Safe Q6600 vcore?

[aznofazns]

I used to just leave my Q6600 at 3.33ghz but now I'm tampering with it a bit... I raised the vcore to 1.475v for 3.5ghz and prime95 failed pretty quickly. Now i'm trying 1.4875v and it seems stable so far. It seems like I've got pretty significant vdroop though, since my CPU-Z reads 1.416v and my HW Monitor shows 1.42v max under load. Is this normal? Also, if it turns out that 1.4875v and 3.5ghz are a stable combination, what do you guys think the risk of my CPU burning out will be? For instance, do you think it'll last 3+ years at such settings? My core temps reach a max of around 69C under prime95. Under normal use, though, they never surpass 55-56C.

EDIT: Hmmm crap, my core temps got up to 72C and my vcore readout increased to 1.45v. I'm doing a Blend test btw, not Small FFTs.

 

[lodingi]

I am currently trying to see how high I can go on my rig. I have a Q6600 1.325 VID. My vcore is set at 1.525v in the bios. CPU-Z is reporting my core voltage at 1.44 idle & 1.392 under load. My FSB is 378x9 for 3.4 ghz & I'm running Prime95 64 bit as I type this post. Cores 1&2 are reading 65C. Cores 3&4 are reading 56C. I would like to get to 3.6 ghz stable if possible. I think the general rule is to keep the vcore under 1.5v after vdroop & your core temps should be kept under 71C under load.

 

[Furious Nerd]

Generally, on top end air cooling (your Xigmatek qualifies) 1.5V should be around the max, but it all depends on your temps.

The max safe temp according to intel is 71C, the max safe temp to people who don't plan to keep their CPUs for 20 years, is ~80C

Keep in mind, it's OK for a stress test to raise your CPU heat so high, it's not like anything in the real world with keep it at 100 percent load for 8 hours a time like a stress test does, so your temps will most likely never reach even 65C in the real world.

As long as your stress temps are below 80C you'll be fine.

Another thing, make sure you use a very high divisor so that your RAM won't interfere with the CPU OC, there's no reason you should be testing both at the same time.

That means use really low speeds for RAM, and OC clocks for your CPU, while testing small TFF's in Prime to ensure you're testing your CPU, not both your CPU and RAM at the same time

Keep in mind the above temps refer to G0 revision. B3 has a ~63C max safe temp and is a terrible overclocker.

 

[unclewebb]

The max safe temp according to intel is 71C, the max safe temp to people who don't plan to keep their CPUs for 20 years, is ~80C

That spec can be somewhat misleading. 71C is a TCase temperature measured at the geometric center of the CPU on top of the IHS with an embedded probe. With a heatsink installed, depending on the load, the actual core temperature is a good 25C to 27C hotter than the TCase tempeature. rge over on the XS forums recently did a thorough test showing that.

Thermal throttling begins about 2C to 3C before TJMax. The theory is that if you keep your IHS based TCase temp at a maximum of 71C then your processor will be very unlikely to ever throttle. 71C does not refer to a maximum core temperature.

There are a lot of crazy overclockers and I rarely read of people killing their CPUs. When they do, it's mostly excess voltage that does it. As long as you're Prime stable then there's no reason to worry about what temperature you're running at. If you are well overclocked your Quad will reach a point where it simply isn't stable. More voltage won't help because it will only create more heat which will further reduce your overclock. Let stability be your guide and you can basically overclock these things as much as you like.

 

[aznofazns]

Thank you all for the replies.

I just spent an hour or two reseating my heatsink and reapplying AS5 to my CPU. This time I used a plastic bag to spread the paste evenly across the surface of the CPU rather than place a glob in the middle. Hopefully that will reduce temps. I'm testing Prime95 (small FFTs) with a 3.33ghz OC at 1.45v so hopefully it'll be 100% stable with manageable temps. My real goal is to reach 3.6ghz but it seems that will be impossible without at least 1.5v which is kind of ridiculous.
UPDATE: after an hour of testing, max core temps are 66,66,60,60... not bad.

Also, I skimmed an article on TCase and whatnot... I'm guessing that Intel's specification of 71C limit means that the actual limit for core temp is quite a bit higher?

