Q9450 overclocking, need a little help.

Justintoxicated

[H]F Junkie
Joined
Apr 10, 2002
Messages
14,522
So I have been scewing with this processor for over a week now and I just can't get it stable with any overclock.

With Motherboard and CPu voltages left stock (no added voltage) I can boot up at 3.2 GHZ with 2 sticks of ram in single channel (slots 1&2). If I put the memory in slots (1&3) to run dual channel mode I can't pass Intel CPu Burn indicating the system is not stable, no amount of added voltage to CPU or MB fixes the issue. I have 4 sticks total so I really need to get dual channel memory working. If dual channel is not working does this indicate a problem with the Motherboard or the CPU?

I also heard that adjusting my GTL values may allow me to run dual channel? Can someone please explain this process or link me to a tutorial I'm having trouble finding info on what to do for my gtl values. I'm ripping my hair out not knowing which component has the problems. I'm about to put my old Opteron 165 back in and part out this new systemas I can't seem to get dual channel memory working or a decent overclock.

I can't boot up at anything over 400FSB. adding voltage to my setup seems to have no effect on the system what so ever (To NB, CPU, VTT), if anything adding voltage seems to make it less stable or not want to boot.

Even with the CPU underclocked to 2.4ghz CPu voltage at 1.36-1.4v, NB at 1.4v and VVT at bumped around a bit I can't do anything over 400 FSB or run dual channel for the memory.

Thanks in advance,
 
That's rather odd....

What motherboard are you running ?

The fact that you can't get dual channel running suggests to me an SPD problem rather than an issue with the cpu/motherboard.

I can't speak to GTL adjustment , I havnt messed with it.

I assume you are running the latest bios , and that your ram will pass memtest @ stock speeds ?
 
yea I tested teh memory and it is fine. Although I can't test the memory at stock speeds because I have not been able to reach 500 FSB (PC2 8800).

the SDp is set correctly.

Aparently it is a GTL adjustment issue. I was able to get it to 3.0 GHZ stable but it took WAY too much voltage to get it there.

1.36v CPU 1.35v VTT GTL adjustments for VTT based on G0 CPU (not the one I have). NB 1.5v

Yet I can easily get prime and even better Intel CPU Burn stable off default voltages using single channel. I would highly suggest everyone start using the Intel CPU burn to test for stability instead of prime for intel chips, my CPU gets about 15C hotter on water than when I run prime and it takes ALOT less time for testing, like 10 minutes instead of 24 hours....

I'm running a 602 beta bios for the UT eddition of the MB since the boards are physicaly the same. motherboard added to sig.

P8020945.jpg
 
I dont think you will be able to reach 500FSB with that board. 460-470 is a stretch.
I left you some of my settings for the UT X48 on XS. Hope it helps.
Copy/paste from there :D:
You shouldnt need so much vtt.
In my UT X48 if i set 1.34vtt it gives around 1.33measured. And with that i could do 3.52 and 86 91 96 gtlrefs which translated into mV in 0.853,0.853 and 0.896, so around 64% for the cores and 67% for the nb. This with 1.3125vcore(bios) which is actually around 1.347(idle) and 1.328(load) if i remember correctly. And for higher FSB you should go much higher with the NB voltage, i needed for 440FSB around 1.579(bios) which is 1.55 measured.
 
I dont think you will be able to reach 500FSB with that board. 460-470 is a stretch.
I left you some of my settings for the UT X48 on XS. Hope it helps.
Copy/paste from there :D:
You shouldnt need so much vtt.
In my UT X48 if i set 1.34vtt it gives around 1.33measured. And with that i could do 3.52 and 86 91 96 gtlrefs which translated into mV in 0.853,0.853 and 0.896, so around 64% for the cores and 67% for the nb. This with 1.3125vcore(bios) which is actually around 1.347(idle) and 1.328(load) if i remember correctly. And for higher FSB you should go much higher with the NB voltage, i needed for 440FSB around 1.579(bios) which is 1.55 measured.


Does not seem to be the case for me, I can't even boot up with a 1.3vtt, and anything under the settings below Hard Locks Blue Screens running Intel CPU Burn. Even now I just had the PC shut down using the settings for 3500 MHZ.

3400mhz 1.36v FSB 425 NB 1.5v Vtt 1.40x GTL 1/2-120 0/3-110 NB-130
3500mhz 1.39v FSB 438 NB 1.5v Vtt 1.463 GTL 1/2-125 0/3-115 NB-135
3600mhz 1.39v FSB 450 NB 1.5v Vtt 1.503 GTL 1/2-130 0/3-120 NB-140

Also I'm noticing some multiplier flux @ 3500 mhz it changes between 6x and 8x constantly drasticaly changing the clock speed. Under load it seems to stay at 3500 though? What can I do to fix this? And which of the above values are safe if any?
 
You have the EIST option enabled in BIOS and that is changing your multiplier from 8x to 6x when your cpu is idling.
As for the voltages, bare in mind Intel states 1.45vcore and also 1.45vtt as absolute maximums. For long term usage i wouldnt put more than 1.38 from both.
As i stated:
Q9450+DFI UT X48: 3.52GHz @ 1.3125(actual 1.347measured) and 1.33vtt
Q9450+Asus P5Q Deluxe(P45): 3.8GHz @1.38125 and 1.28vtt
 
You have the EIST option enabled in BIOS and that is changing your multiplier from 8x to 6x when your cpu is idling.
As for the voltages, bare in mind Intel states 1.45vcore and also 1.45vtt as absolute maximums. For long term usage i wouldnt put more than 1.38 from both.
As i stated:
Q9450+DFI UT X48: 3.52GHz @ 1.3125(actual 1.347measured) and 1.33vtt
Q9450+Asus P5Q Deluxe(P45): 3.8GHz @1.38125 and 1.28vtt

So basicaly I shoudl jsut run the CPU at stock speeds? Right now I have the CPU at 1.39 and VTT at 1.513 Anything less I am not stable. I don't know why. I could back down to 3200x4 but why even overclock at all with such a poor boost? I might have to swap the CPU I guess or just run it as it.

