Q9400 ovcerclocking

[Bman123]

Well I am gonna give it a try but I want some help form the pros here.
What is the max voltage for the Q9400?

I went from 333 to 400 at stock volts and it didn;t like it in intel burn test.I am guessing more volt to the cpu but i dont know how much to give it.

Hopefully you guys can help me get this thing stable to around 3.6 or so

 

[chiefy]

Stock volts? Nah won't work. Not sure what the stock is, but it won't be enough.

Can you tell us?

 

[Zero82z]

What is the max voltage for the Q9400?

1.4V.

 

[Bman123]

I heard the max voltage was 1.3625

Do you think I should just start at 1.3625V and go from there slowly lowering the voltage til I find where it is stable at?

 

[Zero82z]

I heard the max voltage was 1.3625

The normal voltage limit for Intel CPUs is not explicitly stated. The absolute maximum safe voltage is 1.4V.

Do you think I should just start at 1.3625V and go from there slowly lowering the voltage til I find where it is stable at?

You could do that, or you could start at a low voltage and gradually raise it until you reach stability. Either method works fine.

 

[saedrin]

Eh. I find each chip to be different. I've been at 1.5v on a q6600 since launch and I haven't had any issues.

 

[skitlets]

Eh. I find each chip to be different. I've been at 1.5v on a q6600 since launch and I haven't had any issues.

The Q6600 is built on the 65nm process. The newer 45nm (Q9xxx) procs are much more sensitive to voltage.

 

[saedrin]

Oh well now I feel silly.

 

[Bman123]

I will fuck with it when I come home from work Around 6pm tonight will be when I start to ask some questions.This cpu is not fast at stock clocks.I can see when I run the RE5 benchmark that there is alot of ground that can be made up with this thing.

Hopefully I can hit 3.6

 

[Frobot]

I heard the max voltage was 1.3625

Do you think I should just start at 1.3625V and go from there slowly lowering the voltage til I find where it is stable at?

You'll save time if you go up instead of down.

 

[ivandurago]

I will fuck with it when I come home from work Around 6pm tonight will be when I start to ask some questions.This cpu is not fast at stock clocks.I can see when I run the RE5 benchmark that there is alot of ground that can be made up with this thing.

Hopefully I can hit 3.6

You might hit 3.6, but with a AC7 your temps might be high.

I have a q9450 running at 3.33 @ 1.28v, and temps are ~38-40idle, ~60-62 prime load, and I run a AC7 as well.

Also I am not sure a q9400 can always hit 3.6, shooting for ~3.2-3.4 might be more realistic w/ the AC7. I definitely could be wrong though.

 

[Zero82z]

You might hit 3.6, but with a AC7 your temps might be high.

I have a q9450 running at 3.33 @ 1.28v, and temps are ~38-40idle, ~60-62 prime load, and I run a AC7 as well.

Also I am not sure a q9400 can always hit 3.6, shooting for ~3.2-3.4 might be more realistic w/ the AC7. I definitely could be wrong though.

You're worried about temps in the low 60s with a CPU that can go up to 100C? That doesn't make sense to me. There's plenty of headroom left in both your CPU and the OP's. Q9400s can often hit 3.8GHz and higher on air.

 

[Zoson]

You're worried about temps in the low 60s with a CPU that can go up to 100C? That doesn't make sense to me. There's plenty of headroom left in both your CPU and the OP's.

This.
Google is starting the initiative to have the temperature of datacenters raised for a reason.

Computer hardware safely operates at significantly higher temperatures than rated for. For instance I have a now 8 year old AthlonXP 1600+ running at 2.7Ghz with 2.1v(i think 1.4 was stock?) for the entire 8 years. It tops out at about 80C and idles around 70C.

The E8600 listed in my sig runs at 95C full load and about 55C idle, and I'm not worried. The point is my cooler removes heat faster than the cpu generates heat at that load/temperature, so it never reaches TJ max and is completely stable.

Hell, even Kyle's article comparing the TRUE to the corsair self-contained watercooler illustrates this.

 

[Kato1144]

Hey buddy with you voltages you may need to increase your PLL and VTT voltage, but be careful no to push them two in particular to far, i think 1.2 -1.3 is ok for you VTT but 1.4 and up is at your own risk, as of the PLL voltage, i keep reading horror stories about people killing there expensive 45nm chip quad or duo.

As for me I only increased mine 0.1v each for now and it seems to be working, I'm going to try and lower the voltage on the PLL after i achieve fully stable clock on my E8400.

This.
Google is starting the initiative to have the temperature of datacenters raised for a reason.

Computer hardware safely operates at significantly higher temperatures than rated for. For instance I have a now 8 year old AthlonXP 1600+ running at 2.7Ghz with 2.1v(i think 1.4 was stock?) for the entire 8 years. It tops out at about 80C and idles around 70C.

The E8600 listed in my sig runs at 95C full load and about 55C idle, and I'm not worried. The point is my cooler removes heat faster than the cpu generates heat at that load/temperature, so it never reaches TJ max and is completely stable.

Hell, even Kyle's article comparing the TRUE to the corsair self-contained watercooler illustrates this.

