Overclock

A short test of the card including stock cooling system has show the following overclocking results:

VID Fan, % Fan, RPM Tgpu
EVGA GeForce GTX 570 732/950 MHz 0.950 V 64% 2370RPM 86C
EVGA GeForce GTX 570 @800/950 MHz 0.950 V 65% 2430RPM 86C
EVGA GeForce GTX 570 @875/950 MHz 1.000 V 67% 2700RPM 87C

Card settings for the following clocks @800/950 MHz has resulted in total stability during 10hours of non-stop testing.

Next, the card had the stock air cooling system changed to liquid cooling, using Swiftech MCW60 universal GPU waterblock. With the GPU being cooled by water, overlooking results are much better now:

VID Fan, % Fan, RPM Tgpu
EVGA GeForce GTX 570 @910/950 MHz 1.050 V - - 45C
EVGA GeForce GTX 570 @970/1235 MHz 1.100V - - 49C

Further overclocking came harder and card operation was, certainly, not stable enough. On a large scale, overclocking technique was already described in our previous article on how memory overclocking effect the overall performance, so this time we learned our lesson and increased the memory speed together with GPU overclocking, which in the end had a huge impact on final results. With GPU/Memory clocked at 1000/1250 MHz and Vgpu=1.100В we managed to pass some benchmarks. For example, in 3DMark Vantage with Performance profile (PhysX technology turned off) the card got 28546 points:

28546 points in 3DMark Vantage

3DMark 11 benchmark, which less dependence on CPU performance, our 7062 points with Performance settings allows us to to hold pride of place among card of the same model:

7062 points in 3DMark 11

As for final words on alternative cooling system that was used, I have to say that with GPU at 1GHz the following voltage increase did not effect card’s stability.

During testing, the following voltage measurements were:

VID 0.950 0.963 0.975 0.988 1.000 1.013 1.025 1.038 1.050 1.075 1.088 1.100 1.113 1.125 1.138
Vgpu, measure 0.960 0.972 0.985 0.998 1.010 1.024 1.036 1.050 1.063 1.088 1.101 1.113 1.126 1.139 1.152
Vmemory 1.570

So, now we can move to actual performance results of cards with stock cooling systems. For overclocked cards, we displayed clocks, that were safe to set without default voltage change.

Leave a Reply

required

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>