MSI has built a technology into their drivers called D.O.T. Express. D.O.T. stands for Dynamic Overclocking
Technology. MSI claims this is the easiest way to optimize the GPU and memory speed for maximum performance.
MSI also claims this technology smartly detects and saves the GPU from overheating and overspeeding. The
card already comes clocked higher than a normal 7600GT, so anything else that you can squeeze out of it is cool.
Vivid Control Panel
The control panel allows for 6 different overclocking levels, ranging from 2% overclocked to 10% overclocked.
I tried this card on all levels. I was expecting the same type of overclocking from this card as from the previous
MSI NX7600 GT that I reviewed. I first set D.O.T. to the Colonel level. Running 3DMark05, it immediately froze up.
After a reboot, I set it to Private level, and 3DMark05 ran slower than with the stock settings, with several
noticeable hiccups. I was very disappointed with these results, so I turned on coolbits for overclocking, and
used the Detect Optimal Frequencies to try to overclock manually. It wanted to clock to 578 Core Clock and 740
Memory Clock. Using different ForceWare drivers, I experienced similar results. I'm not sure exactly what the
problem was with the board, but it certainly didn't want to overclock.
Clock Control Panel
I went with the latest beta driver (91.33), which meant that my old tests were no longer valid. I decided to redo
all of the tests from the last review. New demos were recorded, and all traces of the HardwareOC programs were removed.
3DMark05 and 3DMark06 were used for synthetic benchmarks. Quake 4, Far Cry, and Half Life 2: Lost Coast were used for
games. A new test was added for one of the latest and greatest games, Prey.
3DMark05 v1.20 and 3DMark06 v1.02 were both used for the synthetic tests of the NX7600GT HD. For 3DMark05, SLi produced
a 51% increase over a single card. For 3DMark06, SLi produced a 57% increase over a single card. These scores were lower
than the scores produced by MSI in their documentation. Since they did not post their system specs, I can only assume that
the system I tested on is not as powerful as their test system.
All games were benchmarked with the single NX7600GT-VT2D256E HD, and in SLi mode with the NX7600GT-VT2D256E
from my previous review. All games were run using 3 resolutions: 1024x768, 1280x1024 and 1600x1200. The games
were tested in 3 modes under each resolution, 0AA/0AF, 4AA/16AF with Gamma Correct and Supersampling Transparency,
and 8xSAA/16AF with Gamma Correct and Supersampling. Frame rates were recorded using FRAPS. For all games, I
ran 4 benchmarks, throwing out the first benchmark and averaging the last 3. For Quake 4 and Prey, I recorded
my own demo, and used that for benchmarking. For Far Cry, I used the built in Regulator demo. For Half-Life 2:
Lost Coast, I used the built in Video Stress Test.