|
In the nest table you can pinpoint the exact position of this product compared
to compared to other GeForce 4's I suggest you take a
look at the following table:
| |
MX 420 |
MX 440 |
MX 460 |
Ti 4200 |
Ti 4400 |
Ti 4600 |
|
Number of Transistors (Millions) |
29 |
29 |
29 |
63 |
63 |
63 |
|
Manufacturing Process (microns) |
.15 |
.15 |
.15 |
.15 |
.15 |
.15 |
|
Rendering Pipelines |
2 |
2 |
2 |
4 |
4 |
4 |
|
Texture per Pass |
4 |
4 |
4 |
8 |
8 |
8 |
|
Core Clock (MHz) |
250 |
270 |
300 |
250 |
275 |
300 |
|
Memory Clock (MHz) |
166 |
200 |
275 |
250 |
275 |
325 |
|
Memory Amount (MB) |
64 |
64 |
64 |
128 |
128 |
128 |
|
Memory Type |
5.5ns |
4ns |
3.6ns |
4ns |
3.6ns |
3ns |
|
RAMDAC |
350 MHz |
350 MHz |
350 MHz |
350 MHz |
350 MHz |
350 MHz |
|
Fill Rate (gigatexels) |
1.0 |
1.08 |
1.2 |
2.0 |
2.2 |
2.4 |
|
Fill Rate (megapixels) |
500 |
540 |
600 |
900 |
1100 |
1200 |
|
Fill Rate
(millions of Triangles/sec.) |
31 |
34 |
38 |
114 |
125 |
136 |
|
Bandwidth
(GB/s) |
2.66 |
6.4 |
8.8 |
8.0 |
8.8 |
10.4 |
This particular model we are reviewing today should competition wise be between
the 4200 and 4400 with it's default 250 MHz core clockspeed. The features of the GeForce4 (read keywords) you need to
remember, for the GeForce4 Titanium / NV 25 are Accuview, nView, nfiniteFX II
Vertex Shaders, nfiniteFX II Pixel Shaders and Lightspeed Memory Architecture
II. Let us look at some innovations and specs of the GeForce4 Ti series with you
that where provided by NVIDIA:
- 63 million transistors (3 million more than GeForce3)
- Manufactured in TSMC's .15 µ process
- Chip clock 225 - 300 MHz
- Memory clock 500 - 650 MHz
- Memory bandwidth 8,000 - 10,400 MB/s
- TnL Performance of 75 - 100 million vertices/s
- 128 MB frame buffer by default
- nfiniteFX II engine
- Accuview Anti Aliasing
- Light Speed Memory Architecture II
- nView
One thing is a sure fact, 3.3ns memory can
easily take that 550 MHz clock frequency. Let's do our usual round of
calculations shall we and bare in mind that a normal Ti 4200 would have a
bandwidth of roughly 8000 MB/sec, we are talking about the faster 64MB model
here btw.
There's a simple way to calculate
everything you need to know, you just need to look at the memory chips and note
the last numbers on the chip as it will state how fast they are or simple write
down the series number and check it on the manufacturer website.
Let's calculate the nominal speed
for this memory:
1000:3.3ns x 2 (DDR) = 606 MHz
Hey .. it's clocked at 555 MHz, that means there is
plenty of room for overclocking. Now that
we have calculated nominal memory bandwidth and can see that the ram modules
deliver the frequency that NVIDIA states we can raise the bar a step and
calculate memory bandwidth:
(2x128bit) x (2:606MHz) : 8bit =
9696 MB/sec That's about a 1696 MB/sec above standard specification.
When we fired up SiSoft's Sandra we where able
to figure out that the new A3 revision of the GeForce4 GPU was used on this
videocard. A3 is the latest build of the GPU and is also being used on any
GeForce4 Ti 4600. Combined with the Ti 4600 PCB I think we'll be able to tweak this puppy to about 300-310 MHz but more on that later in our overclocking bit.

DDR memory rated 3,3ns - click image for
high-resolution photo
|