Running Wolfdales at 1.5v safe or not?

[dino26]

honestly i agree with the "greedy" comment.

its fun to oc and see what you can get but I think that with pretty much any wolfdale out most people should be happy with 3.6-4.0ghz on air maintaining a 1.4vcore or less. anything higher is icing on the cake but why push things that far if you arent stable or damaging your expensive equipment.

 

[OfficeLinebacker]

Incidentally, what is the top recommended voltage for a Q6600?

 

[Idontcare]

Originally posted by: OfficeLinebacker
Incidentally, what is the top recommended voltage for a Q6600?

Intel specs allow for 1.5V on G0 Q6600 provided the temps do not exceed 71C.

http://processorfinder.intel.c...tails.aspx?sSpec=SLACR

Intel specs allow for 1.5V on B3 Q6600 provided the temps do not exceed 62.2C.

http://processorfinder.intel.c...tails.aspx?sSpec=SL9UM

 

[OCGuy]

Originally posted by: OfficeLinebacker

Originally posted by: Ocguy31

Originally posted by: Owls
65nm chips are definitely more resilient to overclocking than 45nm.

He is taking a 3.0ghz stock clock to 4.5ghz (50% OC). It isnt the chips fault that it needs unsafe voltage to do 50%.

I would say that 45nm Wolfdales are far better overclockers, but people get greedy. This chip being able to go to 4.0ghz at safe voltages and temps is amazing for the average user. However, people go over to XS and see everyone with 4.5, 4.8, and think they should also.

Personally, if you plan on keeping this chip 18 months or longer, I would not be going over 1.45V. There are plenty of degredation stories out there.

Wait....the Q6600 does 3.6 pretty easily, which is a 50% overclock. I don't believe you need unsafe voltages to do that, but I could be wrong.

Hmm...I dont think 3.6ghz would be average for the Q6600, I see alot of 3.2, 3.0, etc. Not exactly a scientific sampling I know, but when I see people with 3.6 it is usually with water cooling.

 

[Idontcare]

I agree, I raise my eyebrow whenever I see an air-cooled G0 clocked >3.4Ghz.

There are a few out there, but pretty scarce.

 

[OfficeLinebacker]

Thanks for the info.

I don't use water cooling.

I have a Q6600 running right now at 3.6GHz that runs at full load (2 instances of Folding SMP; load average is ~4.7). It runs at about 61C case temp (in the intel sense, middle of heatspreader), with the highest core temp hovering around 72C.

It's one of the $199 jobbies from MC, on a Gigabyte P31 board.

Note it's an open case.

 

[OfficeLinebacker]

For a e8400, the max livable voltage is 1.4, correct? If the specs say max of 1.3625, that's .0375 over.
By extension, the max livable voltage for a Q6600 is 1.50375.

I will double check the VID on my machine, but I doubt it's over 1.5V.

 

[Owls]

Intel's datasheet says that the 45nm chips should not exceed 1.45v

 

[Big Lar]

Actually, the 1.45v is a stretch at best. If you would read the data sheet fully, of which I have, it states that 1.3675v is the max voltage and it is a probability of damage that is irreversible over 1.45v. So, with that, and after reading that a coupla weeks ago, I will not run my cpu over 1.3675v. Guess I am getting old

Larry

 

[Idontcare]

For all your specs -
http://processorfinder.intel.c...m=2774&sSpec=&OrdCode=

For Q9450 45nm -
http://processorfinder.intel.c...tails.aspx?sSpec=SLAWR

Max Vcore is spec'ed for 45nm at this time is 1.3625V provided you keep the Tcase below 71.4C.

And even at 1.3625V this strikes me as rather abusive considering 65nm upper spec is 1.5V.

Given the ~0.75x shrink from 65nm to 45nm, reducing the 65nm 1.5V so as to maintain the same effective electric field on the metal lines (to have similiar electromigration propensity) means you really shouldn't have the 45nm device operating higher than 1.125V.

I haven't seen a SEM cross-section of a penryn yet, but I am going to go out on a limb and suspect that M1 is embedded in TEOS dielectric rather than a low-k in order to support such higher operating voltages (1.3625V versus 1.125V) without destroying itself from electromigration and TDDB issues.

