Q9550 Voltage - Help needed

[NickelPlate]

Hey Folks,

I recently noticed that CPU-Z is reporting a core voltage of 1.568. I have the mobo on auto voltage control. Intel's website lists the VID range of 0.8500V - 1.3625V. In any case it seems the mobo defaults are running this thing way too high. I suppose I'd better go in and lower it manually but wanted input first from some OC'ing experts. It's been running perfectly stable at 3.4Ghz for more than a month now.

Thanks,

NP

 

[VirtualLarry]

Yes, that voltage is way too high for a 45nm chip. Max is 1.4v.

 

[AzN]

Yes, that voltage is way too high for a 45nm chip. Max is 1.4v.

1.4 volts is officially supported by intel but anyone can go higher if they wish long as the temperature isn't excessive.

But yeah OP you might want to lower it down manually.

 

[F1N3ST]

You can run 3.4ghz at stock voltage probably, I am..

 

[NickelPlate]

Thanks for the replies so far folks. I set the CPU voltage in BIOS to 1.425 which results in a 1.4V idle (cpu-z). I'm going to work downward from there. I'm currently 1 hour into Prime95 with no errors and temps are hovering around 50c for the hottest core (using coretemp).

Strangely, cpu-z and coretemp are reporting different core voltages, 1.320 and 1.300 respectively. I'm not sure which to believe. That seems like alot of Vdroop. Does my mobo suck?

Another thing which I can't seem to understand is that whenever I manually set my memory timings to the manufacturer specs (5-5-5-18) I get Prime errors in less than a minute. It runs much better when I let the motherboard auto configure the timing, which according to cpu-z is 5-6-6-15.

Adjusting the FSB to anything other than 400 results in errors and stability and crashing problems particularly in games. I suspect I have some sort of weird memory timing issue that is going on with the mobo's auto timing configuration.

I'm trying to get this worked out as I believe I can safely hit 3.6Ghz without overheating the CPU. Further comments are welcome.

Thanks,

NP

 

[AzN]

You could probably go much lower than that. Say 1.3. Keep testing orthos until you pass with lowest vcore. That's what I do.

Since you have P5B there's a pencil mod for vdroop. You just need a *B pencil. Just go back and forth with the pencil for vcore doesn't drop @ load.

Picture might be different depending what version of P5B you have so take caution and search the web.

My dual core wolfdale sits @ 4ghz @ 1.44volts. You could probably drop it as low as 1.25 after the vdroop mod. Keep your processor cool or go high like me. I only max load @ 57C and idle @ 38C 32C for both cores so temp wasn't a problem.

 

[NickelPlate]

You could probably go much lower than that. Say 1.3. Keep testing orthos until you pass with lowest vcore. That's what I do.

Since you have P5B there's a pencil mod for vdroop. You just need a *B pencil. Just go back and forth with the pencil for vcore doesn't drop @ load.

Picture might be different depending what version of P5B you have so take caution and search the web.

My dual core wolfdale sits @ 4ghz @ 1.44volts. You could probably drop it as low as 1.25 after the vdroop mod. Keep your processor cool or go high like me. I only max load @ 57C and idle @ 38C 32C for both cores so temp wasn't a problem.

Thanks Azn. Interesting mod. Hard to tell if that's a resistor or capacitor that's being shorted by the pencil mod. I would imagine a dab of solder on top of the component would accomplish the same thing. I'll have to check this out. As it stands right now I'm getting a Vdroop of nearly 70mV.

Regards,

NP

 

[zerocool84]

You can run 3.4ghz at stock voltage probably, I am..

Mine wasn't able to but for some reason my Q9550 was stock volted ridiculously low from what I've mostly seen, 1.12 and runs perfect @ 3.4GHz 1.175v.

 

[fffblackmage]

I would imagine a dab of solder on top of the component would accomplish the same thing.

Do NOT use solder. The point of using pencil is lowering the resistor's value, not shorting it.

In other words, soldering across the resistor drops the resistor's value to 0.
Using a pencil effectively puts another resistor in parallel, lowering the overall value of the resistor you're drawing on.

Do not use solder!

Also, resistors are usually black, caps are usually tan colored.

Edit: wait... that is the point right? lowering the resistor's value, not just simply shorting it?

 

[AzN]

Do not solder!

you could easily screw it up that's permanent. With a pencil it's none permanent and you can easily rub it off with rubbing alcohol.

