This content was originally featured on Amdmb.com and has been converted to PC Perspective’s website. Some color changes and flaws may appear.

Finally we get around to doing some overclocking. Even with a crooked heatsink and some questionable Artic Silver II application, I’m still determined to see how far I can push this card. For testing overclocking, I have included 3DMark 2001 benchmarks as well as UT2K3 to show the “real-world” effects.

Benchmark notes:

  • The following tests were done using two different system overclocks: 1660MHz CPU + 166MHz RAM, and 1633MHz CPU + 175MHz RAM.
  • No anti-aliasing or anisotropic filtering was used.
  • UT2K3 CTF-Face3 was run using 1024×768 with low details.

3D Mark 2001

3DMark 2001 results and overclocking.

You can see here that the card overclocking reached its peak around 315 MHz GPU and 634.5MHz RAM, which is faster than a GeForce 4 Ti4600 at default speeds. For a dramatic effect, you can see the 3DMark scores drop dramatically past this point. I suspect if the cooler was flush with GPU and RAM, we’d be able to squeeze more out of this. Not bad for a crooked heatsink.

Initially, I ran 3DMark 2001 using a synchronous system overclock of 166MHz FSB (1660 MHz CPU, and 166MHz RAM). The 3DMark results were pretty disappointing as I was expecting something in the 10,000 3DMark range. Then taking some of the knowlegde I’ve gained as a moderator here, I decided to increase my RAM speed while keeping the CPU speed about the same (142MHz FSB for 1633MHz CPU and 175MHz RAM). The results were much better. By increasing my RAM by 9Mhz, I was able to squeeze out almost 2000 more 3DMarks! So if you’re looking to max out your 3DMark scores, get some good RAM that you can overclock.

But does any of this translate into a real difference in a game?

Unreal Tournament 2003

Here we use the same procedure used in the 3DMark 2001 tests. We used two different system overclocks and gradually push up the video card’s GPU and RAM until it can’t go any further.

UT2K3 results and overclocking.

The results are somewhat disappointing. Here we see that overclocking the videocard actually does very little for raw performance in Unreal Tournament 2003. There may be about 3 extra frames per second gained by overclocking, but this is such a small amount that it can easily be considered experimental error. The game started to show severe stuttering and polygon tearing after the 315/634.5 overclock. At 327.4/648, the game would freeze.

I think from these results we can say that for this particular card, you’re better off overclocking your system than overclocking your video. The marginal performance increase you see in a real-world game is minimal and probably not worth the trouble. So keep in mind that when you see a 3DMark score that appears to be impressive, be sure to find out what it means in a real-world application/game.

« PreviousNext »