. I think they’ve
shipped concurrently or near concurrently on the Macintosh. Every version of
their game in recent memory, and I know that John and id intends to do a
similar thing with Doom III, and that is, every feature that GeForce 3
enables he will enable on the PC platform as well as the Macintosh platform.
And because we fully embrace OpenGL, and every extension we implement on the
Macintosh side, Windows side, and actually even Linux side, developers can
develop for the GeForce 3 platform and get to every platform.

Now, content developed for the X-Box, because it is basically GeForce 3
content should port transparently. The only thing they really have to do is
the OS specific functionality that supports the Macintosh. The Macintosh
sound manager would be different, for example, than the X-Box sound
manager. The Macintosh device input would be slightly different than the Windows
input. But that is much simpler code to port as opposed to the core
engine code and the core graphics code.

Another issue that is specifically important to Mac gamers and developers is the issue of bus speed and AGP throughput. As Mac systems tend to have far fewer motherboard revisions (and no possible motherboard upgrades for existing systems) they have lagged far behind in memory bus speed and AGP/PCI speeds. Now that Apple's high-end systems have 4x AGP this is less of a factor, but with the PC world already discussing 8x and 12x AGP, will this continue to be a bottleneck to performance?

Tomasi: We view our job as solving the graphics problem. And to the extent possible,
we will try to put in all the graphics functionality for the graphics
problems into our processors. So that means that things like the vertex
programming and pixel programming and continuing to improve that
functionality and make it ever more powerful and flexible, so that maybe one
day we can truly run all of Renderman [ed note: Pixar's plug-in scripting language for shading 3D surfaces] on our graphics processor. That would
leave things for the CPU such as artificial intelligence, game code,
networking, game logic, running the operating system…all the things that are
general.

Fundamentally, we don’t want there to be a bottleneck to the creativity of a
game developer. If AGP or LDT or Hyperstransport or anything is getting in
the way, then we view that as a problem we need to solve. As in the same way
as memory…memory bandwidth is one of the key factors in determining graphics
performance. So we have to be increasingly clever in the way that we use it.
So in the case of GeForce 3, the lightspeed memory architecture is all about
that. We are using the same exactly the same speed memory that we were using
in GeForce 2 Ultra product. And yet we’ve achieved two times [greater]
in terms of performance because we’ve been increasingly clever with the way
we use memory bandwidth and I expect that trend to continue as well.

The problems that the graphics industry, and in particular NVIDIA, have been
improving performance at a rate far beyond the memory industry’s ability to
improve the speed of their memory sub-systems. And so the options you have
are to either be more clever or go much, much wider on your memory
interface.

Be sure to read on for part two!

GeForce 3 Conference Call, Part 2

GeForce 3, DOOM Engine Demoed in Tokyo

Search for other Mac games news stories or browse our Mac Games News Archive.