 

[iceredwing]

1.5v is the labeled max for the 65nm processors. it is fairly normal to have large vdroops during load. mine can get as high as .1v off the amount that is set on in the bios

 

[setter]

Im happy to let mine go to 75'c temp wise ,and volts 1.5500 would be my max, the old 65nm quads are pretty robust, if i can get mine to 3.8ghz on air ill be happy, sell it and buy a 45nm quad (cant afford i7)

 

[prasvt]

So would 1.55v on water be ok? I have an ip35 Pro, which has a hell of a vdrop. So if I set the vcore to 1.565 in the BIOS, it's actually 1.50 volts. So my Q6700 has been running at 3.6ghz at that voltage. Under load, the vdroop is about .03-.04 volts so (avg of about 1.47V). My temps under load are about 49-50C. Thus far I've been careful to keep the voltage at 1.5v max, regardless of temps.

But I've booted at 3.8ghz before (took about 1.55v or 1.6 in bios)...just wasn't sure if I'd kill the cpu. Sounds like it might be ok if the temps are manageable?

 

[setter]

So would 1.55v on water be ok? I have an ip35 Pro, which has a hell of a vdrop. So if I set the vcore to 1.565 in the BIOS, it's actually 1.50 volts. So my Q6700 has been running at 3.6ghz at that voltage. Under load, the vdroop is about .03-.04 volts so (avg of about 1.47V). My temps under load are about 49-50C. Thus far I've been careful to keep the voltage at 1.5v max, regardless of temps.

But I've booted at 3.8ghz before (took about 1.55v or 1.6 in bios)...just wasn't sure if I'd kill the cpu. Sounds like it might be ok if the temps are manageable?

Im not sure about the q6700, but ive seen a few folks sticking 1.6v through a q6600 for prime95 runs (8 hours+) most of theese rigs were water cooled, some on extreme air cooling,(high cfm delta fans) with very low vid chips, my own cpu has a vid of 1.2750. Im gonna try and take it to 3.8ghz this weekend, currently @3.6ghz on 1.4125 vcore with load line calibration enabled (asus board feature) my only worry is temps and trying to stay below 75'c, i think my rig should be ok. 2x2000 rpm fans on my cooler, couple of yate loon 1300's for intake, 1 for exhaust. Soon to be replaced with another 3 silent eagle 2000's.

Ps good read here on quad overclocking.

http://forums.overclockers.co.uk/showthread.php?t=17779323

 

[aznofazns]

Wow 1.4125v for 3.6ghz is pretty impressive. My chip's got a VID of 1.27 or 1.28 as well, but I need close to 1.45v just for 3.33ghz. No matter how many times I reseat the heatsink and reapply AS5, though, my temps are always trying to reach the 70s. I think I might replace my stock Xigmatek fan with a Yate Loon and see if that helps.

 

[setter]

Wow 1.4125v for 3.6ghz is pretty impressive. My chip's got a VID of 1.27 or 1.28 as well, but I need close to 1.45v just for 3.33ghz. No matter how many times I reseat the heatsink and reapply AS5, though, my temps are always trying to reach the 70s. I think I might replace my stock Xigmatek fan with a Yate Loon and see if that helps.

A few words of advice mate, one thing to watch with the p5q boards is that they seriously overvolt the chipset voltages (particularly on earlier bios's) auto and first manual setting are the same (over volted) i currently run mine at +0.02v on the cpu pll, fsb termination and north bridge voltage and my ram which is rated @2.1v i have running at 2.02v again due to the board overvolting. My cooling is a TRUE black with 2x2000 rpm fans in a push/pull config which really benefits a high clocked q6600

 

[prasvt]

Yeah I'm jealous of you guys who hit 3.6 with such low voltages. I'm sure if changed up mobos it might help, but it's too much of a pain in the ass to disassemble my water cooling setup + the expense for 200mhz. Mine's a 1.275 VID chip as well, but below 3.6ghz it doesn't require that much voltage. My guess is, I've probably half-assed my overclocking methods. I've got 4 fans on my radiator ... one fat panaflo, one tuniq fan and 2 thermaltake fans (bottom is push pull, middle is the fat fan - pull, and top is the fan from my tuniq tower - push). Random assortment I know. But they run at 8v until temps hit 35C and then they're dynamic via the BIOS.

