Though the GeForce4 series has been out for a while now, the most recent of the series is the TI4200 model. It's an expansion of GeForce3 technology such as vertex and pixel shaders, and the Lightspeed memory Architecture that we discussed in our Chaintech GeForce3TI200 review. However, NVIDIA has undergone some controversy concerning the GeForce4 name.

Will the REAL GeForce4 Please Stand Up?

NVIDIA actually has 2 completely separate lines of GeForce4 products. The GeForce4Ti series is a direct descendant of the GeForce3 series, containing all of the following wonderful features:

nFiniteFX II engine
-dual vertex shaders
-advanced pixel shaders
-3D Textures
AccuView Antialiasing Engine
Lightspeed Memory Architechture II
-Lossless Z-Compression
-QuadCache
Shadow Buffers
Z-Correct BumpMapping

However, there is ANOTHER GeForce4 series of chip: the GeForce4MX. This does NOT contain the full nFiniteFX II engine, though it includes the Lightspeed Memory Architecture II and AccuView. For information regarding shadow buffers, the LMA and other GeForce3 carryovers, please see our previous GeForce3TI200 review. We are only going to cover the additional and improved features of the GeForce4TI line, and cover the GeForce4MX line in another review.

AccuView

The AccuView Antialiasing is more than simply a new AA mode to replace Quincunx (and it's certainly a better name, though Quincunx AA is still present as an option). It's a complete and dedicated antialiasing engine on the chip itself. Since it's only purpose is to provide antialiasing services, it should do a very good job of it. It's designed with a new 4XS mode that is intended to provide 50% more sub-pixel coverage than any other NVIDIA anti-aliasing mode. What this means is it can take one pixel and divide it into many more pixels than even Quincunx AA mode can. Of course, the question is, how much framerate performance must be sacrificed for this? Since the AA engine is a dedicated one, it shouldn't be a lot, but remember that the rest of the GPU has to process all of these additional pixels as well. However, despite the fact that NVIDIA is into version 40.xx drivers now, AccuView AA is still only available in Direct3D games.

QuadCache

QuadCache is an addition to the Lightspeed Memory Architechture (Hence the "II" designation) that is designed to provide an ultra-quick-access buffer for storage of pixel and vertex data. However, NVIDIA has not provided any actual specifications as to the speed or size of the cache, and it's presence can only be confirmed by any performance increase of the chip itself.

MSI's TI4200 line

G4Ti4200-TD64
  -64 MB RAM, VGA/DVI/TV-Out
G4Ti4200-TP64
  -64 MB RAM, VGA/VGA/TV-Out, TwinBIOS
G4Ti4200-TD
  -128 MB RAM, VGA/DVI/TV-Out
G4Ti4200-VTD64
  -64 MB RAM, VGA/DVI, VIVO
G4Ti4200-VTP64
  -64 MB RAM, VGA/VGA, VIVO, TwinBIOS
G4Ti4200-VTP128
  -128 MB RAM, VGA/VGA, VIVO, TwinBIOS

Basically, there are a lot of cards, with a wide mix of features. The moral of this is to read the description carefully and know which one you're getting.  We will be reviewing the 128 MB card, aptly named "VTP128".

NVIDIA has released 2 reference specifications for the TI4200 video cards. One calls for 64 MB of DDR RAM running at 250 MHZ (500 MHZ effective). The other calls for 128 MB of DDR RAM running at approx. 222 MHZ (444 MHZ effective). In both cases the GPU's clock is 250 MHz. But MSI has different plans for today's review card (The G4Ti4200-VTP), which is their top-of-the-line TI4200. Even though it comes with 128 MB of RAM, MSI has clocked it at 250 MHZ DDR (500 MHZ effective)!

Next Page: (2)

Page 1: Page 2: Page 3: Page 4: Page 5: Page 6: