e6750 needs crazy vcore for some reason

[AriciU]

Hi

I just got a Gigabyte P35-DS3R, an E6750 and a Thermalright Ultra120+Scythe sflex fan. I can reach around 3200mhz on the stock voltage, without having to mess with timings or anything else in the Bios (latest F7 revision).

I take it to 3.3ghz but i need to increase the voltage to 1.375v.

Anything beyond that and i need to add some serious voltages all around to get the damn thing stable. My mobo has a bad vcoore drop also btw, it drops like 0.5-0.7v when i hit it with Orthos.

Anyway, to get ~3550mhz (444 x 8) stable i have to set my latencies to 5,5,5,15, add +0.2v on both the Southbridge and the Northbridge and set the voltage in the Bios to 1.518v (this drops to around 1.44 at full load). The temps are fine.

Now i'm running at 3656Mhz (457 x 8), 5,5,5,15; +0.3v on NB and SB and, get this, 1.59v on the CPU
When i hit it with Orthos it drops to around 1.52v but at least it's stable. Temps never go beyond 55-57 deg. Celsius (with Core Temp, 100deg tjunction) and it idles at around 25/31 (1stCore/2ndCore).

My question is... is this a safe thing to use even if it doesn't pass 60deg C? I've heard people say that i can lower the life of the CPU to 6 months or something along those line.

Also, why the crazy vcore? I know people who are getting 3.6ghz 3.75v (0,25 more then stock). Is it just my luck catching a bad chip who needs loads of voltage to get anywhere?

Anything beyond 3656Mhz and windows doesn't want to load. No matter what vcore i set. I haven't increased the PCI-ex voltage btw but doing it makes no difference in stability and max cpu speed.

I was really hoping to hit 3.8-4.0ghz with this chip. That's why i chose it over the Q6600

http://img216.imageshack.us/img216/2171/intesteya4.jpg

My PSU is a Corsair CMPSU-520HX, 520W unit. GFX is a Gigabyte 8800GTX.

 

[The-Noid]

Yeah could be a bad chip, otherwise could be bad vreg, should be good on that board.

A lot of Core 2's need a bunch of voltage to cross certain frequencies though.

 

[mrfatboy]

what are your you volting your ram to?

 

[MetaDFF]

I think the overclock you are getting is pretty typical for the E6750.
All those "easy" 3.8 GHz+ overclock you see in reviews are likely hand picked chips to generate buzz.

Anyway, to get ~3550mhz (444 x 8) stable i have to set my latencies to 5,5,5,15, add +0.2v on both the Southbridge and the Northbridge and set the voltage in the Bios to 1.518v (this drops to around 1.44 at full load). The temps are fine.

Your overclock and voltage at 3.55 GHz is similar to my current overclock of 3.56 GHz @ 1.432 V with a full load voltage of 1.424 V (ASUS motherboards have a droop compensation setting in the BIOS). The rest of my overclock settings are in my sig .

I would be wary to overvolt the CPU beyond 10% of the specified range. When chips are designed it is typical for them to be tested with 10% over-voltage.

My question is... is this a safe thing to use even if it doesn't pass 60deg C? I've heard people say that i can lower the life of the CPU to 6 months or something along those line.

I previously posted this before but it's worth reposting here since you asked:

When you over-volt a CPU beyond the specs, the transistors in the chip experience high electric fields which cause, "hot electrons" or electrons that have extremely high energy. These high energy electrons can enter the oxide region in the gate where they can become trapped causing oxide charging. Over time, these charges will accumulate and degrade the transistor performance by increasing the threshold voltage V_T. Increasing the V_T of a transistor lowers the I_ON of the transistor which lowers the speed of the transistor. This is probably the reason why their processor could no longer overclock as high as before. Also running the CPU 24/7 aggravates the situation as more switching (electron activity) occurs which will increase the rate at which those trapped charges accumulate.

