Overclocking
Overclocking the two nVidia cards was a breeze. RivaTuner did an exemplary job and it not only allowed us to change the clocks but it also let us see how the shader clocks were affected by our GPU changes. The XFX card with its active cooling solution allowed us to push the GPU speed all the way to 499 MHz (49 MHz above stock) and in turn the shader clock from 1188 MHz to 1350 MHz. You would think that because of the passive cooler on the Zotac card things would be different there, but they weren’t. We again managed to push the card to 499 MHz, but this time the shader clock only reached 999 MHz. The different shader clock speeds are also the major reason we were able to push the GPU clocks to the same level in spite of the very different cooling solutions. For the memory we were quite lucky both times, especially considering the DDR2 memory chips used on both cards. From the stock 400/800 MHz we got all the way to 441/882 MHz without any problems. Further on we started experiencing texture corruption, so we know that was as high as we could take it.
The ATI HD2400XT proved to be a tougher nut to crack. Most of the tools simply refused to recognize the card and even the latest version of ATITool only allowed us to monitor the temperature – changing the clocks was asking it too much. In the end we resorted to the AMD GPU Clock Tool, which although a bit rigid in terms of features let us overclock the card. After some trial and error we settled on a GPU speed of 750 MHz (55 MHz over the stock speed) and a memory speed of 436,5/873 MHz (40,5/81 MHz over the stock speeds).
The increases we achieved are all very standard and all three cards maintained their original positioning. The two nVidia cards benefited more from the overclock, which is especially noticeable in the Lost Planet graph, where the XFX card managed to overtake the HD2400 and the Zotac card finally managed to catch up in the Cave test. In Call of Juarez the green cards also benefited more from the overclock, though the ATI card remained well ahead all the same.
Things get complicated when speaking of the temperatures and noise produced though. The XFX card comes out as the winner here, as the temperatures remained stable (never going above 75°C) and the noise levels didn’t increase. The Zotac card remained silent due to there being no fan on the card, but the temperatures raised to 90°C, a situation that is hardly ideal. The HD2400 went the other way and maintained reasonable temperatures (80°C peak), but became quite noisy. This makes the card a bad choice for multimedia centers and the Zotac card a prime candidate, as long as the system is well ventilated.
