C0 e8400 OC... yet another thread

[imported_wired247]

Before anyone asks, I've been searching around quite a bit (not just AT but also google), and I'm seeing a lot of conflicting information.

I got my e8400 as soon as it came out. It's C0, stepping 6 according to CPU-Z. As soon as I got it, I found it could hit 3.6GHz without much trouble. It has been that way for more than a year, in fact it was on stock cooling and the temps and voltages were always fine, always staying under 55-60C during normal usage, and voltage was right around stock.

Just in the last 24 hours I bought a sunbeam core contact freezer and I'm very happy with the temps it's giving me, I haven't seen anything over 60C even when priming. But this damn chip is giving me issues with OCing.

My mobo is the ASUS P5E / X38. The RAM is G.Skill DDR2-1066 at stock settings/voltage.

I was able to hit 500FSB on this mobo without much issue (very happy with the board!). Hardly even had to bump the NB voltage. So I know it's not the mobo holding me back. However, no matter how I attack it, this chip doesn't seem to want to put out more than about 4.0GHz.

If I run it at x9 multiplier and 450 or higher FSB, the damn chip wants a ridiculous amount of voltage to stay stable. We're talking like almost 1.5V just to prime for more than a few seconds.

If I lower the multiplier to x8.5 and run it at 480-490FSB, same story. The chip wants almost 1.5V.

Right now I'm having good stability for the time being (more prime95 pending) at 4GHz right on the nose (445FSB x 9), without pumping crazy voltage into the damn thing.

Is this typical? Am I missing something? Anyone else hit their wall at 4GHz? I've enabled loadline calibration, but I am not willing to do any pencil mods. IMO graphite belongs nowhere near a motherboard, that's end of discussion as far as pencil mods go. I don't believe FSB termination voltage is holding me back either. I've tweaked FSB termination up, down, and all over the map to 1.4V and it didn't affect stability from what I could tell. CPU PLL was kept at auto.

I'm not scared of eventually destroying the chip or gradual degradation, in fact I wouldn't get upset if I found myself with a good excuse to purchase a quad. But let's just say I have read about Aigomorla's 1.57V story and I don't want to be popping the chip in the span of a week.



So... other than Aigo's 1.57V what other voltages have been proven to destroy an e8400?


If possible I would appreciate SPECIFICS if you're citing people who have killed their e8400, rather than speculation such as "well, 1.4V sounds rather high..."

I have read all of that speculation, but what I've had trouble with is finding SPECIFIC stories such as Aigomorla's.

I know that Intel recommends no higher than 1.36V, for example. I have done my homework, but I still feel that I'm missing something here.

Thanks in advance.

 

[polarbear6]

have the same problem but i see a solution that is to get a north bridge cooler and try to raise the nb voltage and may be lower the vcore

i will post as soon as i get my lazy ass to the local hardware store and get my self a nb cooler

 

[Tempered81]

Originally posted by: wired247

So... other than Aigo's 1.57V what other voltages have been proven to destroy an e8400?

It's usually greater than 1.6v (real voltage measured in cpu-z during prime, or with a DMM). Or 1.2-1.4v when temperatures constantly remain too high. (90C+)

http://www.xtremesystems.org/f...211803&highlight=E8400

 

[imported_wired247]

Thanks jared, looking for more info like that. I feel that the 1.4V estimate is really just a parrot response, as I've not heard of an actual e8400 dying at 1.4, but practically everyone seems to agree that it's "getting a little high"

 

[polarbear6]

hey try uping the north bridge voltages
some times instability there also causes the errors in prime 95
cause some where in my heart i have a feeling that 4ghz can be achieved at vcore less that 1.4v

 

[imported_wired247]

NB is at about 1.41 right now. I'll give it a little more juice and lower the vcore... thanks

I have no idea what to do with CPU PLL. A few OC sites say it doesn't really make a difference. Any thoughts?

my original goal was 4.2GHz, but I realize that setting a number in my head is silly. So if the chip wants to be at 4.0 that's where it will stay.

 

[imported_wired247]

bump... I really don't think this chip wants to go above 4GHz...

tried raising NB / CPU PLL / FSB VTT / etc... didn't seem to increase stability by much.

The only thing I haven't tweaked is the CPU / FSB GTL's. for now I'm totally stable at 4GHz so I'll keep it here for a while.

 

[polarbear6]

well i heard of people wid ln2 cooling going as high as 1.888v vcore on a p4
and some similar thing on a e8600(6 ghz volts unknown) on ln2 thru youtube

so i think we can run pc's at high volts as long as we maintain them below 72c on stress test
but this is untested theory so be ware and always keep ur eyes open
and and even sleep wid eyes open ) hahahahahahahahaha

 

[Markfw]

Wehn I couldn;t get abouve 4.07 without crazy voltage, I said good enough. Had to go down to 3.98, but its been very stable there. Not worth it for the last 100-200 mhz IMO. And I settled on 1.3875.

