Gainward Ultra/750 Golden
Sample GeForce4 Ti4600 Review
- April 7, 2002
So just what is different about the GeForce4 Ti series of
cards? Is this just a hopped up GeForce3 card running at
higher clock speeds? Or if not, what's new with it? I will bullet some of the key features of the
GeForce4 Ti series of cards. I am going to give a word of warning to the naive
users about the "GeForce4" MX series of cards. They are really less
than GeForce3's if you ask me, as they do no programmable pixel or vertex
shading whatsoever. If you are a 3D gamer, you will be disappointed with
your purchase. You may save a few dollars, but you will have a lackluster 3D
gaming card compared to the GeForce4 Ti series of cards.
1) The NV25 chip - Built on a .15 micron process, it has 63 million
transistors. That's 6 million more transistors than the GeForce3 chip and 8
million more than the Pentium4 has. It also adds an extra vertex shader to pump programmable
shading in DirectX 8.1 games at double the throughput of what a GeForce3 can do.
NV25 chip layout
2) BGA memory modules - The GeForce4 cards incorporate a new package design
for their memory modules. Ball-grid array (BGA) uses small balls of solder to
connect to sockets on the card. The older NVIDIA cards used TSOP (thin small
outline package) to connect to the card using leads which where each soldered to the
PCB, creating dozens of possible failure points. BGA offers higher speeds
with less errors. The GeForce4 Ti4600 is using top of the line memory and
plenty of it.
The DDR RAM is in a smaller
The GeForce4's new memory
has really got balls! Solder balls that is .
Each of them fit into a corresponding socket on the card. Resulting in faster memory with
The GeForce4 Ti4600 cards come with 128MB of 2.8ns DDR RAM.
That's twice the memory that came with a typical GeForce3 card, and it's rated about
2ns or 41% quicker than the GeForce3's memory. BGA modules are smaller and pack more of a punch. The
memory on the GeForce4 cards is a big plus over that which came with the GeForce3's. The architecture used on the GeForce4 GPU
also uses this memory in a much more
efficient manner than the GeForce3 cards did. The next bulleted item will
explain how it does this.
3) Lightspeed Memory Architecture II - Adding
new enhancements to the GPU's crossbar memory technology. By drawing only
those pixels which will be shown, compressing Z-buffer data and changing the
way that data is written to and read from the memory banks, the GeForce4
reduces the amount of traffic traveling over the memory bus. All these
features answer the problem of memory bottlenecking, and give the GeForce4 a
theoretical 10.4GB/sec bandwidth. The additional bandwidth makes
FSAA a very usable feature, which does not take such a massive hit on
performance as it did on the GeForce3 cards. This is what NVIDIA says about
"A crossbar-based memory controller: Ensures that every aspect of the memory system is balanced and that all memory requests by the graphics processor are handled properly. Under complex loads, LMA II’s memory crossbar architecture delivers 2-4 times the memory bandwidth of other standard architectures.
A Quad Cache memory caching subsystem: High-speed access buffers that store small amounts of data and operate at tremendously high bandwidth, ensuring that data is queued and ready to be written to the memory. These caches are individually optimized for the specific information they deal with, resulting in almost instantaneous retrieval of key data.
Lossless Z-buffer compression: Reduces Z-buffer traffic—one of the largest consumers of memory bandwidth in a graphics subsystem—by a factor of four, without any reduction in image quality or precision.
A visibility subsystem: Determines whether or a not a pixel will be visible in a scene. If it determines a pixel will not be visible, the pixel is not rendered, saving valuable frame buffer bandwidth.
Fast Z-clear technology: Minimizes the time it takes to clear the old data in the Z-buffer, boosting frame rates up to 10% without compromising image quality.
This is probably one of
the most important changes in the GeForce4 over the GeForce3's technology.
Memory is handled much differently and better with the GeForce4. Note that the
visibility subsystem has NVIDIA now showing signs of the Gigapixel/3dfx hidden
surface removal influence. The Quad Cache, Fast Z-buffer clearing and Lossless
Z-buffer compression are really great features as well. It all
adds up to a totally new architecture in the GeForce4, and points to the fact
that it is not just a juiced up GeForce3 card.
4) Accuview FSAA - The
GeForce4 Ti cards use a unique multi-sampling technique for rendering FSAA
called "Accuview". What it does is renders extra pixels at less of a
performance hit due to the AA subsystem being outfitted with wider internal
data paths which are designed to accommodate additional texture
information. In addition, pixels are now rendered in semi-random spots
rather than the same exact spots all the time. They are continually changing,
which creates, believe it or not, a more mathematically correct FSAA effect
and does away with the FSAA halo effect. So with the GeForce4 Ti cards, you
get a whole new way of handling FSAA. It looks better and performs much better
than the GeF0rce3 cards did.
5) Multi-Monitor Support - The GeForce4 cards come
equipped with two monitor outputs, with at least one having DVI output for use
with flat panel displays. With NVIDIA's nView feature, you can run two
displays at once. This is not offered with all the various companies cards,
you must check for it before buying your card if you need it.
So to sum things up, the GeForce4 uses a whole new architecture
which is greatly improved from that found on a GeForce3 card. It has more and
better memory which is used better, better implementation of FSAA, and two programmable
vertex shaders for use with DX 8.1 games. If you want the
newest and best technology in a 3D video card, the GeForce4 Ti is clearly the
card to get.
Overclocking & Performance