First attempt at overclocking (Q9450)

[dimaevt]

Hi, I have a question about overclocking a Q9450 (intel quad core @ 2.66ghz). I want to overclock this processor by only increasing the FSB and nothing else. Since the stock GHz on it is 2.66, what is a safe, stable MHz speed I can obtain on this processor without getting a new fan, or anything else.

My RAM, motherboard and power supply:

1000 watt power supply
nVidia 790i ultra SLI motherboard (supports 1600 FSB)
4GB Patriot EP+ Dual Channel DDR3 SDRAM at 1333MHz - 2 x 2048MB

Also, I watched a video on YouTube that you can overclock the FSB via the nVidia control panel (ntune?). Since this is my first time overclocking, I want to start out small and safe before touching voltage, or upping the FSB further than what the stock voltage can handle. Having said that, what is a safe and stable speed that my processor can handle, given the mobo/ram?

 

[ausch30]

Wow that PSU is overkill

That CPU runs fairly cool and I think you should be able to get at least 2.8 out of it before needing to up the voltage. Speed isn't what causes heat voltage does so just raise the FSB one step at a time checking for stability until it becomes unstable then back it down one.

 

[dimaevt]

This may sound stupid, but how do I check for stability? And what constitutes stability and unstability? And how much is "one step at a time"? I'm planning on doing all this via the nVidia control panel, not BIOS (only increasing the FSB).

As for my PSU, it's to power dual 9800gx2s, however this processor might bottle neck them, so I want to overclock the processor.

 

[ausch30]

The CPU your planning on has a 1333FSB which is 333 quad pumped. Just set it at 334, boot up, and run Orthos or Prime95 for a few hours. Total stability is when you can run a stress test for about 24hrs without issue.

 

[dimaevt]

Also, my cooling is completely stock. If I only increase the FSB to where I get a flat 3.0GHz, can my cooling handle that?

 

[ausch30]

It should be able to but it might not be enough. As I said just start at stock and go up slowly and when you either reach instability or temps start getting a little high just back it down a notch and your done.

 

[dimaevt]

Much appreciated. Thanks ausch.

 

[MajinInsane]

I just overclocked my Q9450 to 3.2 GHz, although i must say i threw out that cheap OEM heat sink and went with the CoolerMaster TX2 (zipzoomfly had it with a rebate). It runs quite stable, although ive noticed that the ASUS p5e board does well to adjust voltages and what not when you manually run up the FSB clock speed. 3.2 will also put the FSB and DRAM speeds at 1:1 too which is nice if you dont want to overclock any DDR2-800MHz memory.

 

[dimaevt]

When upping the FSB, do you HAVE to mess with the ram timings, or any other settings in BIOS as well?

 

[ausch30]

Your RAM is running 2:1 at stock with your CPU at stock so you don't have to change anything else. The board will find the correct divider to work with your RAM as you increase the FSB.

 

[sciggy]

I have a Q9450 and I'm at 3.0ghz @ stock volts running mid to high 30s for temp. Pretty easy to get there. Going to jump to 3.2ghz to run an even 1:1 ratio with ram (400mhz FSB x 8.0 cpu multi = 3.2ghz)

 

[belgarath420]

Hi, I'm trying to OC my new Q9450 on an EVGA 780i. Need some advice . I can't get it to boot @ 3.0 . Any one with this same combo? Suggestions?

 

[cpampine]

Hi, I'm also having my first attempt at overclocking this processor with 1600 FSB.

The minimum BIOS VCORE to allow stability on my system is 1.325. This is with all other voltages set to normal and no other adjustments, except for the RAM which was volted and timed according to manufacturer's specs (2.1, 4-4-4-12)

During Prime95>large FFT I get core voltage readings of 1.264 - 1.28 from CPU-Z and HWMonitor. This looks a lot lower than my BIOS vcore. Also Real Temp is showing an unusually narrow temperature margin between idle and full load - less than a few degrees... ?? 63,59,57,55c

Ultimately, this isn't stable Prime95 has a fatal error on the second core after about 10 mins. I'm going to try upping the vcore another notch or two. I'm curious to see how CPU-Z reports. I know when I allow the BIOS to auto volt vcore at this overclock it sets it up around 1.34 and that reflects accurately in CPU-Z. Temps there are too hot, and I wonder if longevity might be compromised even though it is withing operating specifications from Intel.

