Posted by Thomas
McGuire on December 26, 2001
Labs Europe Product: 3D
Blaster GeForce 3 Titanium 200
for the lowest videocard
moving onto anything else I'll tell you about my experiences
with the GeForce 3's stability out of the way. Personally
speaking, stability of a Graphics Card (Or other
Hardware/Software) is the most important feature to me and
given my experiences from various newsgroups & e-mails
I've received I was a bit nervous about how stable the
GeForce 3 would be.
receiving this Graphics card I had received Windows XP &
was a bit anxious about installing the latest NVIDIA Drivers
after noticing this
story on The Inquirer regarding the 23.11 Drivers.
Thankfully I didnít have any problems with the new
features in the Driver & VIA also responded with a patch
quickly to circumvent any possible problems also (Though not
Intel or Sis, Ali as far as Iím aware).
stability of the Drivers was good, although not without
issue and I did experience problems with some Games, e.g.
when running the timedemos on Quake 3: the game would crash
sometimes when changing
resolution. At another point Serious Sam wouldnít
load for some reason
GeForce 3 supports DXTC & S3TC Texture compression
modes, although disappointingly the GeForce 3 was allowed to
inherit the same fault as the GeForce 1 & 2 had: the
DXT1 problem. The problem is that in the GeForce 3 DXT1 uses
only 16-Bit interpolation, while other Graphics Cards use
higher interpolation depths. As a result any DXT1 compressed
textures with the GeForce 3 (Or 1 & 2 for that matter)
look hideous. The most commonly used example to illustrate
this is the sky in Quake 3. NOTE - S3TC=DXTC, albeit DXTC is
used in Direct3D while S3TC is used in OpenGL, although
other than that no differences exist, both use the same 5
Compression algorithms (DXT 1 - 5).
of sorts however does exist for this problem in OpenGL only.
The NVIDIA driver team added a workaround
whereby DXT1 requests are changed to DXT3 instead. It must
be stressed this is a workaround only
& that DXT3 offers poorer a compression ratio than DXT1.
The screenshots below illustrate the Quake 3 sky with
Texture compression disabled,
without the Workaround in effect, then with the Workaround enabled.
sake of comparison I've also thrown in a shot of the same
scene using a Vivid!
XS Kyro 2 that supports DXT1 only.
2 - DXT1 Compressed
Screenshots are fairly self explanatory as you can see, the
Sky with the Kyro 2 Graphics Card looks fine, albeit
with a small amount of colour banding (Noticeable with all Graphics Cards I might add) while the DXT1 compressed Sky on
the GeForce 3 is a rather highly banded mess.
asked NVIDIA (Brain Burke & Diane Vanesse specifically)
for some comments on this matter a few times over half a
year ago, although to this date I've never received any
reply/acknowledgement to my Questions (One of which being
would the GeForce 3, unannounced at the time, have resolved
NVIDIA did make some statement on the matter, that being
according to spec". This is partially true as
the DXTC spec makes no mention of what interpolation to use
for DXT1, but anyone with a basic sense of logic can
figure out that in order for Texture compression to be
viable it must compress textures with a minimum amount of perceivable artefacts, something which NVIDIA
has failed to pick up on sadly. Epic's
Daniel Vogel had
the following to say over on GameBasement about the
matter (Partially in response to the previous NVIDIA claim
that it works correctly);
don't work according to the specification & the below is
just PR talk. I doubt you would get one of their developers
to state that the texture compression works according to the
specs & even if it were working according to the specs
(which it is not) it wouldn't be an argument as every other
company has better DXT1 support.
wouldn't matter if it were working "correctly" if
it is looking worse than ATI's. The average gamer doesn't
care about technical details - she/he wants nice looking
graphics on her/his screen.
can read the full article on about DXT1 on NVIDIA GeForce
Graphics Cards over at Gamebasement.
wise I did a quick test in Quake 3 & using DXT1 the
average frame rate was 141.9, whereas with DXT3 (workaround)
the frame rate dropped slightly to 140.6. That said the
difference will be more noticeable in Games which use a
greater amount of DXT1 compressible textures.
the point to get from all of this is that DXT1 on the
GeForce 3 does not work correctly. It does not do
what texture compression was intended
to do Ė reduce video memory storage size with minimal
image quality difference. Whereas NVIDIA may insist it works
according to spec the fact that newer Drivers contain a
workaround in the OpenGL driver for this would seem to
indicate even they realise itís a problem. Letís hope
the next GeForce finally can perform DXT1 compression
properly, in much the same way that the PowerVR Kyro 1/2