Quick E8400 voltage question.. Too much?

[maxwolfie]

Hey guys,

I'm currently Orthos'ing my CPU at 4.2ghz, 1.49v in BIOS, however I get massive vdrop & vdroop

Idle in CPU-z is 1.456v
Load in CPU-z is 1.424v

Any lower at all and my 4.2ghz overclock becomes unstable in Orthos only after an hour. So far, with the above settings, it's 11 hours Orthos stable.

I have decent air cooling, a Zalman, with a lapped IHS on the E8400. Reported temps in CoreTemp 0.96.1 are 72c/70c. Although I've read from multiple sources that the temp sensors in these things tell lies quite often.


Is this overclock "safe" for 24/7 use? I'm a little worried about the slightly-high vcore

[cadaveca]

Although mine seems to be dying form the same voltage, or something, your individual core temps are pretty much the same.

On my dying cpu, there's a 15c difference. This difference could be part of why it died..I dunno. Intel states 1.45v max, and it will be at least a few months before anyone can accurately answer your question.

[GLENBOY]

sad to hear that is it your q746a one, you had nice clocks on that , at least it a cheap cpu, not like a qx9650

[maxwolfie]

My idle temps are 45c/28c

I've stopped Orthos'ing after you said that!

[cadaveca]

sad to hear that is it your q746a one, you had nice clocks on that , at least it a cheap cpu, not like a qx9650

you must be thinking of someone else, it's a 740 with a "t" at the end.



maxwolfie, my idle temps are 40/40, regardless of cooling, even with slightly-subzero water. And it's not compeltely dead yet, but you can rest assured I'll be posting when it is completely dead, and it won't be some other component causing the problem...other than maybe the board. The second cpu might confirm it's the board...but i think it won't.

I noticed degredation on 65nm too, but not this severe...approx 50mhz after 6 months of crunching @ 1.55v(Intel stated max for 65nm).

[GLENBOY]

sorry yes you're right it cordovader, sorry about that ,

[JohnZS]

I would hold fire on overclocking especially if that involves raising the vcore on the 45nm CPU's at the moment as there are serious issues for concern.
Check out the last few pages of this thread here (E84000 thermal related issues)
Basically there have been reported cases of Sudden Death Syndrome, or "Electrodegridation" of many 45nm CPU's
Now, many people say it is due to over volting (and this is true) as the Intel Technical Specifications Datasheets list 1.45V as the absolute Maximum that the core will tolerate, it's usually wise to back down from this by a fair way due to inaccurate BIOS settings of the Voltage... or power spikes.
The other issue of grave concern is that some motherboards.. (including my ASUS P5E3) appear to overvolt certain settings @ STOCK!

The setting of concern being "FSB TERMINATION VOLTAGE"

The Intel Datasheets specify that the QX9650 CPU should have a FSB Termination Voltage of no more than 1.15V.... yet the LOWEST value I can set on my motherboard is 1.20V!

I would hold fire seriously hold fire on ANY overclocking until some sort of clarification has been made as to why so many people are having CPU's die, erratic temperatures or degradation.

I would also read the back end of the thread I linked.. and read up on the datasheets.....

Perhaps we need BIOS updates?

I don;t know, but one thing is for sure... 45nm is VERY sensitive to voltages.

John

[unclewebb]

I think everyone pushing these new chips with Conroe type voltages needs to read the Intel datasheet very carefully. Max voltage is defined as 1.45 volts which Intel states is outside the functional limits of the processor. When outside the functional limits, neither functionality nor long-term reliability can be expected.

The Max for the 65nm Conroe processors was 1.55 volts but they seemed more tolerant of a little bit of excessive voltage. The jury is still out for the 45nm. People might need to keep them at 1.40 volts under full load for long term reliability.

Also avoid the AUTO setting for core voltage in the bios. I know my P5B Dlx gets very generous, over 1.60 volts, when overclocking if the board is set on AUTO. That might fry up a 45nm chip pretty quick.

http://download.intel.com/design/pro...s/31873201.pdf

2.6 Voltage and Current Specification

2.6.1 Absolute Maximum and Minimum Ratings

Table 3 specifies absolute maximum and minimum ratings only and lie outside the functional limits of the processor. Within functional operation limits, functionality and long-term reliability can be expected.

At conditions outside functional operation condition limits, but within absolute maximum and minimum ratings, neither functionality nor long-term reliability can be expected. If a device is returned to conditions within functional operation limits after having been subjected to conditions outside these limits, but within the absolute maximum and minimum ratings, the device may be functional, but with its lifetime degraded depending on exposure to conditions exceeding the functional operation condition limits.

At conditions exceeding absolute maximum and minimum ratings, neither functionality nor long-term reliability can be expected. Moreover, if a device is subjected to these conditions for any length of time then, when returned to conditions within the functional operating condition limits, it will either not function, or its reliability will be severely degraded.

[maxwolfie]

Cool. So 1.45v max. The problem is that the BIOS, CPU-Z, Gigabyte EasyTune, SpeedFan etc etc ALL read completely different vcores. And then I also have to think about vdroop differences. I'm contemplating dropping my BIOS voltage down to 1.4v, and running just at 4ghz for the time being. BIOS voltage at 1.4v means about 1.360v in CPU-z idle, and about 1.344 or thereabouts CPU-z under load.