Antec NeoPower 550 | Antec P182 | GA-EX38-DS4 | QX9450 | ZALMAN CNPS9500 | MUSHKIN XP 4GB DDR2-800 PC-2 6400 SDRAM | Sapphire Toxic Radeon HD 3870

 

[shadowthor]

As long as temps are below 70 under load, it is fine. Did you try doing small ffts in prime95 to just test cpu stability.

 

[cpampine]

No, I haven't tried small ffts since I experimented initially with the BIOS autovolts - when i stability tested then, everything was pretty stable but I was a bit wary of the 1.34 showing up in CPU-Z. If I understand correctly, that reading in CPU-Z is effected by vdroop and it would have actually been higher in the BIOS (if I could see - when it autovolts you can't). I'm really scared of setting my vcore too high - I don't want to compromise the longevity of the processor, and I would be out $$$ if I actually damaged it or my new mobo somehow.

I thought the large ffts would be the best to get stable first based on some of the OC guides I've looked at. I'm worried that I'm not doing something right because I've read in a few places that users are getting 3.2GHz easily on 1.175, but that seems impossible to me! I wish I could figure that one out. Maybe I should crunch some simple numbers to determine if I can get a 1600 FSB some other way, but nobody has endorsed that method yet that I am aware of, and it seems that the 400 w 1:1 is the preferred method.

Is it possible that I might achieve a stable 3.2 at a lower vcore by increasing the voltages for the MCH (NB) and the FSB? Or perhaps figuring out something using a 333 (or something) 2:1 setup needs more consideration?

 

[shadowthor]

it is possible to get 3.2, experiment and see, btw nvidia chipsets are harder to overclock in my opinion.

 

[iluvgillgill]

hm...what interest to me is when the 9500 will reach its limit.sorry to say you should have got something else.im honest.the better 9700 wouldnt even let me clock 3.6ghz on Q6600.

 

[level101]

I have a Q9450 @ 3.4GHz (425 x 8) CPU Voltage in BIOS is 1.225v. I'm using Zalman CNPS9500

 

[cpampine]

Prime95 blend running stable after increasing the MCH by .100. NB is rather hot with the case open, but not unreasonably so. I can hold my finger on the hs for a ten count, although I cook and my asbestos finger tips might not be the best gauges!

Once I finish stability testing with my new MCH NB settings, I'm going to significantly reduce the vcore and by doing so gain some much needed piece of mind... I hope.

 

[cpampine]

i still want to torture test for the rest of the night, but it would appear as though my BIOS vcore of 1.325 is the best I can do right now. With load temps during Prime95>smallFFTs around 60c on my hottest core, I'm not exactly enthusiastic about my o.c. - I was really hoping to get 3.2 on stock voltage with this chip.

What should I do?

RMA and hope for a better chip?

Re-examine how I'm overclocking?

 

[Paris Lee]

first you can try to increase FSB to 400 in bios of your motherboard
i guess the system will be ok and your cpu will be stable at 3.2ghz
and then just increasing the FSB once a bit till can not start up your system
,and slow down your cpu's speed also by FSB till you can finish 3DMark stablely or paly 3d gaming for huors
maybe after one day' try you will find your cpu's best speed
haha maybe tired but gains more

 

[iluvgillgill]

24hours torture is 100% guarranteed stability.

 

[dimaevt]

I've now upgraded to liquid cooling, from the stock crappy fan cooling. My computer should come in a few days. I am hoping to hit 3.2ghz with everything on stock.

 

[cpampine]

the system is stable, but sadly not 24 hrs prime stable, and with the voltage as high as it is already , i'm not happy. i'm gonna think about it for one more day, but i'm probably going to RMA.

 

[shadowthor]

Its going to be hard to rma a chip. They might test it and say it works fine.