CPU Electron Migration and You!

[cadaveca]

In fact all Electromagnetic Radiation is AC in nature and would have to be coming from a high power transformer for a current to even exist. EM will never occur from natural or household sources with current day consumer electronics/computer parts.

What about surface induction chargers? I guess they don't use EMR, exactly...but will they cause EM, no?

http://www.shoppureenergy.com/wire-free-solutions/

This one uses magnetic induction:

http://www.todayshottrends.net/computers-electronics/powermat-wireless-charger.html

In order for the right info to come out someone does not need to be wrong. There only ever needs to be the truth. If you are purposefully pushing out the wrong information to entice conversation than you are doing the people of TPU a great disservice. I don't take your comments as disrespectful, but your blatant disregard for being at least semi-accurate or poignant is room for me to consider your responses to be more trollish than helpful. I don't like to throw around the term troll, and I hope if you are not then you assume that you are not. The issue I have is not with your source (even though you provided none I assume you may have some sort of background... at least a highschool degree since you use the term Covalent Bond) but why you present your information and what you hope to achieve with your bias which has proven to me and my sources to be harmful. Remaining innocent or helpless in a discussion like this is not what a reader should feel. This subject is not beyond anyone, and it's an issue that can be explained with little to no special vocabulary.

I can see your point of view, but had I not posted, you may not have either. Like I said, I don't mind being wrong...you see it as a disservice...but the education provided is a service rendered. People used to think the world was flat...

The difficulty in accepting another's culture is understanding what created that culture.

EDIT: I do not think people should feel comfortable overclocking, BTW...it's voiding thier warranty...if it was perfectly fine, it wouldn't have any effect on warranty. Myself, warranty gone, I feel perfectly comfortable pushing the limits.

 

[3volvedcombat]

Will i be fine if i keep the voltage of my q9550 at 1.232volts at 4.0Ghz, hahaahahahahaa. I might not need 4.0Ghz and i might run my processor at 3.8Ghz at 1.168volts. I think i have a golden q9550, it can pass 30 runs of prime 95 at 3.1Ghz at 0.992volts. I was pretty amazed with the processor when it did that, and managed to be stable. Does running a processor below voltage increase its lifespan? I mean I havnt read a thread on over-clocking were the facts show processor lifespan increase with lower voltage . Also if i run at 3.8Ghz my loads will be less then 55c on intelburntest.

 

[Binge]

Will i be fine if i keep the voltage of my q9550 at 1.232volts at 4.0Ghz, hahaahahahahaa. I might not need 4.0Ghz and i might run my processor at 3.8Ghz at 1.168volts. I think i have a golden q9550, it can pass 30 runs of prime 95 at 3.1Ghz at 0.992volts. I was pretty amazed with the processor when it did that, and managed to be stable. Does running a processor below voltage increase its lifespan? I mean I havnt read a thread on over-clocking were the facts show processor lifespan increase with lower voltage . Also if i run at 3.8Ghz my loads will be less then 55c on intelburntest.

Bingo! You see the objective behind iron-clad "safe" OCs. The best anyone can hope for is to achieve a clock speed at stock voltages. Voltages higher than stock may be safe to a degree taking into consideration the margin of manufacturing process error, but the real win is when someone receives a golden chip like the one described in this quote.

 

[TheShad0W]

am I right in thinking that in practice, so long as you don't push crazy voltages/temperatures then you should be able to stave off degradation for a few years? My chip's within intel's recommended votage limits by a fair margin - and despite being at 4Ghz it runs under load at slightly lower temperatures than at stock with the stock cooler...

 

[cadaveca]

In practice, yes. IN reality, process variations prevent anyone knowing for sure. AMD recommended 1.55v is safe, yet others do not feel comfortable pushing that voltage.

 

[Sasqui]

Here's a really good article about processor life and failure. It doesn't get into any nitty-gritty about what physically causes the failure, just a high level of the statistics behind warrantee lengths and overvolting:

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3251&p=6

Some time ago (years), I wrote a thread asking who has seen a processor fail for no apparent reason... other than some accident or obvious problem - like forgetting to put the heatsink on or total abuse. The answer was no one. W1zzard even chimed in and said he had never seen a processor just fail for no reason.

