Advanced Tweaking Performance

When we look at the previous pages and the cooling technology used by Sparkle did invite us to start some of our advanced tweaking, so at first we will look at some basic overclocking. In fact, since the Calibre 8800 GTX OC edition did use a good cooler we wanted to squeeze a bit more out of the card. So what we have done is quite easy, we use ATITool to find a sort of maximum clock speed for both the core and memory and afterwards set the clocks via RivaTuner. On top of that we also increased the shader clock which is set at default speed even with this pre-overclocked version. So we pushed the Calibre 8800 GTX OC edition a bit further then initially set by Sparkle, followed by some benchmarking with Futuremark and some games. So read on and find out the gains of our easy tweaking on the performance and frame rates.

Futuremark 3DMark06

Benchmark Results - 1600x1200 - E6300 2.80GHz - 2GB RAM
	Calibre 8800 GTX 768MB	3DMark06 Score
4x AA/8x AF	630/1960MHz - 1350MHz (Shader)	7669
4x AA/8x AF	630/1960MHz - 1450MHz (Shader)	7782
	Calibre 8800 GTX 768MB	3DMark06 Score
4x AA/8x AF	650/2000MHz - 1404MHz (Shader)	7895
4x AA/8x AF	650/2000MHz - 1512MHz (Shader)	8010
Results Interpretation: Higher index values are better.
Benchmark Results - 2560x1600 - E6300 2.80GHz - 2GB RAM
	Calibre 8800 GTX 768MB	3DMark06 Score
4x AA/8x AF	630/1960MHz - 1350MHz (Shader)	4730
4x AA/8x AF	630/1960MHz - 1450MHz (Shader)	4787
	Calibre 8800 GTX 768MB	3DMark06 Score
4x AA/8x AF	650/2000MHz - 1404MHz (Shader)	4903
4x AA/8x AF	650/2000MHz - 1512MHz (Shader)	4965
Results Interpretation: Higher index values are better.

It might be a bit disappointing to those who expected to see a greater overclocking of the core, but even with the TEC cooling from MACS Technology the Calibre 8800 GTX OC edition couldn't beat 650MHz clock speed. And yes although our previous GeForce 8800 GTS could be run at 675MHz with tuning up the fan speed it seems thought that the GTX isn't that flexible. But when looking back at the differences between the GTS and the GTX, then we would like to point out that the 128 stream processors might have something to do with the overclocking limit of the GTX chipset. In the end 128 stream processors running at higher speed, with on top of that a higher core will generate a bit more heat then only 96 stream processors available on the GTS chip. So in the end not a bad at all when we look at the performance increase with some minor tuning.

UbiSoft Far Cry

Gaming Results - 1600x1200 - E6300 2.80GHz - 2GB RAM
	Calibre 8800 GTX 768MB	Game Score
4x AA/8x AF	630/1960MHz - 1350MHz (Shader)	134.88
4x AA/8x AF	630/1960MHz - 1450MHz (Shader)	135.34
	Calibre 8800 GTX 768MB	Game Score
4x AA/8x AF	650/2000MHz - 1404MHz (Shader)	135.51
4x AA/8x AF	650/2000MHz - 1512MHz (Shader)	136.07
Results Interpretation: Higher index values are better.
Gaming Results - 2560x1600 - E6300 2.80GHz - 2GB RAM
	Calibre 8800 GTX 768MB	Game Score
4x AA/8x AF	630/1960MHz - 1350MHz (Shader)	70.37
4x AA/8x AF	630/1960MHz - 1450MHz (Shader)	70.93
	Calibre 8800 GTX 768MB	Game Score
4x AA/8x AF	650/2000MHz - 1404MHz (Shader)	73.13
4x AA/8x AF	650/2000MHz - 1512MHz (Shader)	73.75
Results Interpretation: Higher index values are better.

As the synthetic benchmarks are one thing, they don't always show a clear picture on the actual gaming performance and experience. So to that extend we use our own favorite game, Far Cry to see the difference while overclocking and tweaking the shader clock. Somehow it seems our configuration was not fast enough for the Calibre 8800 GTX OC edition, as the performance difference are quite low. Although keep in mind that we are use a very modest Intel Core 2 Duo E6300 which is overclocked to 2.8GHz and which is quite affordable. With a better CPU, you might see a more clear difference while playing Far Cry and certainly with HDR is enabled. But with the higher shader clock, the game left a much smoother impression on the game play...

Raven Software Quake 4

Gaming Results - 1600x1200 - E6300 2.80GHz - 2GB RAM
	Calibre 8800 GTX 768MB	Game Score
4x AA/8x AF	630/1960MHz - 1350MHz (Shader)	156.6
4x AA/8x AF	630/1960MHz - 1450MHz (Shader)	158.1
	Calibre 8800 GTX 768MB	Game Score
4x AA/8x AF	650/2000MHz - 1404MHz (Shader)	158.2
4x AA/8x AF	650/2000MHz - 1512MHz (Shader)	160.4
Results Interpretation: Higher index values are better.
Gaming Results - 2560x1600 - E6300 2.80GHz - 2GB RAM
	Calibre 8800 GTX 768MB	Game Score
4x AA/8x AF	630/1960MHz - 1350MHz (Shader)	80.2
4x AA/8x AF	630/1960MHz - 1450MHz (Shader)	81.4
	Calibre 8800 GTX 768MB	Game Score
4x AA/8x AF	650/2000MHz - 1404MHz (Shader)	83.5
4x AA/8x AF	650/2000MHz - 1512MHz (Shader)	84.7
Results Interpretation: Higher index values are better.

To conclude the advance tweaking and overclocking, we also wanted to see the OpenGL performance impact and we figured that Quake 4 would be our choice to gather our input. Once again we started with the default settings, tuned these up a bit in the various ways and noted down the performance impact. With Quake 4 the difference is a bit clearer with some extra frame rates to be gained. Indeed, it isn't that much, but what did you expect when we only could increase the core with an extra 20MHz, the memory with an extra 40MHz, leaving the shader clock as our only way to tweak up the performance. But in the end the main core and memory clock will steer the graphics card performance, with the shader clock controlling the stream processors. So a small increase, but which can't be neglected as it shows it will make a difference between the different brands who have released their pre-overclocked versions on the market.