Tuesday, September 21, 1999
This would be an interesting
strategy if they felt their product could not match 3dfx's raw
performance. This also would land them a lot of initial sales of their
cards before consumers knew what Nvidia's and S3's card are capable of
doing. We PLEAD that everyone wait until all the video cards from S3,
Nvidia and 3dfx are released and benchmarked so we can help find the
one that is best suited for your computer! The ONLY damage you can do
by waiting for the reviews is to make a better marketing decision! Do
not be led by hype to make a premature decision!
Q) During the recent
interview at Thresh's FS, Scott Sellers said the following, "It
appears as if the raster engine in the GeForce 256 is really just 2
TNT2 raster engines put in parallel (with some feature additions
added)." Given that statement, do you think it is possible that
what we are seeing in GeForce 256 is really a hybrid form of two
TNT2's merged?
A: Well, by my statement I was saying
just that - that it appears based on their spec sheets and
functionality that the GeForce's raster engine is likely 2 TNT2 pixel
pipelines put in parallel. Clearly, the geometry capability is brand
new for GeForce, but it appears as though the raster functionality is
virtually identical to the TNT2 raster engine (except it's obviously
doubled in width and they've also added cube environment mapping…)
Well, we have heard this speculation before. Is GeForce 256 really a
NEW product if you remove the GPU and a few other features? Not really
sure yet since we have not seen one. In fact, they were claiming a
launch date of any day now and no one outside of Nvidia has seen one.
Q) Creative Labs
admitted in a recent Annihilator preview that their GF256-based board
was going to lose to your next part in terms of fill rate. . . care to
elaborate on why they believe this?
A: Well, as we have not announced our
next generation product we do not know why Creative would say
something like this. We have, however, said that we believe our next
generation product will outperform the GeForce boards on fill-rate. I
guess maybe Creative just takes us on our word. Who knows….
It was very shocking to hear this news come from Creative. 3dfx has
not even revealed their fillrates yet. This could be the beginning of
damage control, or it could be that Creative feels that fillrate will
be irrelevant.
Whether this "defeat" in fillrate turns out to be a real
issue or not is unknown (until we can test the cards side by side and
find out) but usually a company will choose not to talk about features
from a competitor who's "specs" are not released. This is
definitely a deviation from policy in this industry and one that could
cost Creative popularity and sales.
Q) Now that NVidia and
S3 have introduced Transforms and Lighting (T&L) on their chips
many people have assumed that with the introduction of DX7, this
feature will suddenly become "mainstream on all games" and
"be REALLY fast especially in games like Quake". Is this
really true or is there more to the story with T&L?
A: Oh, I think there's a lot more to
the story than that. We have gone into details about this in other
interviews, but the reality is that both the GeForce and Savage 2000
appear to be very unbalanced parts (meaning they have focused on
geometry without substantially increasing 32-bit fill-rate). We
believe that fundamentally gamers are looking for a solution that
allows them to experience new technology immediately, right out of the
box, and not wait for the "promised" games to come down the
line.
To this end, we have focused on
fill-rate as being the number one problem to solve for this next
generation of 3D accelerators. The problem with the GeForce and
Savage2000 is that neither has done much to improve fill-rate, so just
like you can't run Quake3 today at 1024x768 with sustained 60 fps in
32-bit color, we don't believe you'll be able to do it on those
products either.
We have, instead, focused on delivering
very high frame rates at high resolutions and 32-bit color, as this is
what gamers have told us is very important. So instead of having to
wait 6-9 months for games to start taking advantage of geometry
acceleration, we're going to have a solution which dramatically
improves today's games in a substantial way, in both high resolution
fill-rate and overall image quality. No one else will be able to
deliver real-time full-scene spatial anti-aliasing either, and once
people experience this they'll never go back to the aliased rendering
of the GeForce and Savage 2000 products.
I must admit that this single issue has been my gripe with Nvidia for
a long time: releasing hardware with features that are unusable to the
consumer for quite some time. In the case of T&L, it does appear
that it will be 6-8 months before we see any support for these
features. Cube Mapping is like doing Anti- Aliasing in software. It is
a great feature to talk about but useless unless you think single
digit FPS are "acceptable" and this is at a resolution of
640x480!
I give Nvidia great respect for being the first company to release
32-bit color support. In some games the framerate hit was not that
large and the increased color depth looked good. However, trying to
run the game from any resolution above 800x600 meant hitting a
fillrate limit and my FPS dropped rapidly which in turn caused me to
go back to 16-bit color and usually Glide (for F4), and then I was
running my SLI V2 or V3 because of the huge FPS jump over TNT and
TNT2.
Here we are again. As expected we have T&L. Now, this is where I
have to make a split. Most of you out there are probably pretty
hardcore gamers, who admittedly upgrade at least twice a year and some
of you, more than that. Even IF 3dfx does not have support for T&L
we can count on them having it in the next product when we see
large-scale support for T&L.
[ Next Page ]
|