My theory is that people seeing "degredation" are witnessing other, non-solid state components on the motherboard such as caps falling apart. One exception to this I've seen first hand are IC chips on memory modules - I don't know what the difference is there. Silicon is pretty reliable stuff.

 

[phanbuey]

Here's a really good article about processor life and failure. It doesn't get into any nitty-gritty about what physically causes the failure, just a high level of the statistics behind warrantee lengths and overvolting:

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3251&p=6

Some time ago (years), I wrote a thread asking who has seen a processor fail for no apparent reason... other than some accident or obvious problem - like forgetting to put the heatsink on or total abuse. The answer was no one. W1zzard even chimed in and said he had never seen a processor just fail for no reason.

My theory is that people seeing "degredation" are witnessing other, non-solid state components on the motherboard such as caps falling apart. One exception to this I've seen first hand are IC chips on memory modules - I don't know what the difference is there. Silicon is pretty reliable stuff.

Ive never seen a processor just stop working... but I have seen a 100% stable OC become unstable after 3 months... and when I lowered the OC slightly but kept the voltage high, after another 3 months even the lowered OC was no longer stable...

Basically after 6 mos of my Q6600 on 1.5v - it went from 3.72Ghz, to 3.6Ghz, to 3.53 Ghz. All of those were linpack stable for 4hrs at first, then after continuous use, the 3.72 Ghz OC would BSOD after 2 min of linpack. It could have been the MB but I doubt it.

I put the chip in a different motherboard... and it would no longer run at stock on stock volts.

The chip was watercooled and never broke 65C EVER... even linpack didnt do it - so majority of the time it was running in the low 40's.

 

[Sasqui]

Ive never seen a processor just stop working... but I have seen a 100% stable OC become unstable after 3 months... and when I lowered the OC slightly but kept the voltage high, after another 3 months even the lowered OC was no longer stable...

Basically after 6 mos of my Q6600 on 1.5v - it went from 3.72Ghz, to 3.6Ghz, to 3.53 Ghz. All of those were linpack stable for 4hrs at first, then after continuous use, the 3.72 Ghz OC would BSOD after 2 min of linpack. It could have been the MB but I doubt it.

I put the chip in a different motherboard... and it would no longer run at stock on stock volts.

The chip was watercooled and never broke 65C EVER... even linpack didnt do it - so majority of the time it was running in the low 40's.

I had a similar story with a P4 chip, but never another mobo to test it with. So the question begs if it's the silicon itself, or the other components that make up the CPU? I have no answer.

 

[TheShad0W]

So, my i5 is stable at the current voltage and at a setting 8mV below (currently set to hit 1.336 under load and 1.304 at idle). Anyone got a ballpark idea for what sort of lifespan I can expect?

 

[Binge]

So, my i5 is stable at the current voltage and at a setting 8mV below (currently set to hit 1.336 under load and 1.304 at idle). Anyone got a ballpark idea for what sort of lifespan I can expect?

If you're within spec then the average lifespan should be 5-10 years with correct cooling. Every 10C you reduce on temperatures will lengthen the life of the part 2-5 years and the opposite for every 10C greater average temps.

The i5/i7 lifespan is really tentative because of the on-chip components which may or may not fail faster than the CPU core itself.

 

[Wile E]

Here's a really good article about processor life and failure. It doesn't get into any nitty-gritty about what physically causes the failure, just a high level of the statistics behind warrantee lengths and overvolting:

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3251&p=6

Some time ago (years), I wrote a thread asking who has seen a processor fail for no apparent reason... other than some accident or obvious problem - like forgetting to put the heatsink on or total abuse. The answer was no one. W1zzard even chimed in and said he had never seen a processor just fail for no reason.

My theory is that people seeing "degredation" are witnessing other, non-solid state components on the motherboard such as caps falling apart. One exception to this I've seen first hand are IC chips on memory modules - I don't know what the difference is there. Silicon is pretty reliable stuff.

Your theory would be incorrect then. I killed 3 Brisbanes with severe degradation at 1.55v for a few months. Placing them in entirely different boards still had them failing at stock speeds and voltages. Once you take the board out of the equation, and the chips still fail, the fault lies within the chip, not the other components.

