Too high voltage for 24/7?

[AriciU]

Asus p5k-e wifi, e6750 @ 3824mhz. The vcore is set to 1.66v in bios with the vdrop limitator ON. This gives me 1.64v in idle and 1.63v in load, CPU-Z shows it like this i mean.

Everest motherboard temp (Northbridge temp i assume) stays at 23C idle / 31C load with 1.55v set in bios.

CPU temps are 21/26C Idle and 61/61C load (Orthos after like 2-3 hours).

Do you think i should keep it like this or is the voltage a wee bit too high? Temps seem OK to me.

BTW, this is the minimum voltage i can get it orthos stable at this speed. If i drop it i also have to drop the cpu speed (don't really want that ).

[KTE]

Too high voltage for 24/7?

I do, definitely.

Not sure if others would agree.

[skynet666]

Depends if you want your cpu to last two weeks
In all honesty, from experience with presHOTT, I wouldnt go past 1.55, allthough core 2 may have a higher tolerance? maybe someone else could comment? Your temps seems really good

[zsamz_]

i get nervous after 1.6 lol i was doin 4.2 @1.62v n my load was 60c still i wouldnt keep it there for 24/7
maybe you should try high fsb @3.6

[AriciU]

i get nervous after 1.6 lol i was doin 4.2 @1.62v n my load was 60c still i wouldnt keep it there for 24/7
maybe you should try high fsb @3.6

I can't. My ram sucks and i'd really have to get some serious voltage in there for anything about 500FSB. These sticks would burn like there's no tommorow IMO They are hot as it is at 2.2v.

My options for the CPU are: 3.72ghz @ 1.57v / 3.6ghz @ 1.52v / stock @ 1.31v

Offtopic: how do people mount an 80/92mm fan above their RAM slots? I can't think of anything inside my case that i can screw the fan on to stay put.

I think i'll just keep the CPU as it is. I'm gonna get a Wolfdale E8500 when they come out in January anyway. This should be fine untill then i guess/hope

[zsamz_]

aim for 3.65

[flopper]

We as a community of extremes dont know when and how core2duo breaks with volts?


One of the reasons is that temps is one side of the equation, the other is electron migration.
So, I run a conroe 6600 a year using 1.56vcore, no problem.
I also can run this new E6750 at 3.8ghz 1.59/1.56 vdrop and tried 4ghz and can run at 1.67 bios set for bench, its pretty doable for me to run it at 1.67vcore, I just dont know if its worth it, I rather run at 1.56 24/7 and then bench a little higher since the other factor that comes into play is overall temperatures.

I just sit back and wait for 45nm cpu´s, the wolfdale can go pretty high at low voltage and temps.
as soon its out I sell my 65nm.

[Pcuser]

I think that's pretty high .
My E6750 needs 1.40 vcore for 3.8 stable

[AriciU]

Yeah. My CPU is a pig though. I think i caught the worst e6750 on the market. At 1.40v i can only get around 3300

[timber715]

i think thats ok as Monitoring your TEMPS is more important, haven't seen any proc break/die due to voltage but some due to excessive temps.

[Falkentyne]

Well, my X6800 lost max overclock at 1.575v. That voltage caused it to lose a few FSB (or requiring slightly higher vcore for 100 mhz lower at what used to be stable at 1.475v). And I've had it briefly at 1.625v (never for long). So that should tell you something.

Basically, every chip is different, but I have experience with these lower micron processes; Ever since the die shrink that led to the P4/netburst, you are -always- taking some risk when you go above the maximum operational voltage. Intel does NOT draw up those charts for their good looks. They know far more about what their chips are capable of than anyone on these forums. It's sort of like a human being trying to tell God how to design a galaxy...

I had a P4 that very slowly degraded (over many months) at just .25v above the max operational, in this case, 1.575v ! The lower process is -far- more voltage sensitive than the old Pentium 3's were. The P3's never suffered from this type of problem on this scale, and those chips could run *HOT*. Heck I still remember an old Pentium MMX with the stock heatsink, which was overclocked, and the fan FAILED on the heatsink, and the chip still didn't crash O_O... Those things could take some serious heat and voltage. (I remember going to to 3.3v on one, and may have gone as high as 3.5v).

With the hotter/larger process, it took a lot more abuse to see those chips degrade, whereas, now....but something definite DID change between the P3 and the P4, and some of those P4s responded to voltage like tissue paper (put the voltage to a similar % increase that you did on a P3, and you could have an overclock degrade so bad, you can notice it within week....)

Basically, if you want what *MAY* be a safe 24/7 air voltage, try to look for 0.1v above the Intel maximum operational voltage (NOT on northwoods; remember I told you I had one slightly degrade at just .25v above), so, for a 1.35v 65nm chip, go no higher than 1.45/1.462v (after vdroop), for a 1.25v 45nm chip, go no higher than 1.35v/1.362v. This may not seem like enough voltage, but this is for -safe- or reasonably safe 24/7. You're perfectly able to go higher if you want, but you do so at your own risk. Someone recently saw degradation on a c2d above 1.55v, so it *can* and probably *will* happen. But each chip responds differently, and usually on how well it overclocks. This may be XS, but not everyone here can afford degrading a steady stream of hardware...

[KTE]

I used a Q6600 on air open benchtop which I benchmarked at 1.625V BIOS/1.594V Windows load at work for 3 days/nights continuous in ambient 10-12C to try and finish a project 2x quicker than it would otherwise take. Mainly some very heavy virtualization and medical load/arithematic calculations but I ran some common enthusiast benchmarks to check at the beginning too. It ran 24/7 doing medical work similar to folding when I left. At the start the max clocks it was reaching and staying 65C load was ~3650MHz. It was inside a hard plastic deep tub with a transparent acrylic sheet over the top which had a circle cut out above where the mainboard was and inside the tub were 4x Delta fans 2 were 252CFM and 2 were 150CFM with 2 more 40mm and 60mm fans for the NB and MOSFETs.
I primed it hard for 8hrs to check at the start and it showed no errors whatsoever. It all did well until the 5th day when I went back to check in the evening and it was still running. However the calculations were giving very poor scores of what you'd expect at 3.1/2GHz.

So I rebooted it and ran P95 again just out of curioisty. Running these volts continuous with not enough cooling makes it very common to damage these low node chips, so it was always in the back of my mind.
Anyway, it gave errors below 2 minutes. So I backed down to find my stable again to continue on. It errored at 3.5GHz under 2mins. I moved it and I kept going more and more lower until I reached 3.1GHz. At that time I just set it to stock 2.4GHz and left all my RAM very high timings after checking it with Memtest86 to see if P95 gives errors now. What do you know, it errored within 5 minutes again. I had damaged a CPU that belonged to the work firm. I had to pay the charges for a new replacement.

[mad_skills]

Intel recommends that vcore should stay below 1.5V, everything above 1.5V shortens the life expectancy of the cpu.
This of course is not important 'cause you swap cpu's every year or more right.
I just think that you don't need 3.8GHz for 24/7, you will get 99% of the performance with 3.6GHz but that will require significantly less vcore, mine does 3.6GHz @1.37500V.
I also need something like 1.5V or a bit more for 4GHz but i just don't need that, i use that vcore and speed only for benchmarking.
1.6V is a lot, even if you have liquid cooling and you don't.

[Bigchrome]

Ran my E6600 at 1.7v (real) for an entire year with absolutely zero problems.... If you need that to get your max and it degrades you just end up at the frequency you were going to be at anyway...