I'm not worried about the transistors seeing the higher electric fields at 45nm, those hi-k dielectrics and metal gate buy you back some robustness.

 

[aigomorla]

1.57Vcore, bios was set at 1.6

Lasted a little over 1 month under constant full load 24/7

Ironic how the day i decide to lower the voltage on it, was the day it died. :X


But definitely not safe, im guessing 1.45V should be absolute max if you dont want that thing to die a painful horrible death.

And as amber said my cooling was kept in check. 44C loaded temps across both cores. Never saw 50 ever.

Oh wellz, i picked a QX because of that. :T


I believe phase units will probably extend it another few months. Im getting one customed by philly. But im not grabing anymore wolfdales.


My Next big project after the one im doing right now which is a rebuild, will probably be a full blown server, bsel moded and watercooled. I think that might be a fun project.

 

[Mondoman]

As Larry writes, it's always a good idea to check out the datasheet for the CPU; as noted, in the case of the e8xxx series, it lists the "absolute max voltage" as 1.45V (vs 1.55V for e2/4/6xxx).

 

[inverse]

Originally posted by: Owls
Like I said before, I'm not going to RMA it. I also doub't that using an aftermarket HS affects warranty.

edit: okay, I don't think what I'm experiencing has anything to do with my chip. I have my Q6600 in and installation is slow as crap. I guess something is wrong with my board

I reset the CMOS and all is well. Both chips are doing fine and are currently napping.

You have Vista don't you?

 

[error8]

Originally posted by: aigomorla
1.57Vcore, bios was set at 1.6

Lasted a little over 1 month under constant full load 24/7

Ironic how the day i decide to lower the voltage on it, was the day it died. :X


But definitely not safe, im guessing 1.45V should be absolute max if you dont want that thing to die a painful horrible death.

And as amber said my cooling was kept in check. 44C loaded temps across both cores. Never saw 50 ever.

Oh wellz, i picked a QX because of that. :T


I believe phase units will probably extend it another few months. Im getting one customed by philly. But im not grabing anymore wolfdales.


My Next big project after the one im doing right now which is a rebuild, will probably be a full blown server, bsel moded and watercooled. I think that might be a fun project.

Oh, so a CPU can die with high voltage, even if it's cooled properly.

 

[Owls]

Originally posted by: inverse

Originally posted by: Owls
Like I said before, I'm not going to RMA it. I also doub't that using an aftermarket HS affects warranty.

edit: okay, I don't think what I'm experiencing has anything to do with my chip. I have my Q6600 in and installation is slow as crap. I guess something is wrong with my board

I reset the CMOS and all is well. Both chips are doing fine and are currently napping.

You have Vista don't you?

No, I am running XP. The behavior was strange because everything loaded very slowly. I should have just reset the cmos instead of doing another reformat.

 

[amenx]

An 8400 at 1.784v:

http://www.xtremesystems.org/f...2843582&postcount=2171

I have a feeling he's already written it off.

 

[Owls]

Is it worth getting a vapochill so you could clock higher at the same voltage? I'm assuming you can use the same or less voltage if you are subzero?

 

[bryanW1995]

the cooling solution on your cpu doesn't affect the required voltage for a given clock speed, it just makes it turn cooler. for example, you might not want to run a Q6600 b3 at 3.6 with 1.5 vcore b/c it could load at 90c+ on air, but with a vapochill you might load at 60c and be ok. the discussions in this thread are centered around cpu death-by-voltage overload, not heat death.

 

[Idontcare]

Originally posted by: bryanW1995
the cooling solution on your cpu doesn't affect the required voltage for a given clock speed, it just makes it turn cooler. for example, you might not want to run a Q6600 b3 at 3.6 with 1.5 vcore b/c it could load at 90c+ on air, but with a vapochill you might load at 60c and be ok. the discussions in this thread are centered around cpu death-by-voltage overload, not heat death.

I disagree. Having both a vaporphase cooler and a B3 QX6700...I can hit clockspeeds with voltages that are unacceptable at room temp.