I still maintain that 1.55v is pushing it on Phenom II chips. 65 and 90nm weren't safe on 1.55v for 24/7 usage, I somehow doubt an even smaller process would be, not to mention I've already seen people running at these voltages complaining about how they've had to lower their OCs to remain stable. That screams degradation.

 

[Sasqui]

Your theory would be incorrect then. I killed 3 Brisbanes with severe degradation at 1.55v for a few months. Placing them in entirely different boards still had them failing at stock speeds and voltages. Once you take the board out of the equation, and the chips still fail, the fault lies within the chip, not the other components.

I still maintain that 1.55v is pushing it on Phenom II chips. 65 and 90nm weren't safe on 1.55v for 24/7 usage, I somehow doubt an even smaller process would be, not to mention I've already seen people running at these voltages complaining about how they've had to lower their OCs to remain stable. That screams degradation.

Perhaps. I'm right most of the time, not all

The question is it in the silicon wafer itself, or the other components that comprise the CPU package.

 

[Wile E]

Perhaps. I'm right most of the time, not all

The question is it in the silicon wafer itself, or the other components that comprise the CPU package.

It's generally the insulators between the transistors on the die.

 

[aCid888*]

I'll just chime in real quick here. I used to go through E8500's almost daily. I probably clocked over 50 of them. I noticed after about 6 months of running my "golden" chip at 4.75 GHz @ 1.52v under water, for occasional benches, and 4.5 GHz @ 1.45v under water, it needed a voltage bump to keep hitting those speeds.

I can say I've had similar "issues" with one of my chips.

My E8400 "suicide" run to 4.95GHz @ 1.56v added 0.05v to the voltage I needed to run that chip at its usual speed of 4GHz.

 

[Binge]

Perhaps. I'm right most of the time, not all

The question is it in the silicon wafer itself, or the other components that comprise the CPU package.

It's generally the insulators between the transistors on the die.

Metal lithography is how traces are made that connect all components in a chip. It's these traces that are most prone to EM.

 

[exodusprime1337]

Your theory would be incorrect then. I killed 3 Brisbanes with severe degradation at 1.55v for a few months. Placing them in entirely different boards still had them failing at stock speeds and voltages. Once you take the board out of the equation, and the chips still fail, the fault lies within the chip, not the other components.

I still maintain that 1.55v is pushing it on Phenom II chips. 65 and 90nm weren't safe on 1.55v for 24/7 usage, I somehow doubt an even smaller process would be, not to mention I've already seen people running at these voltages complaining about how they've had to lower their OCs to remain stable. That screams degradation.

the data sheets which i posted in another thread here, and read through extensively point out that 1.5 is the deteremined "nominal Voltage" for most of the phenom 2 series.

Keep in mind that electromigration is not the only factor that limits and oc over time, stress on the nb(on the board), memory, motherboard itself, power supply, vrm, pcms, and many many other internal features can cause an overclock to fail over time, not just electromigration in the core.

Do you all remember "burning in a processor" that was supposed to help increast and stabilize a proccy at oc speeds by causing it to experience high heat... I actually posted a guide for it myself a couple years ago. As an electrical engineer myself, theory says that this is possible.. and practice says electromigration is a real threat.. however how much of a threat is it..

my core2 in my notebook runs at 60-70c full load and this is considered normal at 1.1 or so volts... and desktop core 2s are kept between 50 and 60 and most consider that pushing it.. given our chips are made using the same high k soi process, am i succeptable to a greater rate of em then others... probably.. but how long is it going to take.

EM as i stated can happen faster with high heat and can be slowed with lower temps. My phenom 2 has been running happily at 3.9Ghz 1.535v for almost a year now, loads at 43c and idles at 27 on water... it's probably not going to last as long as say an un oc'd p2 940, however the time it'll take for em to take place is much longer then i find the intended use to be, so i don't worry about it.

the video and image you showed was an accellerated imaging used to show em taking place. the voltage and amperage is far far more then the rated trace line can handle and so i burns up quickly, using this to show it should not influence or scare people into voltages

I find this thread altogether useless in the fact that it takes away from what we do here. We void warranties, we burn up chips and cards and ram and mobo's, we experiment with the cutting edge and the push the envelope of everything we can get our hands on, if i could oc the computer in my car i would, it's just what we do. EM is such a hard item to pin a timeframe on and whether it happens faster or slower for someone doesn't mean you'll see the same results as the tolerance used in engineering a processor is broad despite it's small size.