I think there is a common misconception that overvolting a CPU while keeping the temperature below a certain threshold won't damage the CPU. This is true to a certain extent because lowering the temperature helps meet the metal current density (so metal interconnects don't burn out due to the increased currents). But applying a high voltage to the CPU beyond the spec will certainly cause "hot electrons", which is something the CPU makers consider when setting the maximum voltage spec on their chips.

The bottom line is that overvolting will cause a degradation in the performance of the CPU over time. How long before you begin to notice the degradation is hard to tell.

Basically overvolting will damage the chip and you do so at your own risk. The higher the voltage the quicker the damage will accumulate.

 

[AriciU]

Originally posted by: mrfatboy
what are your you volting your ram to?

2.1v as recommended by Geil. I can set it to 2.2v and run the fsb asyncronously at around 1010Mhz with the multi dropped to 7x but i still can't get any more clockspeed when using 2.2v.

MetaDFF: yeah, i've read about e6750's being hit and miss. I should've gone for an e6850 to get the 4ghz that i want... oh well.

 

[mrfatboy]

just make sure you set the DDR2/DDR3 overvoltage to +.2 and not +.3 to get 2.1v in your bios.

 

[imported_SeeFu]

I'm running a MSI P6N Platinum board w/ the E6750. I can get it to 3.6ghz, but when I get it to that speed, my processor core temps rise to 67C. What's interesting is my vcore under cpu-z is never above 1.325v. Is this normal?

 

[TC91]

SeeFu, what kind of cooling do you have? the temps might be a bit high due to your cooling or the dts on the cpu or motherboard may be uncalibrated as well but that is less likely.

 

[bryanW1995]

I need 1.53 in bios to run stable at 3.5 with mine. It droops to 1.49 under load. I've been north of 3.7 with temps in mid 60's running orthos but vcore was crazy and my case was loud as hell with all the fans set to high and the tuniq at 2500 rpm.

 

[swing848]

Is your motherboard reporting 46 degrees C at idle?

 

[AriciU]

Originally posted by: swing848
Is your motherboard reporting 46 degrees C at idle?

Yeah. It shows it at 45C in Everest... I'm assuming it's either the northbridge or southbridge. Any problems with that? I don't think it's too high IMO.

 

[n7]

3.65 GHz @ 1.52V = bad?

Obviously that's not amazing, but it's pretty damn good in my books.

OCing involves a lotta luck.

If you really want to find an awesome OCer, you'll likely need to buy a couple chips, test them all, & keep the best one.

 

[swing848]

Originally posted by: AriciU

Originally posted by: swing848
Is your motherboard reporting 46 degrees C at idle?

Yeah. It shows it at 45C in Everest... I'm assuming it's either the northbridge or southbridge. Any problems with that? I don't think it's too high IMO.

Not too high, right now, as I type, my motherboard is reporting 29 degrees C, but I have great case cooling. When I saw 46 degrees C for your motherboard I began to wonder what kind of case and cooling you have as well as your room temp.

 

[littleOaf]

I am running my e6750 @ 3.720Ghz w/ 1.464 VCore using the mighty Thermalright-120 (I lowered my VCore after reading MetaDFF's repost of high energy electron migration due to high core voltages). I have previously hit 3.8Ghz at ~1.5VCore but my RAM seemed to be the limiting component in my setup...Orthos runs kept giving the ugly loud BEEP. My temperatures are at ~35C using both SpeedFan and CoreTemp, though this might be 15C lower than the true value(I have been told my BIOS version somehow results in lower reported temperatures). I am thinking of reseating the heatsink and shaving some of the thermal compound(Arctic Silver) as this has been claimed to increase thermal transfer.

 

[jmmtn4aj]

Originally posted by: swing848

Originally posted by: AriciU

Originally posted by: swing848
Is your motherboard reporting 46 degrees C at idle?

Yeah. It shows it at 45C in Everest... I'm assuming it's either the northbridge or southbridge. Any problems with that? I don't think it's too high IMO.

Not too high, right now, as I type, my motherboard is reporting 29 degrees C, but I have great case cooling. When I saw 46 degrees C for your motherboard I began to wonder what kind of case and cooling you have as well as your room temp.