I do have a quick question though...what voltages are you guys running your MCH at? I tend to push more voltage if I use a lower multi and higher fsb (360 x 10 currently, but say 400x9 or 450x8).

btw thanks for the link, I'll check it out now...maybe do a more methodical oc.

 

[setter]

Yeah I'm jealous of you guys who hit 3.6 with such low voltages. I'm sure if changed up mobos it might help, but it's too much of a pain in the ass to disassemble my water cooling setup + the expense for 200mhz. Mine's a 1.275 VID chip as well, but below 3.6ghz it doesn't require that much voltage. My guess is, I've probably half-assed my overclocking methods. I've got 4 fans on my radiator ... one fat panaflo, one tuniq fan and 2 thermaltake fans (bottom is push pull, middle is the fat fan - pull, and top is the fan from my tuniq tower - push). Random assortment I know. But they run at 8v until temps hit 35C and then they're dynamic via the BIOS.

I do have a quick question though...what voltages are you guys running your MCH at? I tend to push more voltage if I use a lower multi and higher fsb (360 x 10 currently, but say 400x9 or 450x8).

btw thanks for the link, I'll check it out now...maybe do a more methodical oc.

Theese are the chipset voltages i use.

PLL - 1.52v, FSB Term - 1.12v, and NB - 1.12v)

Vcore of 1.4125 set in bios, load line calibration enabled, iirc mch is north bridge (im to used to nvidia boards)

 

[iceredwing]

should read up on fan choices and test results here. I went with what Vapor recommended and that helped dropped temps on my setup

 

[yenniedn]

I've got a lapped Q6600 with a VID of 1.2625v that's capable of doing 3.6GHz prime95 stable at 1.4v. Load temps are in the mid 60's*C on my lapped TRUE 120. The highest I've ever taken it is 1.45v but didn't get any noticeable OC increase so I run it at 1.4v. I wouldn't ever take it over 1.5v I don't think.

 

[iceredwing]

it's quite hard to compare temperatures because lapping/type of heatsink/TIM/etc. all play 2nd fiddle to your case temperatures & ambient temperatures going into the case.

If your case is hot or the ambient is hot (like in california... it's like summer here all year round) then you have quite a battle at hand.

 

[aznofazns]

I'm testing Prime95 @ 3.5ghz @ 1.49 something volts with +.02 on NB, FSB termination, and PLL but small FFTs failed after 10 minutes again... disappointing. Blend has been stable for nearly 20 minutes now though.

 

[prasvt]

You're right...California makes it hard to maintain cool temps - even w/ water cooling. It's mid november and it's like 70 outside (east bay)...bizarre. Honestly I need a full tower case...I've got a midtower with all the watercooling stuff in it + a gtx 260 + x-fi + wifi card ...very tight fit. Tack on 2 40mm fans on the northbridge, a 120mm fan blowing at the pwm mosfets, and front and window intake fans and it seems to be just enough to make it through back to back 720p blu-ray -> h264 encoding sessions (100% cpu for about 4 - 6 hrs straight). It's probably not comparable to prime, but it's a realistic day to day load.

Back to oc'ing..so aznofazns ..you may need a bit more on the NB. I've noticed I need 1.44v on the MCH whenever i run 360 x 10 ...maybe a bit more at 400 x 9. So I'd say depending what fsb you're using, maybe add a few notches to it and run the small FFTs again and see if it helps.

Then dial it back a bit. 1.49 vcore should be more than enough for 3.5ghz.

 

[mista ting]

you must be doing something wrong most Q6600's don't need that much vcore for that low OC.........mess around with other settings

 

[aznofazns]

This is interesting. I raised my PLL to 1.58v, my FSB term. to 1.36v, and my NB to 1.30v, keeping vcore at 1.49375. After 25 minutes of small FFTs, my max temperature is only 65C!!! I'm guessing that raising those other voltages had something to do with this...

 

[mista ting]

This is interesting. I raised my PLL to 1.58v, my FSB term. to 1.36v, and my NB to 1.30v, keeping vcore at 1.49375. After 25 minutes of small FFTs, my max temperature is only 65C!!! I'm guessing that raising those other voltages had something to do with this...

hey mines do the same thing!! i'm guessing it kind of messes with the temp readings?

