Chaintech SA6600G Video Card Review - Overclocking
(Page 5 of 9 )
When it comes to overclocking, it doesn't get any easier than enabling the "Coolbits" registry tweak to get access to clock speed modifications within the nVidia drivers. There's even an auto overclocking feature attached to it, though it doesn't work all that well. It overestimated what the core on this chip could take, and well underestimated the memory clocks. However, it's at least a place to start looking for artifacts. I found FarCry, HL2 and 3DMark05 to be the best programs for bringing out causing instability, and through a combination of all three I found the best this particular card would do was 552/600MHz. The memory overclock was quite impressive, as I said above it's probably getting hotter due to the heatsink design as opposed to being open to case airflow.
Testing
- Intel Pentium M 735 @ 2.26GHz (141x16)
- DFI 855GME-MGF
- 2 x 512MB Kingston HyperX PC3000 @ 188MHz
- 2 x 80GB Hitachi 7K250 in RAID 0
- Windows XP SP2
- Chaintech 6600GT @ 500/450MHz (default) using ForceWare 67.03
- ATI Radeon 8500 @ 275/275MHz using Catalyst 4.11
Driver settings:
- NVIDIA - image settings at Quality, vsync off, optimizations on, 85Hz refresh rate
- ATI - all settings to "Balanced"
Benchmarking with these two cards is less than a perfect scenario. They are very different in age, and available driver settings. It's far from an "apples to apples" comparison. With the way drivers, settings, optimations, and AA/AF methods are all different from one architecture to the next, I'm not sure there can be an "apples to apples" comparison outside of tests comparing cards with the same GPU.
So, what is the next best way to compare cards? Instead of "identical" settings, I went for which setting could provide the best gaming experience, in terms of playability and eye candy. This is usually attempting to garner a minimum frame rate above 30FPS, and an average somewhere north of that. In the case of some the benching with the Radeon, that just wasn't possible. It's a little weak in the knees for some modern games. All numbers were taken from FRAPs in various parts of the games, over 90 second intervals. So other than 3DMark benches, none of these relate back to standard canned benchmarks, instead they are from actual gameplay. Please be sure to read the paragraph under each graphic to find out the settings used for each card.
Next: Call of Duty, Doom3 >>
More Video Cards Articles
More By DMOS
