Categories


Πέμπτη 20 Ιανουαρίου 2011

Benchmarking and Overclocking Q9650 Part 2

After our previous feature on overclocking the Q9650, we've been wondering how much we can push this CPU. We won't be bothering with water cooling solutions, as we want to see how much we can get with the air cooling solutions. So, time to see how much we can push the Q9650 on air.

We used a Gigabyte UD3LR EP45 motherboard with 4GB DDR2 Memory Kit from Kingston, a 1000W PSU from Coolermaster and a GTX295. As we said in our previous feature, in order to achieve a 4.0Ghz frequency we needed 1.45V VCore. That was before our latest tweaks. After messing with the CPU Termination, MCH Core and MCH Reference we managed to achieve 4.2Ghz with 1.4125V Vcore. This had a huge impact in our temperatures, so we strongly suggest not raising your Vcore above this value. Still, 1.4125V is quite high and you will definitely need a good CPU cooler.

Temperatures were averaging around 33-38c in idle mode and around 70c in Prime95's stress mode. The temperature junction of the Q9650 is 105c, so don't be afraid if you get these temps, as most applications won't stress your system as much as Prime95 does.



But what about the performance gains? Is it worth overclocking the Q9650 this high? The answer is possitive. In our benchies we experienced noticeably improvements and in some cases we witnessed 10fps increase. All of them were performed in 1080p. Going from 3.6Ghz to 4.2Ghz, we gained 1500 points in '3D Mark Vantage'. In 'The Last Remnant' we saw the biggest boost which was almost 20fps. In 'Cryostasis' we gained 10fps, in 'HAWX2' 15fps and in the second benchmark of 'Lost Planet 2' 7fps. We also saw a minor increase in 'SF4' and in the fixed benchmark of 'Resident Evil 5'. All in all, we can see an increase in every benchmark. This also means that we were CPU limited, even with a Q9650 at 3.6Ghz.

So there you have it. If you can push the Q9650 to 4Ghz or higher, go for it as you will see gains in almost every game. Pay attention though and monitor your CPU temperature, as you might damage it if you don't have a good and effective CPU cooler. Enjoy!



2 σχόλια:

  1. It's good thinking that you used a dx9 card, based alot on cpu performance than a dx11 graphic card will.
    However to see true cpu results in 3dmark vantage i would suggest a non phys-x card, as (from what i have seen)the hardware phys-x chip affects the cpu benchmark tests alot. I refer to pure cpu benchmark parts of 3dmark vantage (like the crashing dissasembling planes test )

    ΑπάντησηΔιαγραφή
  2. The PhysX settings from Nvidia's control panel were set to CPU when we run 3D Mark Vantage. Which basically means that the CPU handled all the PhysX calculations and not the GPU ;)

    ΑπάντησηΔιαγραφή