Buddy, I know the chips have up to 100oC but at 95oC at load your pretty close to burring the chip, and you probably should consider thermal degradation to your chip, you should try to stay below 90 at max load even then your core senors are not giving you a 100% accurate value there is also speculation that each chip model may have different TJ max's.

Anyway I found i nice little read on the topic here, I hope you find this informative.

BTW I'm not trying to tell you how to run your rig, I'm just trying to save you on replacing that expensive chip.

-Kato

 

[Zero82z]

at 95oC at load your pretty close to burring the chip

No, you are not. The CPU will shut itself off before you get anywhere near a temperature that could cause damage to the CPU. It is impossible to damage an Intel CPU with high temperatures, period.

there is also speculation that each chip model may have different TJ max's.

That's true. However, since temperatures are computed based on the distance to Tjmax, the actual Tjmax value doesn't actually matter very much. It's not like you can accidentally push your CPU to a temperature above the Tjmax, since once the distance to Tjmax gets near 0, the CPU's thermal protection will kick in regardless of what your temperature-monitoring program is set to.

 

[Zoson]

No, you are not. The CPU will shut itself off before you get anywhere near a temperature that could cause damage to the CPU. It is impossible to damage an Intel CPU with high temperatures, period.

That's true. However, since temperatures are computed based on the distance to Tjmax, the actual Tjmax value doesn't actually matter very much. It's not like you can accidentally push your CPU to a temperature above the Tjmax, since once the distance to Tjmax gets near 0, the CPU's thermal protection will kick in regardless of what your temperature-monitoring program is set to.

If you have thermal control on and you hit Tjmax, all that happens is your multiplier drops to 6 and your voltage down to minVID. No crash, no shutdown, just a temporary loss in speed(and a huge reduction in temperature). Works quite like speedstep.

 

[Zero82z]

If you have thermal control on and you hit Tjmax, all that happens is your multiplier drops to 6 and your voltage down to minVID. No crash, no shutdown, just a temporary loss in speed(and a huge reduction in temperature). Works quite like speedstep.

Not exactly. First of all, thermal throttling will happen regardless of whether or not thermal control is enabled. Thermal control is a BIOS feature and has nothing to do with the thermal protection mechanisms built into the CPU, which cannot be disabled at all. Throttling also begins a little bit before the temperature hits Tjmax (usually around 97-98C). And it's true that it will only drop down the CPU's speed, but if the temperature continues to increase, there is a thermal shutoff point at a temperature above Tjmax (that temperature is not specified in Intel's datasheets).

 

[Frobot]

If you have thermal control on and you hit Tjmax, all that happens is your multiplier drops to 6 and your voltage down to minVID. No crash, no shutdown, just a temporary loss in speed(and a huge reduction in temperature). Works quite like speedstep.

I can personally confirm that the processor does indeed shut down at about Tjmax.

 

[Bman123]

Right now I have it @ 3.46 ghz with 1.36250 volts

Ram is at 861 mhz and running great.temp is awesome 55 under load 33idle
This is bus speed of 430 If I go to 440 I get the ole BSOD

What do I need to do to get this thing stable at 3.6ghz (LOL)

 

[Zoson]

Not exactly. First of all, thermal throttling will happen regardless of whether or not thermal control is enabled. Thermal control is a BIOS feature and has nothing to do with the thermal protection mechanisms built into the CPU, which cannot be disabled at all. Throttling also begins a little bit before the temperature hits Tjmax (usually around 97-98C). And it's true that it will only drop down the CPU's speed, but if the temperature continues to increase, there is a thermal shutoff point at a temperature above Tjmax (that temperature is not specified in Intel's datasheets).

It doesn't make sense that the cpu temperature would continue to increase with the multiplier drop and VID drop to minVID. Unless of course your comp was overheating at idle already..

 

[Zero82z]

It doesn't make sense that the cpu temperature would continue to increase with the multiplier drop and VID drop to minVID. Unless of course your comp was overheating at idle already..

The thermal protection is designed to shut down the CPU if the temperature cannot be controlled. That means if the CPU is not being properly cooled, or is in an excessively hot environment, and not even thermal throttling is enough to lower the temperature to a safe level, the CPU will shut itself off to prevent damage from occurring.

 

[Sysjack]

Right now I have it @ 3.46 ghz with 1.36250 volts

Ram is at 861 mhz and running great.temp is awesome 55 under load 33idle
This is bus speed of 430 If I go to 440 I get the ole BSOD

What do I need to do to get this thing stable at 3.6ghz (LOL)

You're going to need more NB voltage, and doing that may require you to put an aftermarket cooler on it. I had to replace the ineffective NB/SB heat sink on my Rampage Formula with some Thermalright coolers. I would bet you don't need anywhere near as much voltage as you're giving the CPU, and that you're undervolting the bus at the speed you're trying to run. My Q9450 runs at 1.176V measured, 1.19V set in BIOS, at 3.6GHz. My settings are here: http://www.hardforum.com/showpost.php?p=1032724375&postcount=18

I have since raised my RAM to 1081MHz which required some tweaking of those settings.