As always here on mods-n-clocks an article covering the overclocking of a processor follows in the footsteps a few weeks after the processor’s review. So here to compliment the Q9550 review is an overclocking article that aims to provide an example of how far you might be able to overclock the Q9550 with just fairly standard air cooling.
I’ve now had the Q9550 in my possession for a good few weeks and that’s allowed me enough time to get my teeth into overclocking it. For the majority of enthusiasts out there cooling their CPUs with dry ice of N2 is not a feasible option. Water cooling is however much more widespread and becoming more so as retailer sell a greater range of beginner friendly water cooling kit. This still doesn’t come close though to the number of enthusiasts and overclockers who still want to run their systems with a simple but effective air cooling solution. This article is aimed at those people and as you’ve read down this far there’s a good chance that you’re one of those people.
Like any CPU it would be silly to consider the Q9550 as a component on its own, as it is entirely dependant on other components to work. It therefore makes sense that these other components will also effect how well it will overclock so to start with we need to at least note what the specs of the whole system are.
I been running the Q9550 on my ASUS P5E3 Deluxe motherboard based system which has the following system specs:
- Motherboard : ASUS P5E3 Deluxe
- Memory : OCZ Reaper DDR3 1333MHz
- CPU : Intel Q9550
- CPU Cooler : Akasa Evo 120 (modified with a Nexos Real Quiet 120mm PWM fan)
- Other Cooling : Hiper Hiperflow fans intake x2
- Power Supply : Kingwin Mach 1 900W
- Hard Drive: Samsung 250GB SATA2
- Case : Coolermaster Centurion 532
The most influential system spec amongst these is the motherboard, using a different motherboard can greatly affect what maximum overclock you can achieve. Memory is not quite so important as you should be able to keep the memory at or below its rated speed by adjusting the memory divider. Obviously having a stable and capable power supply unit is crucial to obtaining and maintaining a decent overclock but one good stable PSU will provide much the same overclock as another stable PSU. Cooling is similar to the PSU, it is important but as long as its up to the job there are many options that will perform much the same.
3.70GHz – 23th September 09
This overclock is 31% above the standard clock speed and not that far short of the 4GHz level that many people are aiming for with their overclocks at the moment. I’ve confirmed this overclock as stable by running Prime95 on both the small FFT and Blend tests for greater than 8hrs on each. Often people refer to being 24hrs stable on these tests as the benchmark for stability but I’m yet to see an 8hr+ failure so if its 8hrs+ then I’m happy it would also have been 24hrs+ too.
The screenshot from CPUz above shows the overclock and its basic settings. I ran with the default multiplier of 8.5 and increased the fsb up to 436 which gives the clock speed of 3706 MHz (8.5 x 436). I had to increase the voltage to make this work and CPUz shows an increased voltage of 1.232V. This voltage was achieved on the ASUS P5E3 Deluxe motherboard by setting a CPU core voltage of 1.250V, clearly there is a discrepancy between the set voltage and the read voltage. The fact that I’m running ‘load line calibration’ contributes to this voltage mismatch and these levels of voltage it doesn’t really matter so much which is right just as long as it’s stable and the temperatures are kept down to a reasonable level.
If I run the Q9550 with the RAM set to lowest divider option then I can push the overclock up to 3.74GHz and remain fully stable but the system then looses stability when I move the divider back up to set the RAM to it’s rated spec of ~1333MHz. So with the choice of 3.70GHz/1308MHz or 3.74GHz/881MHz it’s got to be the 1st option.
With this 31% increase in clock speed the CPU can complete the 1M Super PI test in 12.852s, 7-Zip compression benchmark with an average of 10628 KB/s, decompression with an average of 128879 KB/s and the RealTemp XS Bench with a score of 1230. Lets add these to the CPU comparison charts drawn up the other week…
SuperPI 1M (Lower is better)
A nice reduction in time taken to complete the task seen here. Which works out to be approximately 29%, clearly in the SuperPI test you get back in performance what you put in by overclocking.
7-Zip data speed – Compression (Greater is better)
The 7-Zip tests are much more like real life computing tasks so it’s nice to see a significant increase here as well. The overclocked Q9550 breaksthrough the 10,000 KB/s barrier. If you do the sums you’ll also notice that it’s a whole 3GHz Pentium 4 (630 Prescott) faster than it is at stock speeds. Of course with this being more like real life we don’t expect the performance increase to necessarily mimic the clock speed increase and as such the increase of around 21% is reasonable.
7-Zip data speed – Decompression (Greater is better)
Unfortunately the Q9550 overclocked isn’t so far ahead when it comes to the decompression test with an increase of just 11%. Of course this is a performance increase for zero cost so it’s still not something to complain about.
RealTemp XS Score (Greater is better)
As you can see the RealTemp Score increases handsomely too. Here the overclock yields a 30% increase in performance which is very close to the 31% clock speed increase.
I must admit that I’m a little disappointed with the overclock that I was able to squeeze out of this CPU. Having managed 80%+ out of my E2160 I was at least hoping to push this chip past the 4GHz mark although obviously wasn’t expecting any where near 80% as this Q9550 is a much faster chip to start with.
Even so, gaining an additional 31% for free and with a small amount of work really isn’t to be sniffed at and well worth having. I will of course continue overclocking this chip and there are so many options to play with in the P5E3 Deluxe BIOS that I’m confident of getting even more sometime soon – I’ll be sure to let you know when I do of course!