Everyone is telling me my Votlages are too high. So believe me when I tell you I have tried to lower them but the system becomes unstable! I found the setting in the bios that was causing the multi change it was C1E function.

desktop3600.gif


I'm so frustrated with this CPU needing high voltage im about to say screw it and just run it as it is. But everyone has me worried now that I'm killing an expensive piece of equipment :(

@ 1.33vtt I really can't even achieve 3.4 ghz, in fact I don't think I can even achieve 3.2 ghz.... However sticking the ram in single channel (2x2gb) I can run 3200 MHZ with default voltages (CPU @ 1.18v and Vtt at 1.0v and NB 1.2v) no GTL changes.

But for dual channel GTL and VTT is needed and at crazy voltages. I do not understand. Should I just clock back to stock speeds? I bought intel and this whole $$$ watercooling system to overclock....Extreamly Disapointed. can someone help me decide what I should do?

Is it safe to measure the votlages with regular Fluke DMM probes or do I really need the special SMD probes?
 
What voltage is your RAM running at? I don't see that listed anywhere in the previous posts.

On my X48 motherboard my RAM was defaulted to 1.8v. Now that made it stable and I could run Dual-channel but I wasn't able to do any overclocking until I corrected that as my RAM is 2.1-2.2v rated. As soon as the RAM was getting the power it wanted my overclock immediately stabilized.

-V
 
Fluke DMM probes should do fine if you do not short something.
I had the same problem as you with the UT X48(i wasnt able to pass over 3.55 without putting a lot of voltage) and i thought i had a lazy cpu. Turns out that the mobo is really limited to "low" fsb with quads.
You could either switch your mobo(this is my recomendation) with something based on P45 or with a faster X48 like the Rampage Formula or decrease the frequency, perhaps at 3.4 or 3.5 it will let you run it without seriously overvolting.
Oh, and try to increase the NB voltage to 1.579 or higher and put a fan or provide adequate cooling to the NB. See if that helps. Mine asked for 1.55volts(1.579 in bios) for 440FSB.
 
What voltage is your RAM running at? I don't see that listed anywhere in the previous posts.

On my X48 motherboard my RAM was defaulted to 1.8v. Now that made it stable and I could run Dual-channel but I wasn't able to do any overclocking until I corrected that as my RAM is 2.1-2.2v rated. As soon as the RAM was getting the power it wanted my overclock immediately stabilized.

-V

In bios the Mem is set to just over 1.9v however, from MB it was 1.87. The memory is PC1100 and sold as a kit, so it should be perfectly fine at these voltages, otherwise something is wrong.

Fluke DMM probes should do fine if you do not short something.
I had the same problem as you with the UT X48(i wasnt able to pass over 3.55 without putting a lot of voltage) and i thought i had a lazy cpu. Turns out that the mobo is really limited to "low" fsb with quads.
You could either switch your mobo(this is my recomendation) with something based on P45 or with a faster X48 like the Rampage Formula or decrease the frequency, perhaps at 3.4 or 3.5 it will let you run it without seriously overvolting.
Oh, and try to increase the NB voltage to 1.579 or higher and put a fan or provide adequate cooling to the NB. See if that helps. Mine asked for 1.55volts(1.579 in bios) for 440FSB.

The stock probes on the Fluke are too large to get the GTL readings. I bought some new probes from Frys for $32. They have very fine point and are insulated up to the tip and the tips are adjustable up to 3" long!


I'm not copnvinced the Rampage is any better. Alot of people leaving the VTT on auto. Someone decided measure his and it was boosting to 1.53v (measurement from the MB). So is it really better? Also how the hell would I water cool the asus? I would have to spend another bundle of money on custom sinks because they use heatpipes for everything so everything is connected...(main reason for choosing the DFI board)

My NB is liquid cooled (see picture above)

After manualy adjusting the GTL's with the probe I am able to achieve 3.4GHZ w/ 4GB ram in dual channel stable on the DFI board @ 425 FSB with only 1.33Vtt (1.34 measured). However with 8GB ram in place I am once again unstable according to IntelBurnTest, but I'm not getting lockups at least...

What is the advantage of the P45 board, these were not out when I bought my X48 a couple weeks ago. I remember X48 was supposed to be 65nm but I'm not sure if it is. I'm also not sure it is worth taking a $250 loss to gain another 100 MHZ of FSB even if the NB is 65nm on the P45. Reviews show the P45 is not as fast in games, esp when using crossfire (which I may purchased eventualty)?

Basicaly I jsut bought this setup and have not even had it running right yet, and the solution is to sell it all? WTF
 
Well even though my PC can run prime for 12 hours, IntelCPUBurn fails in about 40 min.

I decided to bump my mem slightly over its rating at which point Intel CPU burn failed after 4-5 hours.

Just bumped it another notch. I'm probably 100% prime stbale at this point, but since a better stress test is out I want to meet the requirements of that one instead. it also test memory much harder than prime. My Memory is NOT overclock but seems to require a little more voltage than it is speced for.
 
Huh?

Well I still have not achieved stability at 3400MHZ...I seem to be prime stable but I can't pass more than 4 hours of IntelBurnTest...(30 passes)
 
Back
Top