When you say 1.45v MAX, would it be better to assume 1.45v idle max, or 1.45v load max? Assuming that there is about 0.03 - 0.05v difference just in vdroop. It also seems apparent that people are posting up many of their "proof screenshots" while prime/orthos is still loading the CPU, along with a screenshot of CPU-z. This indicated that people are showing off their LOAD vcore's, not their idle vcores.

1. Does this mean that I can push my CPU (theoretically) to 1.45v under LOAD "safely"?
2. Would I have damaged the CPU at all, running it at 1.456v idle/ 1.424v load Orthos for 12 hours straight? I suppose it would need more time to make any real difference, plus the vcore wasn't *that* high anyways

[emoners]

so 1.45v max for the vcore... how about the fsb vtt?

[unclewebb]

Let me try this again.

Intel says that the maximum core voltage for the E8x00 series is 1.45 volts but 1.45 volts is outside the functional limits of the processor where neither functionality or long-term reliability can be expected.

How long do you want to use your new CPU for before burning it out? Even at only 1.45 volts there are no guarantees. People feeding them 1.60 volts and trying to Prime at 4500+ MHz are living on the edge.

If you go over 1.45 volts, even once, Intel says that when you return your voltage to a normal level that it either may no longer function or its reliability will be severely degraded.

I've been reliably running my 65nm E6400 for the last year at 1.50 volts at idle and at about 1.48 volts at full Prime load. Intel has reduced their Max volt number by 0.10 volts so I would reduce those numbers by 0.10 volts for the new 45nm series. Until this issue gets cleared up and more user reports come in I wouldn't go over 1.40 volts under full load.

CPU-Z seems accurate at reporting core voltage on most motherboards.

[maxwolfie]

Until this issue gets cleared up and more user reports come in I wouldn't go over 1.40 volts under full load.

Cool - I can probably do 4.2ghz stable @ 1.408v under load (according to CPU-z). Idle vcore is about 1.456v. Or is that pushing it too for 24/7?

[unclewebb]

maxwolfie: Just to be on the safe side you should have a look around for a volt mod for your board to reduce vdroop. No use feeding your CPU extra voltage at idle if you don't have to.

[maxwolfie]

Will do

Currently Orthos'ing at:

4.2ghz
1.39v load (CPUID HM)
1.42v idle (CPUID HM)

Stable so far

EDIT: Make that not stable after 3 minutes :P Not much I can do now other than drop the FSB. I don't want to go any further with vcore

[cadaveca]

If a device is returned to conditions within functional operation limits after having been subjected to conditions outside these limits, but within the absolute maximum and minimum ratings, the device may be functional, but with its lifetime degraded depending on exposure to conditions exceeding the functional operation condition limits.

SO, maybe this is approached wrong, even by you....maybe we must not go back to lower voltages after going high...

But this is how I've always "burnt-in" my parts....start low, scale clocks within volt limits, then I'll do the reverse, trying to find the point where even scaling stops(when voltage must need a large increase for same clockspeed increase).

I remeber an INtel guy posting on here that we should scale voltage slowly...perhaps we should truly take this to heart.

[JohnZS]

MaxWolfie
What motherboard do you have?
If it is an ASUS P5E based (X38) be very wary as those at stock overvolt the FSB Termination Voltage.
I am not sure but the datasheet suggests that the FSB Termination Voltage should be no more than 1.15Volts?
P5E3 has a minimum selectable value of....1.20V!
Likewise maximum CPU PLL is 1.55Volts
I am holding fire until ASUS resolve this issue.
BTW To achieve stability at a lower VCore, try enabling "Load Line Calibration"
John

[GMX]

I would hold fire on overclocking especially if that involves raising the vcore on the 45nm CPU's at the moment as there are serious issues for concern.
Check out the last few pages of this thread here (E84000 thermal related issues)
Basically there have been reported cases of Sudden Death Syndrome, or "Electrodegridation" of many 45nm CPU's
Now, many people say it is due to over volting (and this is true) as the Intel Technical Specifications Datasheets list 1.45V as the absolute Maximum that the core will tolerate, it's usually wise to back down from this by a fair way due to inaccurate BIOS settings of the Voltage... or power spikes.
The other issue of grave concern is that some motherboards.. (including my ASUS P5E3) appear to overvolt certain settings @ STOCK!

The setting of concern being "FSB TERMINATION VOLTAGE"

The Intel Datasheets specify that the QX9650 CPU should have a FSB Termination Voltage of no more than 1.15V.... yet the LOWEST value I can set on my motherboard is 1.20V!

I would hold fire seriously hold fire on ANY overclocking until some sort of clarification has been made as to why so many people are having CPU's die, erratic temperatures or degradation.

I would also read the back end of the thread I linked.. and read up on the datasheets.....

Perhaps we need BIOS updates?

I don;t know, but one thing is for sure... 45nm is VERY sensitive to voltages.

