Graphic Cards?

[Edward Fitzgerald]

I have a nvidia 4400 with 128mg ram, and I want to upgrade. My question is
regarding two cards that I am looking at
an Nvidia 5700 256mg ram and a Radeon 9600xt with 128mg of ram, both of
these cards are in the $200 range. In doing some research it seem that the
consensus is the Radeon is the better card, but before I buy one I would
like to understand why some more cards with 128 mg of ram are more
expensive that some cards that have 256mg. Which is more important the
chipset or the memory. I have a Pentium 4 2.4 processor on a Intel
motherboard with 800mhz front size bus that supports hyper threading
technology, and two 512mg ddr memory modules. I work allot with digital art
using several programs and do 3d rendering and I am hoping a new card will
speed up the rendering time. I also am a flight simulator user and like to
have a card that when I set the high end display settings it doesn't slow
down or degrade the graphics. I know that my question is fairly general but
would appreciate having anyone's knowledge and experience with these or
similar graphic cards.
Edward

 

[Michael Arm]

I have a nvidia 4400 with 128mg ram, and I want to upgrade. My
question is regarding two cards that I am looking at
an Nvidia 5700 256mg ram and a Radeon 9600xt with 128mg of ram, both
of these cards are in the $200 range. In doing some research it seem
that the consensus is the Radeon is the better card, but before I buy
one I would like to understand why some more cards with 128 mg of
ram are more expensive that some cards that have 256mg. Which is more
important the chipset or the memory.

As far as I know, 256MB models of nVidia use a slower memory than for
the 128MB models for their cards. Making them slower when filling a
scene with textures and the likes.
Since a lot of operations take place in the video memory, the memory
speed is an important feature.
However the chip architecture is also important.
For example: I upgraded my "old" GeForce 4 Ti 4200 to a GeForce FX 5600.
Both with 128 MB Ram.
While being slightly faster in several 3DMark '03 benchmarks, the FX was
significantly slower than my GF4 when Multi-Texturing was tested. This
is due to the fact, that the GF4 has more shading pipelines for
textures. This means, a polygon for a wall can be rendered parallel when
you put a brick-texture on it, that is overlapped with a texture
simulating an ivy growing on that wall, and a dirt texture that
simulates wet, glistening spots on the wall.
The FX has to do some texturemapping one after the other.

Furthermore to the comparison of ATI and nVidia:
Microsofts DirectX9 Pixel Shader use an accuracy of 24 bits.
The Ati cards compute these 24bits in Hardware.
The nVidia Pixel Shaders compute in 16 bits or 32 bits. Therefore the
driver always has to decide if the Pixel effect is important, meaning it
has to be converted to 32 bits, or not that important (e.g. its too far
away to notice the difference) and is done in 16 bits.
However the decision and conversion from 24 to 16 or 32 bits takes time.
And therefore the nVidia Cards are slower in terms of Pixel Shader 2.0
Performance in DirectX9 games.

I have a Pentium 4 2.4
processor on a Intel motherboard with 800mhz front size bus that
supports hyper threading technology, and two 512mg ddr memory
modules. I work allot with digital art using several programs and do
3d rendering and I am hoping a new card will speed up the rendering
time.

Not really. 3D Rendering software like Maya or 3D Studio Max use highly
specialized Pixel and Vertex Shaders (that determine how a surface
looks, e.g. how it is lit, if it is shiny, if it is a rough material,
transparent etc.)
In games, the necessity for fast frame rates comes at a cost of blocky
objects with a compromise regarding visual texture quality and speed.
You will not see a brushed aluminium surface in a real-time rendered
game. In a 3D Program the shader has all the time in the world to
simulate the small scratches that make the interesting lighting effects
of a brushed metal surface and a moving light.

True is, that while interactively constructing the scene, you will
profit from a fast 3D card. But the cards mentioned above are only
slightly faster than "old" DirectX8 cards like the GeForce4, with the
additional DirectX9 Features. And I haven't seen these DirectX9 features
utilized in the modeling panes of such 3D programs.
Als long as you are actively moving stuff around in your work area, the
3D cards play with their muscles.
As soon as you hit the "render" button, everything is calculated by the
far more complicated shaders of your 3D Modeling software.

I also am a flight simulator user and like to have a card that
when I set the high end display settings it doesn't slow down or
degrade the graphics.

That depends on the age of your Flight sim. I have found, that my
GeForce4 Ti is able to play almost all of the actual games in the
highest detail settings on my Athlon 2800+ and 512MB 333MHz Ram.
Since games that utilize the real DirectX9 Features have yet to come
(i.e. Doom3, Half Life2 etc.), an upgrade to a newer graphics card in
your situation is (in my opinion) currently a waste of money.

 

[jimmybob]

Reguarding your question of whether you should get a
256MB or 128MB version , has to do with what resolution
you play your games at. If you play at 1024 x 768 , or
smaller, 128MB will be sufficent. The 256MB versions are
for running games at EXTREMELY large resolutions, like in
the 2000x by 2000x ranges. I personally think that the
Radion series is a good one, i personally have a ATI
radeon 9600 pro 128MB and i love it. I havent looked much
into the xt series, but Half-life 2 is going to work
closly with that series, so that makes it a A++ in my
book. if you look at www.pricewatch.com you'll notice
good prices too.

 

[Anon]

I work allot with digital art

using several programs and do 3d rendering and I am hoping a new card will
speed up the rendering time.

You would get a better hit by adding more RAM than you would by getting a
better graphics card consider the graphics card for gaming only. It is the
applications Render engine that will determine rendor times e.g. Bryce
renders have been known to take days for professional standard renders.