 

[JMapleton]

I didn't think E8400s got above 4Ghz on air. I know only E8600 are assured a 4Ghz oc, but with 8400s it's a gamble.

 

[Tempered81]

A lot of At'ers here have E8400's at greater than 4ghz on air. This is my rig in sig on air cooling (the 8800gs sli has since been replaced by a gtx260):

Inside the case (same HSF as the OP):
http://img12.imageshack.us/img12/1217/caseem8.jpg

4.3ghz 5 & 1/2 hours prime:
http://www.xtremesystems.org/f...090&stc=1&d=1233706459

 

[polarbear6]

jared place i saw the pic and i even saw some fan on the north bridge so i figure u achieved this oc wid some nb volting only

so i wanted to ask u some question which i didnt so far managed to figure out as only few ppl seems to know

my question is what r the temperature ranges for the nb(System) i mean what is the max permissible temp and what are the max permisible nb volts
and what are the range for these nb volts

 

[Tempered81]

whew, i don't know man. you might be better off asking someone else.

my nb is at like 1.42v, 44C - I added a fan to it to keep it cool. Passive it was scalding!

 

[polarbear6]

well my nb is behaving in a very different manner i even posted this in a new post
when i increase thenb volts by 0.1v it becomes instable and crashes but if i decrease it runs fine(not on stress test on normal usage stress test require 1.418v so i dint do stress test)

what does the 0.1v mean i mean what are the starting value

 

[imported_wired247]

i'm pretty sure I just didn't get the best chip. not a problem really... I'll probably upgrade CPU in 6 months or so. It's not even running very hot, just won't get stable over 4ghz without insane voltage

maybe I just suck at OCing, this is very possible

 

[Tempered81]

wired247: perhaps take a screenshot of cpuz/prime & tell us your bios settings?

Here's an excerpt from xbitlab's P5E review:

From overclockers? standpoint Asus P5E mainboard features a very good BIOS Setup. Let?s see how Asus P5E will behave in practical overclocking experiments. Our overclocking testbed included not only this actual board, but also 2GB DDR2 from Corsair (Dominator TWIN2X2048-8888C4DF), OCZ GeForce 8800GTX graphics card, Western Digital Raptor WD1500AHFD HDD and SilverStone SST-ST85ZF power supply unit. We used a Zalman CNPS9700 LED processor cooler, too.

First of all we decided to determine the maximum FSB frequency when the board would remain stable and reliable. We tested with a Core 2 Duo E6750 processor. For stability check we ran Prime95 25.3 program for 30 minutes in Large FFT and Blend modes.

Asus P5E proved totally up to our expectations. Our processor was running stably without any problems at up to 440MHz FSB frequency. We didn?t even have to change any voltages other than Vcore and Vmem for that. However, further overclocking led to evident instability, so we had to dig a little deeper into system fine tuning.

Further tests showed that FSB frequency increase above 440MHz requires raising the North Bridge voltage. By raising this parameter only to 1.4V we could push the FSB speed to 465MHz. with the North Bridge voltage at 1.55V the board runs flawlessly at about 500MHz FSB, although in this case an additional fan is required on the chipset cooling system.

The maximum overclocking result was achieved at 1.65V North Bridge voltage. In this case the board would start and remain stable at 520MHz FSB.

We didn?t have to increase CPU PLL Voltage and FSB Termination Voltage, they were at their default values of 1.5V and 1.2V respectively. The memory was clocked using 1:1 divider at 1040MHz with 4-4-4-12 timings.

I have to stress that further increase of the North Bridge voltage didn?t have any positive effect. The FSB overclocking potential would drop that indicated chipset overheating and required replacing the chipset cooling system with something more efficient.

However, contemporary mainboards have been coping successfully with dual-core processors overclocking for quite a while now. It is the quad-core solutions that seem to be of most interest these days. Therefore, the second part of our overclocking experiments was devoted to Core 2 Extreme QX6850 overclocking.

I would like to say right away that quad-core processors overclocking on Asus P5E mainboard is as simple and quick as dual-core processors overclocking. To achieve maximum results it is enough to simply raise the processor Vcore, Vmem and chipset North Bridge voltage following almost the same rules as in the previous case.

However, maximum FSB frequency when the board remained stable turned out considerably lower than in the previous case. Our best result is 460MHz.

We increased the chipset North Bridge voltage to 1.55V; CPU PLL Voltage, FSB Termination Voltage and CPU GTL Reference were at their nominal values. The memory was working at 1105MHz with 4-4-4-12 timings, i.e. we used 5:6 FSB:Mem divider.

