The Big 3D Fight: Part II
Written: 11/23/99
Written by: Jellebee
   
GeForce
They are currently the only company to have delivered what they have
said they would deliver. They are the only company that are exciting the public. They are
the only company that are excited the developers. They are NVIDIA. And the excitement is
the GeForce 256.
It has to be said that all they ever promised was a card with
improved fillrate and T & L onboard. By relieving the CPU from the number crunching
calculations required for this process the speed of rendering the action onscreen has been
enhanced massively. One programmer mentioned it would take a 3 GHz CPU to pull off the
speed at which the Geometry Processing Unit (GPU) onboard the GeForce can do. Not bad at
all if that is even half true. The GeForce is the first consumer-level GPU. But lets look
a little back on what happened when NVIDIA let slip this new thing called T & L.
Some people on the internet thought it would be able to push 20, 30
40 million polygons a second onscreen. It would be faster than a PlayStation2 and make you
breakfast in bed as well. It would do the washing up, ask that girl you fancied, who sat
in front of you in class out, and sing you lullabies when you finally lay rest to 40 winks
of sweet nothingness.
Well at least they woke up and smelled the roses. The GeForce is a
generational leap ahead of the current cards (TNT2, Voodoo3, G400). However some people
were believing a lot of crap. Well there you go you RIVA lovers, it won’t do 40
million polygons per second, nor does it compare to the PlayStation 2. And tell me what is
the point of comparing a six month old card to a PlayStation 2 which is out on March 2000?
It is worth noting that admist every single screenshot I have seen of Sony’s next
console has left me underwhelmed apart from Tekken. Even GT2000 looks so-so. These games
are nowhere complete but I get the feeling that either the developers are having a
difficult time getting to harness it power (which is true anyway because of its design) or
that it is not as powerful as we are being led to believe. I mean Soul Caliber beats hands
down some games seen on PlayStation 2 previews. Naturally these are just prescreens yet so
hopefully the situation will change by March 2000.
So what am I trying to say? Don’t even think about comparing
the two technologies. Apples are nice and so are Oranges OK? Glad you thought so too.
Anyhow. . .
NVIDIA did something really quite clever before ECTS. They showed
alpha boards to a few select developers who then fell in love and began working on ways to
implement the T & L code and then worked on the DirectX7 code after it arrived. This
has resulted in games like Evolva from Computer Artworks and Experience from WXP coming
out with some way cool shots for us to drool over. . .T&L is here to stay and you can
bet that it will have an easier ride of becoming a used standard unlike something like
S3TC which ids great just not so widespread yet. OpenGL already is ready for it (been like
that for time anyway) and DirectX7 was engineered with T&L code with help from NVIDIA’s
engineers as well Cube Environment mapping and other GeForce specific features. Tim
Sweeney said this off T&L "Holy Shit!"
The beauty of T&L is that it will not be so bad if you have a
processor like a PII 300 when you compare it with a P III 450. The difference in
performance in resolution at or below 1024x768 is not so profound. The sweetspot as some
people claim, 1024x768x32bit should be easily achievable in most games by the GeForce card
even if T&L is not enabled. At higher resolutions a higher fillrate would help. The
current clockspeed of the GeForce is 120Mhz and perhaps when the chip moves to 0.18 micron
technology we will see some rapid speed increases in the core clock speed. However for
this we will probably have to wait until March 2000 when the second revision of GeForce
comes out for competition with 3dfx’s new card and the rest of the competition. Also
if 3dfx claim a 1 gigapixel fillrate and NVIDIA decide that the need to at least approach
that, then they will with T&L also. However it is doubtful whether NVIDIA will
implement FS AA in the next batch of 3D cards. A move to 1 gigapixel would require a
250MHz processing speed which is beyond the graphics cards of today. S3 promised something
approaching 150Mhz to 200 MHz with their 0.18micron technology but have not achieved this
at all.
For this time next year you can expect brand new technology from
NVIDIA and S3 (Project X), as for 3dfx well their past record is not so good on this
point. Perhaps more of the same with added T&L? Embedded DRAM is sure to become the
buzzword next year. Ask Glaze3D about it or go to Sony, Nintendo or even Rendition on the
advantages of eDRAM.
If you are stuck for something to do read my other stuff:
[Page
1] [Page 2] [Page 3] [Home]
|