The voltages in the linked graph are the minimum Vcore for small FFT stability on a B3 QX6700 with vaporphase cooling.

http://i272.photobucket.com/al...ableVcoreversusGHz.jpg

For example I can run my B3 at stock clockspeed on a mere 1.05V. On air (tuniq tower) this requires 1.25V. I'm talking small FFT 12hrs stable in both cases.

Temperature very much impacts the ability to clock a semiconductor at a given speed with a given voltage.

 

[inverse]

Originally posted by: Owls

Originally posted by: inverse

Originally posted by: Owls
Like I said before, I'm not going to RMA it. I also doub't that using an aftermarket HS affects warranty.

edit: okay, I don't think what I'm experiencing has anything to do with my chip. I have my Q6600 in and installation is slow as crap. I guess something is wrong with my board

I reset the CMOS and all is well. Both chips are doing fine and are currently napping.

You have Vista don't you?

No, I am running XP. The behavior was strange because everything loaded very slowly. I should have just reset the cmos instead of doing another reformat.

Oh sorry I thoguht you meant copying files was slow. I've had a similar experience where after OCing, everything would load slower than stock. A BIOS reset fixed it also

 

[Owls]

Originally posted by: Idontcare

Originally posted by: bryanW1995
the cooling solution on your cpu doesn't affect the required voltage for a given clock speed, it just makes it turn cooler. for example, you might not want to run a Q6600 b3 at 3.6 with 1.5 vcore b/c it could load at 90c+ on air, but with a vapochill you might load at 60c and be ok. the discussions in this thread are centered around cpu death-by-voltage overload, not heat death.

I disagree. Having both a vaporphase cooler and a B3 QX6700...I can hit clockspeeds with voltages that are unacceptable at room temp.

The voltages in the linked graph are the minimum Vcore for small FFT stability on a B3 QX6700 with vaporphase cooling.

http://i272.photobucket.com/al...ableVcoreversusGHz.jpg

For example I can run my B3 at stock clockspeed on a mere 1.05V. On air (tuniq tower) this requires 1.25V. I'm talking small FFT 12hrs stable in both cases.

Temperature very much impacts the ability to clock a semiconductor at a given speed with a given voltage.

That's great info. Unfortunately, vapochill is really expensive. I think $860 currently and I thought of hooking up a mini AC to it for $250-300 but it's bulky and not as effective. Do you recommend any alternatives (besides water) or is vapochill the only game in town?

 

[bryanW1995]

Originally posted by: Idontcare

Originally posted by: bryanW1995
the cooling solution on your cpu doesn't affect the required voltage for a given clock speed, it just makes it turn cooler. for example, you might not want to run a Q6600 b3 at 3.6 with 1.5 vcore b/c it could load at 90c+ on air, but with a vapochill you might load at 60c and be ok. the discussions in this thread are centered around cpu death-by-voltage overload, not heat death.

I disagree. Having both a vaporphase cooler and a B3 QX6700...I can hit clockspeeds with voltages that are unacceptable at room temp.

The voltages in the linked graph are the minimum Vcore for small FFT stability on a B3 QX6700 with vaporphase cooling.

http://i272.photobucket.com/al...ableVcoreversusGHz.jpg

For example I can run my B3 at stock clockspeed on a mere 1.05V. On air (tuniq tower) this requires 1.25V. I'm talking small FFT 12hrs stable in both cases.

Temperature very much impacts the ability to clock a semiconductor at a given speed with a given voltage.

that's good info. $860 will buy you a LOT of tuniq's, however

 

[aigomorla]

if you can squeeze those sub degree temps, your cpu will load higher with less voltage.

how much higher? only God can answer this question.

IF your not an advanced overclocker i wouldnt bother looking at phase. You need to insulate the board, and also, things can get really messy sometimes.

 

[Idontcare]

Originally posted by: Owls
That's great info. Unfortunately, vapochill is really expensive. I think $860 currently and I thought of hooking up a mini AC to it for $250-300 but it's bulky and not as effective. Do you recommend any alternatives (besides water) or is vapochill the only game in town?

Vapochill is fun and crazy spendy. I'd recommend water before vaporchill unless you are really interested in pushing performance to the extreme.