Overclocking and Temperatures
Overclocking the GTX 560 Ti was done using eVGA’s Precision to set the clock speeds on the GPU and memory and monitor temperatures. The OC Scanner program, also from eVGA, was used to check the speeds and test the stability of the overclock. Stock speeds on eVGA’s GTX 560 Ti Superclocked are 900MHz on the core and 2106MHz (4212MHz effective) on the memory. The card was overclocked speeds of 950MHz core and 2386MHz (4772MHz effective) on the memory. This represents a modest overclock of 50MHz on the core and an impressive overclock of 560MHz (effective) on the memory. Attempts to push the GPU core speed further were unsuccessful. To keep the card cool, the fan was set to run at 70% of the maximum speed.
First, a table detailing the increase in average and minimum frame rates is shown below. Then some figures for the above benchmarks will be given to demonstrate the increase in performance.
As is expected, the overclock increased the average and minimum frame rates in all test cases. The average increase in frame rate was around 5-7% for each test, with increases of as much as 24% reported. Next, we will revisit the benchmarks and see if our overclock can retake the performance crown for the GTX 560 Ti over the HD5870 and GTX470 in the cases where it fell behind. First up is the Metro 2033 benchmark.
At stock speeds, the GTX 560 Ti performed well at the low settings. However, at maximum settings, the HD5870 edged out the GTX 560 Ti.
With our overclock the GTX 560 Ti was able to gain a couple of frames per second at medium settings and able to edge out the HD5870 at high settings. Next up is the Battlefield: Bad Company 2.
Battlefield: Bad Company 2
In the benchmark at stock speeds, the GTX 560 Ti outpaced the other two cards at medium settings, but once again, the HD5870 outpaced the GTX 560 Ti at high settings.
With the overclock, we find the same thing as with the Metro 2033 performance. The GTX 560 Ti gained a few frames per second at medium settings and once again was able to edge out the HD5870 at high settings. The frame rates are all still within the playable range at both settings. Finally, let’s revisit the 3DMark 11 benchmark.
In the original test at performance settings, the GTX 560 Ti outpaced the GTX470 and HD5870, but for a third time, the HD5870 edged out the GTX 560 Ti at extreme settings.
With the overclock, the GTX 560 Ti edges out the HD5870 in 3DMark 11. So, as was mentioned before, although the HD5870 has an advantage with its larger number of shaders, the GTX 560 Ti makes up for this with raw speed, and adding more speed pushes the card beyond the performance of the HD5870. But this speed is no good if the card runs too hot. Next, we will take a look at the temperatures for stock speeds and overclocked speeds.
The figure below gives the idle and load temperatures for stock settings and the above overclock.
As is expected, the card idles at the same temperature of 32C for all three cases. This makes sense since the card downclocks when in 2D mode at the Windows desktop. For the load case, we can see the difference the fan speeds make. At stock speeds and overclocked, the auto fan speeds keep the card around 80C. While this isn’t terribly hot and isn’t near the upper temperature limit, it is quite hot. When the fan speed was increased to 70% of the maximum speed, the card was cooled down to a much more acceptable temperature of 65C. This drop in temperature allowed me to push the overclock of the card further as well.
While these temperatures are a good guideline, they do not represent what you may actually encounter. As with all temperatures, they depend strongly on your cooling setup and how well your case performs at removing hot air and supplying cold air, and my case features a cold air intake that passes cold air directly over the graphics card.