My chipset temp is current reported as 58 degrees I'd attribute that to the cooler being a little too close to the chipset HS.

 

[imported_SeeFu]

Originally posted by: TC91
SeeFu, what kind of cooling do you have? the temps might be a bit high due to your cooling or the dts on the cpu or motherboard may be uncalibrated as well but that is less likely.

COOLER MASTER RR-CCH-L9U1-GP 92mm Hyper TX2 CPU Cooler - Retail

http://www.newegg.com/Product/...p?Item=N82E16835103031

 

[imported_SeeFu]

My m/b died about a week ago and newegg sent me a replacement. This new one runs MUCH cooler. E6750 o/c to 3.4. NB was bumped up 2 steps as was vcore. core temp reads ~55C under load with prime95. With my first m/b I would be in the mid 60C's

 

[BonzaiDuck]

That's pretty darn good, and cost-effective with the CPU price at around $190 today.

My personal "policy" is to avoid over-clocks that require VCORE in excess of between 5 and 10% of the retail-box "maximum." In fact, I think I stop at 5%. I even felt uncomfortable pushing my E6600 to 1.48V.

I think it's fine that we have a (sort of) enthusiast hobby for pushing these processors to their limit, but I also have a pocket-book that I must consider in the balance. Still, for a CPU costing less than $200, I can see pushing it just a little bit . . .

The conventional wisdom -- if there is any and IF I understand it correctly -- is that keeping the chip cool doesn't reduce the increased risk to CPU longevity due to electron migration. In fact, you can fool yourself with cooling.

I also think it's a myth that extreme cooling has an across-the-board positive effect. Conductivity on the silicon-circuits actually increases with temperature, while metal parts have increased electrical resistance. And again -- even if you reduce load temperatures to room-ambient, I'd limit my over-clocking so that it only captures higher speed to the limit of 5% VCORE in excess of Intel's maximum. You may want to push for a bit higher than that, but why throw away dollars on a speed-record when the manufacturer has already tested the effect of voltage to alleviate their warranty returns?

 

[TC91]

yah, limit vcore to ~5%, that is a very safe margin, especially if your temps are not so great. if your temps are great, then 10% over max recommended on box is the absolute maximum i would recommend using on a daily basis.

 

[imported_SeeFu]

CPU-Z still shows my vcore voltage at 1.288, I keep wanting to push the CPU so much harder since I have a lot of vcore to go before i hit intels termal limit.

 

[imported_SeeFu]

I still have .2v to go before i reach that limit.

 

[BonzaiDuck]

The voltage "maximum" limit printed on the retail box, I presume? Actually, that's a lot of wiggle room ya got there if my presumptions are accurate.

 

[imported_SeeFu]

yea. i've barley increased it over stock. 3.4ghz is fine w/ me though. it runs everything i need at that point.

 

[legcramp]

yeah, your chip is maxing out probably at those voltages because it takes such a big increase in vcore to get that extra 100 mhz.

 

[MaDMvD]

From my posts on another thread:

I spent a majority of the day seeing the minimum amount of voltage per 200MHz step. I only did 3.2GHz - 3.6GHz, but at least now I know what works and what doesn't.

3.2GHz - 1.3375v in BIOS
3.4GHz - 1.41875v in BIOS
3.6GHz - 1.5v in BIOS

It's really interesting to compare voltage consumption amongst other E6750's. If you notice above, the difference in voltage needed between 3.2 and 3.4 is roughly .08v and the difference in voltage needed between 3.4 and 3.6 is about the same. It seems that at lower clocks speeds, I am at par with those of you guys that have non-voltage hungry CPUs. However, as the speeds ramp up, my processor needs considerably more voltage. If I were to compare my 3.6GHz voltage requirements with that of mrfatboy's, the difference would be about 5 hundredths of a volt, or .05. Sounds small, but it very quickly adds up!

In closing, I would definitely stay under 1.5v actual voltage, not BIOS voltage.