I would just worry about the basic numbers, phenom 2s are rated for anything up to 1.5, over that your pushing it, will your chip die tomorrow probably not unless you went way over, then your a moron, Time and time again amd extreme oc's have push 1.7, 1.6, 1.5, and for a while we were undervolting our process which cause another form of failure simialar to EM when the trace collapses because of the lack of force behind the current. All of these things can kill our processors, leave EM up to the engineers to deal with, just be smart about your oc's and don't kill your proccy's out of stupidity

 

[MilkyWay]

so what is said is thast higher voltages increase degeneration? i thought that was obvious

even on my X3 720BE i am pretty sure it needs more voltage to run at 3.5ghz than it first did but then again im still getting used to this new motherboard so it could be other factors

i thought heat was less of an issue than voltage, as long as its not to hot its okay but a high voltage could be bad in the long run

 

[r9]

Everyone is assuming that stock volts are the safest they can get. But what if the manufacturer is using higher stock volts to achieve higher clock in situation where it has low efficiency CPU and needs higher frequency to be competitive ?

What about looking at GPU 4850 and 4870 use the same GPU even though 4870 GPU are selected they still need higher voltages to achieve their stock volts. So which one is overvolted or undervolted ?

The CPU on stock volts would last for 30 years. I don`t plan to use my E5200 for that long so I`m consciously using higher volts and I don`t mind if I loose 25 years of my cpu life.

My point is that even stock volts aren`t calculated by formula they are relative also.

 

[3volvedcombat]

Everyone is assuming that stock volts are the safest they can get. But what if the manufacturer is using higher stock volts to achieve higher clock in situation where it has low efficiency CPU and needs higher frequency to be competitive ?

What about looking at GPU 4850 and 4870 use the same GPU even though 4870 GPU are selected they still need higher voltages to achieve their stock volts. So which one is overvolted or undervolted ?

The CPU on stock volts would last for 30 years. I don`t plan to use my E5200 for that long so I`m consciously using higher volts and I don`t mind if I loose 25 years of my cpu life.

My point is that even stock volts aren`t calculated by formula they are relative also.

I think im still safe if i run at these settings anyways, when it comes to processor's and loss of life and performance. How about we just all run under a volt and use water cooling, and make 500+ dollar motherboards with future tech, and rare RAM that can withstand anything with magical materials we havnt formulated yet, HEY PROBLEM SOLVED!.

 

[TheShad0W]

The CPU on stock volts would last for 30 years. I don`t plan to use my E5200 for that long so I`m consciously using higher volts and I don`t mind if I loose 25 years of my cpu life.

Depends if you're at 1.41v/4.2Ghz as your "system specs" states, or 1.7v/4.8Ghz as in your sig.

I wouldn't be just tremendously confident at an E5200 making it through five years at 1.7v

 

[3volvedcombat]

Depends if you're at 1.41v/4.2Ghz as your "system specs" states, or 1.7v/4.8Ghz as in your sig.

I wouldn't be just tremendously confident at an E5200 making it through five years at 1.7v

He obviously dosnt run at 1.7volts because he would need phase and his motherboard might just go out anytime. He runs at 1.4volts which could possibly run for 5 years but i acctualy think it wouldn't hit 4 years but you never know. I had a e5200 and it burned out in 5-6months at 1.55+ volts. But i overclock with a little more restriction and spend a little more money and research on to getting a good stepping low voltage processor.

 

[cadaveca]

There is a confusion. You are a confusing person.

Or maybe I'm just confused?

you try to confuse the average user with a complex wording of a simple phenomena.

Um, where's my malicous intent? Could you consider maybe I only have partial understanding of the subject?

while you grasp the concept of molecular structures you're failing to prove your original statement. Instead you are rewording a simple phrase into something much more complicated and I doubt you are even qualified to make that statement.