 

[tweakboy]

That is a insane voltage ammount your lucky you didn't burn the CPU

Although Im sure it goes down but @ idle goes back to your 1.56

Anyhow to run this chip @ 3.4 you barely need to touch the CPU voltage. 1.37 or 1.39

Dont go any higher then that. Try for 3.6 , with that voltage.. let us know. gl,,

 

[tweakboy]

I just read about the soldering. Do not do that very dangerous. Search google and see if your mobo has a pencil mod... let us know

 

[NickelPlate]

That is a insane voltage ammount your lucky you didn't burn the CPU

Although Im sure it goes down but @ idle goes back to your 1.56

Anyhow to run this chip @ 3.4 you barely need to touch the CPU voltage. 1.37 or 1.39

Dont go any higher then that. Try for 3.6 , with that voltage.. let us know. gl,,

I must have a really lousy chip or crappy motherboard or both. I hope it didn't get damaged because it's very unstable at 1.3625V. I get BSODs, crashing etc. Running at 1.4250V @ 3.4Ghz it's stable for games and apps, but I still get Prime errors within an hour.

 

[Kenmitch]

The q9550 I just sold had a vid of 1.150v to get to 3.60ghz it needed 1.185v to get to the big 4.02ghz it took 1.30v. I stopped at that.

Wonder what it would have done at that voltage

 

[soccerballtux]

1.4 volts is officially supported by intel but anyone can go higher if they wish long as the temperature isn't excessive.

But yeah OP you might want to lower it down manually.

Depending on what kind of values you mean when you say "higher" you most certainly should NOT go to 1.55v if temps aren't excessive.

 

[AzN]

I must have a really lousy chip or crappy motherboard or both. I hope it didn't get damaged because it's very unstable at 1.3625V. I get BSODs, crashing etc. Running at 1.4250V @ 3.4Ghz it's stable for games and apps, but I still get Prime errors within an hour.

That's why you need to vdrop mod. You are basically @ 1.37 once the cores are loaded.

I was having hard time getting 3.85ghz without vdrop @ 1.45volts but once I did the vdrop mod I got to 4ghz at the same voltage.

That sucks you can't get prime stable. Are you using the latest bios? How's your temperature?

I've been told quad has hard time on the p35 chipsets and older when overclocking but there's no proof. I've found people on the net get 3.8ghz with a P5B so... You might just have a high vcore chip. Get it stable don't run it when it's no prime stable.

 

[AzN]

Depending on what kind of values you mean when you say "higher" you most certainly should NOT go to 1.55v if temps aren't excessive.

I've seen people go higher on water and run it fine. There's no rule for voltage it's just that if you want to be safe for prolong use.

From what I gather 1.45volts is the max these wolfdales should be used for longevity.

 

[NickelPlate]

That's why you need to vdrop mod. You are basically @ 1.37 once the cores are loaded.

I was having hard time getting 3.85ghz without vdrop @ 1.45volts but once I did the vdrop mod I got to 4ghz at the same voltage.

That sucks you can't get prime stable. Are you using the latest bios? How's your temperature?

I've been told quad has hard time on the p35 chipsets and older when overclocking but there's no proof. I've found people on the net get 3.8ghz with a P5B so... You might just have a high vcore chip. Get it stable don't run it when it's no prime stable.

I am using BIOS 1806. I see Asus has released an 1807 which supposedly fixes HD size detection problems but nothing about updated CPU support.

Temps are right around 50°C while running prime. Idle temps are in the mid to high 20s. I ran prime again overnight last night and two cores made it through, and the other two failed after about 3.5 hours. I guess I have more work to do. I'm hoping to get the time this weekend to pull out my motherboard and brave the Vdrop mod.

 

[NickelPlate]

Finally got around to doing the pencil mod. So far worked better than I ever hoped. I have absolutely zero Vdroop now (according to CPU-Z) and have a stable 3.5Ghz at 1.35V. Thanks Azn for the tip.

Regards,

NP

 

[Shmee]

awesome! Now, push farther till you get errors, then back off. Keep in mind, at high FSB, you will need to adjust the NB voltage, especially with your 8 GB of RAM. 4 Sticks puts more strain on the MCH. Also, make sure your ram is at appropriate frequency/voltage/timings.

Remember, not all failed OC's are vcore related. Keep FSB levels, memory, and NB in mind too.

 

[Daedalus685]

Never realized there was such a massive range of VID for the q9550, mine is 1.15V, takes 1.2 (read in cpuz, 1.22 or so in bios) to reach 3.6ghz comfortably.