John

this is a very good point. I've updated to BIOS 705 on my P5k and it says in description for FSB termination Voltage: When 45nm CPU detected all voltage settings are basically -.1v.

I forgot the exact settings but I think they were 1.2v, 1.3v, 1.4v, when 45nm is detected they become, 1.1v, 1.2v, 1.3v

[OnDborder]

On my dying cpu, there's a 15c difference. This difference could be part of why it died..I dunno. Intel states 1.45v max, and it will be at least a few months before anyone can accurately answer your question.

"difference"? Like mine? Mine is 22 degrees difference at idle..
http://www.xtremesystems.org/forums/...6&postcount=48

[unclewebb]

I've been comparing the Intel datasheets today for the E6000 vs E8000.

As previously mentioned, the FSB Termination voltage has dropped from a recommended 1.20 volts +/- 5% to 1.10 volts +/- 5% with the switch to 45nm.

The E6000 had an absolute maximum voltage of 1.55 volts but the maximum VID was 1.50 volts. When operating between 1.50 and 1.55 volts you were in that gray area where "neither functionality nor long-term reliability can be expected."

With the E8000 the absolute maximum has dropped 0.10 volts to 1.45 volts but the maximum VID has dropped even further down to only 1.3625 volts. This implies that the new gray area for the 45nm cpus is from 1.3625 to 1.450 volts where "neither functionality nor long-term reliability can be expected."

It might be excessive FSB Termination voltage that is degrading some of the new 45nm chips or it might be good old excessive CPU voltage. Most XS overclockers would not consider 1.3625 volts to be excessive but it's possible that even one run above 1.3625 volts was enough to degrade the performance and long-term reliability of your chip.

[emoners]

I've been comparing the Intel datasheets today for the E6000 vs E8000.

As previously mentioned, the FSB Termination voltage has dropped from a recommended 1.20 volts +/- 5% to 1.10 volts +/- 5% with the switch to 45nm.

The E6000 had an absolute maximum voltage of 1.55 volts but the maximum VID was 1.50 volts. When operating between 1.50 and 1.55 volts you were in that gray area where "neither functionality nor long-term reliability can be expected."

With the E8000 the absolute maximum has dropped 0.10 volts to 1.45 volts but the maximum VID has dropped even further down to only 1.3625 volts. This implies that the new gray area for the 45nm cpus is from 1.3625 to 1.450 volts where "neither functionality nor long-term reliability can be expected."

It might be excessive FSB Termination voltage that is degrading some of the new 45nm chips or it might be good old excessive CPU voltage. Most XS overclockers would not consider 1.3625 volts to be excessive but it's possible that even one run above 1.3625 volts was enough to degrade the performance and long-term reliability of your chip.

great info uncle!

so let me just get it straight (sorry, it's 3am where i'm at so my brain is kinda slowing down )... for 45nm procs - VCore 1.3625 to 1.450 & VTT 1.10 volts +/- 5%...?

thanks!

[Mav451]

Good lord...so what happens when we need more voltage? Just be grateful with what we get?



It's almost like...high-end mobos or cooling no longer matter. The chip ITSELF is the bottleneck, based on electrical limits. *sigh*

[unclewebb]

so let me just get it straight (sorry, it's 3am where i'm at so my brain is kinda slowing down )... for 45nm procs - VCore 1.3625 to 1.450 & VTT 1.10 volts +/- 5%...?

According to the docs for 45nm, 1.3625 to 1.4500 volts for the core seems to be in the gray area where "neither functionality nor long-term reliability can be expected."

Intel is likely being a little conservative here to keep the warranty claims down. You might want to take it easy until we find out a little more info about what percentage of the new E8x00 processors are having degraded performance issues after overclocking.

Minor problems can get blown out of proportion on a forum. Running reliably 24/7 at 4.0 GHz to 4.2 GHz should be doable on a good board with good memory and without having to get too crazy with the core voltage.

Here's a good review over at Legion Hardware with lots of benchmarks comparing an E8400 at 4.0 GHz to 4.2 GHz to 4.4 GHz. I must be getting old because I no longer see the use of risking your CPU in order to gain an extra 1 or 2 fps.

[maxwolfie]

MaxWolfie
What motherboard do you have?
If it is an ASUS P5E based (X38) be very wary as those at stock overvolt the FSB Termination Voltage.
I am not sure but the datasheet suggests that the FSB Termination Voltage should be no more than 1.15Volts?
P5E3 has a minimum selectable value of....1.20V!
Likewise maximum CPU PLL is 1.55Volts
I am holding fire until ASUS resolve this issue.
BTW To achieve stability at a lower VCore, try enabling "Load Line Calibration"
John

It's a GA-P35-S3, and I've just recently done a vdrop/vdroop mod.

It's now NEVER over 1.39v, idle OR load. Stable @ 4.1ghz so far..

[jvandecar]

Question.

My E8400 just came in today, not even out of the box.

The side of the box says 1.225V max

Is this what y'all are referencing?

[Oderus]

jvandecar: That's the MAX voltage of the CPU according to Intel, yes. There is also a VID which shows under Coretemp which is different. My e8400 shows a VID of 1.08v but the max is 1.22v and people are saying up to 1.4v is safe.