By the way, the North Bridge temperature hits 70-75ºC during such quad-core processor overclocking, despite the additional fan included with the mainboard. It is another indication that Intel X38 platforms are more dependent on efficient chipset system cooling than Intel P35 based ones, because the new North Bridge dissipates almost twice as much heat.

 

[imported_wired247]

yeah I've read that... appreciate all the help

 

[soccerballtux]

Originally posted by: wired247
Thanks jared, looking for more info like that. I feel that the 1.4V estimate is really just a parrot response, as I've not heard of an actual e8400 dying at 1.4, but practically everyone seems to agree that it's "getting a little high"

It's not, Anand ran a suicide run with a 45nm chip at 1.5v and found it broke after 24 hours.

That's why we say 1.4v. Also, Intel's spec sheets.
There's some math that goes on behind the scenes that I'm not interested in working right now-- some electron probability stuff, chances of the electron tunneling into the gate dielectric and getting stuck there-- this is why chips eventually die-- the silicon doping levels become permanently changed. As the node shrinks the probability of the electron showing up outside the boundary of the potential well rises. For larger nodes it's not a problem, but as the node shrinks you have to lower the voltage (IE 1.5v @ 65nm to 1.4v @ 45nm) to shrink the probability that Mr. Electron jumps outside of where he's supposed to be.

 

[manimal]

I have had my C0 running 24/7 at 1.3875 as well at a hair under 4 ghz for over 6 months now. The difference really between it at when I ran it at 3780 at 1.31 is very small. I do keep it at 4 since my 4850s are starved for cpu in some games I am playing now.Loaded mine has a tad of vdroop down to 1.344v so I feel comfortable with that. I could get stable at 4.2 ish with 1.5 but that scared the begeezus outta me so Ill play it relatively close to the vest.

I am sure you did try loosening up that ram to try to isolate your instability right?

 

[nerp]

I think getting a C0 @ 4ghz stable is a luck of the draw. Some people manage it, others never will. I think settling at a safe voltage/speed is the way to go. I could run 3.8ghz 24/7 but I opted for 3.6 at very low volts.

 

[imported_wired247]

I appreciate all the advice, I'm going to reconsider my goals and options and if I need anymore assistance I'll start a different thread. Right now I'm getting fed up with noise out of this setup...

Originally I thought this sunbeam CCF would be a much quieter CPU cooler than intel, but I'm finding the supplied fan controller (cheap ass potentiometer) is a piece of crap, and the fan has 3 settings:

about 40% of the range of the potentiometer gives me: FAN OFF :disgust:
a tiny sliver (0.1%) of the potentiometer gives me: FAN ULTRA HIGH SPEED ZOMG :shocked:

The rest (twist the knob in either direction away from ultra high speed) gives me fan low speed, but get this... I get occasional fan shutoffs??

I think I got a FUBAR potentiometer, so I plugged the CPU fan directly into the mobo.

Now it is on Ultra high speed again... and ASUS Q-FAN cannot figure out how to regulate the fan below 2000RPM.

So right now I'm a bit pissed off about this sunbeam fan controller. I am going to purchase some low noise case fans... use one of them to replace the sunbeam fan, replace the others on my case as well.

Now that means I will have less cooling. So I think I'm probably going to set my CPU back to 3.6GHz, and settle for a 20% overclock with low noise.

I am almost certainly going to upgrade the CPU once before I switch over to a corei7.



So as you can see, I'm feeling rather schizophrenic about my options. So thanks for the help... and trust me I've tweaked every parameter beyond what most people would say is safe, to try to get this chip stable at or above 4GHz. This chip is just not going to work with me here, so for now, it's back to the safe zone of lower volts and 3.6GHz.

 

[Markfw]

I got the Sunbeam cooler. The biggest problem it has ? not enough pressure. If I could figure out a way to make those springs stronger, it would down the heat a LOT. I did that on my TRUE's, as I had to mount them like my big typhoon. since there was already a backplate on my DQ6. The more pressure, the lower the temp seems to get at the same fan speed !

Oh, and I didn't even install the fan regulator, I just left it full, and I can't hear it anyway.

 

[imported_wired247]

That's funny you mention it... I thought the retention spring on the sunbeam is great! (although you would know more about CPU coolers... the TRUE is a bit expensive for what it is if you ask me!)

I know all about contact pressure and thermal resistance, I'm a chemical engineer!

Well as I said, TEMPS are not my problem at this point... it's just the chip itself refusing to get stable.

 

[M1A]

As in my sig I am runnin a CO for over a year now at 4.2 with 1.36v in bios but I got an early chip and they seem to do better than the later ones....

 

[imported_wired247]

Yes I must have gotten unlucky. When other people were saying their chips could do like 3.6GHz at stock voltages, mine never could.

I got the EARLIEST e8400 (other than engineering samples), bought it on launch day, right before every vendor in the western world ran out of them.