 

[aznofazns]

I wouldn't say it "messes" with the readings, I think it actually helps reduce temps when you allow for a bit more voltage. This is purely speculation, though.

3.6ghz testing failed horribly at 1.5 vcore, 1.6 pll, 1.4 fsb term, 1.5 nb. Testing 3.52ghz now...

EDIT: 3.528ghz, 1.5 vcore, 1.58 pll, 1.38 fsb term, 1.34 nb, stable. Do you guys think it's safe to leave it at these settings on a daily basis or the voltage (especially vcore) a little excessive?

 

[prasvt]

Well see that's where each system is unique...I think those are my settings for 3.6ghz. Anyway, I'm suspending this high speed pursuit b/c my side of the house gets all the sun and it's pretty hot here today. 3.7ghz is requiring 1.55V easy = 61C load temps on prime95 right off the bat.

 

[Furious Nerd]

I wouldn't say it "messes" with the readings, I think it actually helps reduce temps when you allow for a bit more voltage. This is purely speculation, though.

3.6ghz testing failed horribly at 1.5 vcore, 1.6 pll, 1.4 fsb term, 1.5 nb. Testing 3.52ghz now...

EDIT: 3.528ghz, 1.5 vcore, 1.58 pll, 1.38 fsb term, 1.34 nb, stable. Do you guys think it's safe to leave it at these settings on a daily basis or the voltage (especially vcore) a little excessive?

don't change your PLL over anything above the default 1.5, it does almost nothing for overclocking and will certaintly vastly shorten the lifespan of your CPU. and I mean VASTLY

 

[aznofazns]

don't change your PLL over anything above the default 1.5, it does almost nothing for overclocking and will certaintly vastly shorten the lifespan of your CPU. and I mean VASTLY

What causes that? I ran small FFTs again with pll set to 1.5 manually instead of 1.58 but the test failed after 10 minutes.

Also, as a side question, why does vcore lower when the cpu is stressed and go back up when idle?

 

[Eruditass]

don't change your PLL over anything above the default 1.5, it does almost nothing for overclocking and will certaintly vastly shorten the lifespan of your CPU. and I mean VASTLY

is this true? i've heard this about fsb termination / VTT but nothing about PLL. I've found that my setup needs 1.6v on PLL to be stable at 3.6

 

[BillParrish]

why does vcore lower when the cpu is stressed and go back up when idle?

(some of which is by design, really, it is enginered that way to keep overshoot spikes from fast load changes from damaging the cpu).

Under load the current through the components of the cpu voltage requlation components increases greatly resulting in more I squared x R (voltage drop) losses. The cpu voltage requlation circuity compensates for this a lot but not perfectly and there are other concerns about voltage spikes created when the load is suddenly applied or removed , starting up or shuting down the load program, and the voltage requlation circuitry suddenly has to deal with large changes current flow.

The intel data sheet has a section on vdroop and overshoot and the Intel site also has the specifications for the VRD (voltage regulation down) Intel recommends. Both contain sections that are not too technical and can be understood with a careful read. Intel does not design CPUs to be OCed and thus takes extreme care in engineering the system to be very robust. Enthusiast motherboard makers must follow the Intel guildlines to some extent because the reasons Intel does what it does is vaild if you OC or not. Or the short version - nothing is perfect and you cant have your cake and eat it too.

 

[prasvt]

Just to resurrect this thread ... I'm using an Asus Maximus II Formula board now - kind of a pricey replacement for a busted ip35 Pro, but I don't plan on moving to an i7 any time soon. So w/ the new board, a Cosmos S with the rad mounted internally, 6 rad fans, and a new Zalman 850w psu ...going to try for 3.7 or 3.8 today. Love the load line calibration feature..no drop in voltage under load!! Anyway I'll report back...this board has a lot of crazy tweaking options...but right now I'm running 400x9 @ 1.475V (which is lower than the 1.51 on the p35 board..but still an arbitrary choice on my part).

 

[prasvt]

Okay..wussed out for now I've been running prime for 3 hours at 3663 (407x9) @ 1.50V and it's fine thus far...58C under load (rad fans @ 70%). I can probably adjust it to 3.7ghz tomorrow and I think that's good enough for a few fps more for my 720p blu-ray -> h.264 encodes. Just unwilling to push much past 1.5Vcore for 24/7 use.