Of course I'm not qualified!!! Who said I was? I am very interested in the topic however, so maybe I know some stuff, but I never claimed to be an expert? That's part of the discussion...me learning too...who says my uneducated opinion is directly fact? What the hell?

If you are purposefully pushing out the wrong information to entice conversation than you are doing the people of TPU a great disservice.

You are reading too much into my words here, friend. I'm just stating things as I understand them, but make no claims to being 100% right. I mean, thanks for that vote of confidence, but I make no claims to such aspirations. What I meant is I merely keep posting what I think is right...in hopes that those in that do know correct me. How else can I learn, except to post what I think, and expect feedback?

your blatant disregard for being at least semi-accurate or poignant is room for me to consider your responses to be more trollish than helpful.

Wait a second. Now I'm acting like a troll because I don't know everything? WHAT?

I don't like to throw around the term troll, and I hope if you are not then you assume that you are not. The issue I have is not with your source (even though you provided none I assume you may have some sort of background... at least a highschool degree since you use the term Covalent Bond) but why you present your information and what you hope to achieve with your bias which has proven to me and my sources to be harmful. Remaining innocent or helpless in a discussion like this is not what a reader should feel. This subject is not beyond anyone, and it's an issue that can be explained with little to no special vocabulary.

OK, why is it that I must have everything right? I don't get it? Harmful? WHUT? Being wrong is harmful? who the hell are you and "your sources", such that my opinion is harmful? I'm some dude typing away from his basement...


Your posting a PM on the forum is out of line. Your misinterpretation of my PM is obviously the source here, and obviously you took it for me to be claiming to know everything, rather than an admission that I do not. I'm sorry if my choice of words set you off, but please make no mistake, I feel you owe me an apology too...you didn't post this for any other reason but to discredit me...why you feel you need to attack my character, I do not know, but if anyone is acting the troll, it's you.

Uh, why am I even that important? Am I missing something here? The best I can do is post the truth as I know it, which I do. But what's seemingly logical to me, may not have any base in reality...so like I said, I have no problems being wrong so someone else can be right...that way I can learn.

The holes I leave...are things I know I don't know. I'm not gonna seek out every little detail, but I do have a pretty good general understanding of how things are, as you've said. That's why there is no passion or emotion in my posts...because I know there's a good chance I could be wrong, and readily accept that!


I don't know why I gotta be an expert...I merely try to help people. I cannot expect to be right with that help 100% of the time, and yes, I expect the rest of the members that know better than I to correct me when I'm wrong...it's part of being a community.



So, if ya have a bone to pick with me...I don't know why, but please, don't assume too much on my part! And leave the bone in the garbage where it belongs!

EDIT: and I got a solution for you...I'll put it right in my sig that anything I post may be wrong. Silly that I would have to do so, but if it will make my posting more comfortable for "you and your sources", I'm happy to oblige.

 

[sneekypeet]

This thread is not the place to air grievances between members, this BS stops here!

 

[somebody]

I find this thread altogether useless in the fact that it takes away from what we do here.

I don't see that way at all. Personally for me it brings to light the possibility of electromigration effect and although there are disagreements it's up to the reader, as always, to make up their mind as to how they use the information. I didn't read into it that you shouldn't overclock.

I also found it interesting that posters noted having loss of stability over time, even total failure. Since electromigration has to do with high current densities then perhaps looking at the cpu current draw instead of just the voltage would be advantageous.

So thanks for this thread.

 

[cadaveca]

I also found it interesting that posters noted having loss of stability over time, even total failure. Since electromigration has to do with high current densities then perhaps looking at the cpu current draw instead of just the voltage would be advantageous.

I think ALOT of people have seen loss of clocks on Intel chips after pushing them, but not a total loss of stability.

I think 4 of my q6600's, and 5 E8400's showed this, but my QX9650 did not. But the QX ran so hot that I never gave it more than 1.425v.

It only occurs in areas where the trace is too small or becomes small after years of passing electrons.

I guess maybe my Core2's were wore down, or experienced EM when pushed, and hence couldn't handle the same current to push the same frequencies. Makes sense given Binge's comment here.

But at the same time, I have a Maximus Formula X38 board that liked to push CPU FSB volts to 2.0v randomly...I only happened to notice due to increased temps. It could have been this that caused damage.