E8400/8500 degradation myth possibly busted?

[RunawayPrisoner]

Intel doesn't really confirm it. It's just that on their specs sheet, 1.10v is stated as the maximum. Just like how 1.45v would be max vCore (but then the recommended max vCore is 1.3625v anyway). If I understand that they are talking about BIOS settings or paper values, then a difference of 0.250v max (in real value) is expected, and then the range should be 1.34 - 1.39v recommended and 1.42 - 1.48v depending on the motherboard. That's true since there's no way to exactly engineer a motherboard that can transfer exact vCore or any other voltage values like that.

What I'm saying is... VTT is probably the only thing we are running that is out of Intel's specs, and with that, there should be risks for those who are taking VTT too high as well.

[mrcape]

I know, I'm aware of that spec sheet. I was referring to the anand report, I thought you might be too. Intel told them directly that high VTT killed their chip.

On my second chip, VTT has never crossed 1.20v. Still had weirdo symptoms though.

[RunawayPrisoner]

Oh... yeah. I didn't read that article too thoroughly. And that reminds me... I am limited to 1.50v (maybe 1.52v real) on this board so maybe that's why my extreme clocks haven't killed my chips yet.

[RealTelstar]

I'm planning to keep my new e3110 to <= 1.30

[Infa]

And I'm not

I like what crazy says

Specially when i started to use My premium X48 board, the VTT are very sensetive and wont boot the board if its just slightly to high.

keeping it 1.20, maybe i can go lower, will try.

Me to have this symptom, first bootup i was prime stable 4Ghz 1.26
4,3 rock stable 1.45, now its just about 0.02 higher that needs. strange though ... maybe this unaware of VTT before, many "degraded" their chips slightly, i didnt know this, before it was said not above 1.50. .. its only on x48 1.10 are STOCK i think.

ahh well cheers!

[CrazyNutz]

On my cpu/mobo combo I have not needed to go beyond the standard 1.10v
for VTT, even with FSB @ 500mhz, stable! However I very well doubt having
it set to 1.20, or 1.30v would cause any damage. The spec document says
1.10v +/- 5% is the norm, however 1.45v is the absolute max, just as it says
for vcore.


The burn-in is complete on this CPU, and it is solid!

[RunawayPrisoner]

Well, just be careful. Auto settings on some boards really do set weird VTT when you're not at stock. Auto doesn't mean it'll stay at one value forever. The only real way to know is to get a multimeter and measure real VTT.

[EXPANDER]

Tests Systems

E8400 Retail 1.64 Vcore
DFI P35 UT T2R
Silverstone Op 1000 WATT Psu
2X 512 MB DDR2 800 MHZ RAM
SINGLE STAGE -57C IDLE BLOK

E8400 1.64 Vcore 5244 MHZ FSB: 582

Validate: http://valid.x86-secret.com/show_oc.php?id=346970



SP1M No Tweak

[slugzkea]

How about a prime result? Super PI doesn't proove any stability sorry to say.

[jaredpace]

i required no burn in for my chip. 4.2ghz 24/7. E8400. Took some SS's of priming and messing around at 4.5ghz 500 x 9. I don't see any degradation. kept my Vcore under 1.42v @ 4.5g
1.36 @ 4.2

[Monstru]

Wolfdale E8400

Q746A476
VID 1.125
Pack date: 07-01-2008
Period of usage until now - 3 months


Maximum voltage fed to the CPU:

vcore air - 1.75v
vcore LN2 - 2.05v
VTT - 1.55v
vPLL - 1.95v


Used for:

A number of air, dry ice and LN2 cooled sessions, for SuperPi 1M and 3D benchmarks, many hours each


Mobos:

-Asus Commando
-Asus Maximus Formula
-Asus Blitz Extreme
-Dfi Infinity P965


Initial behavior:

Orthos - 4000Mhz @ 1.28v (Maximus)


Behavior after high volts:

Orthos - 4000Mhz @ 1.33v (Commando)


Conclusion:

After applying extreme voltages (sometimes almost double than stock vcore)for a long period of time, going through temperatures between -150 and 65 degrees Celsius, the voltage needed to be Orthos stable at 4Ghz increased with 0.05v. This affects daily use, but until now I haven't noticed any changes in the behavior in extreme conditions, the CPU is still fit to run 3DMark at 5650Mhz with 1.9v at ~ -92/-100 degrees. The conclusion is logical and simple...If you care about 0.05v in daily use behavior, do not use the same CPU for extreme purposes. But if you bought the CPU for high end benchmarking, you shouldn't worry about this. This was always an issue, similar degradation happened to a lot of other CPU's that worked with extremely high volts in extreme conditions. We're not talking about something new, just that in the case of 45nm CPU's it happens earlier and it is easier to notice.


A few observations:

-vcore was measured with a voltmeter
-on Asus Commando at least, the voltage measured by in Hardware Monitor in bios (1605), the voltage shown by CpuZ 1.44.2 and the voltage measured with the meter are identical.

[CrazyNutz]

Wolfdale E8400

Q746A476
VID 1.125
Pack date: 07-01-2008
Period of usage until now - 3 months


Maximum voltage fed to the CPU:

vcore air - 1.75v
vcore LN2 - 2.05v
VTT - 1.55v
vPLL - 1.95v


Used for:

A number of air, dry ice and LN2 cooled sessions, for SuperPi 1M and 3D benchmarks, many hours each


Mobos:

-Asus Commando
-Asus Maximus Formula
-Asus Blitz Extreme
-Dfi Infinity P965


Initial behavior:

Orthos - 4000Mhz @ 1.28v (Maximus)


Behavior after high volts:

Orthos - 4000Mhz @ 1.33v (Commando)


Conclusion:

After applying extreme voltages (sometimes almost double than stock vcore)for a long period of time, going through temperatures between -150 and 65 degrees Celsius, the voltage needed to be Orthos stable at 4Ghz increased with 0.05v. This affects daily use, but until now I haven't noticed any changes in the behavior in extreme conditions, the CPU is still fit to run 3DMark at 5650Mhz with 1.9v at ~ -92/-100 degrees. The conclusion is logical and simple...If you care about 0.05v in daily use behavior, do not use the same CPU for extreme purposes. But if you bought the CPU for high end benchmarking, you shouldn't worry about this. This was always an issue, similar degradation happened to a lot of other CPU's that worked with extremely high volts in extreme conditions. We're not talking about something new, just that in the case of 45nm CPU's it happens earlier and it is easier to notice.


A few observations:

-vcore was measured with a voltmeter
-on Asus Commando at least, the voltage measured by in Hardware Monitor in bios (1605), the voltage shown by CpuZ 1.44.2 and the voltage measured with the meter are identical.

These CPUs will require more vcore than the initial, even without applying extreme vcore. For example:

I had mine initially @ 1.32vcore @ 3960mhz 8 hours prime stable
After two weeks it needed 1.352vcore for 3960mhz to be prime stable

However, I say it's a burn-in period rather than degradation, a conditioning
of the cores if you will, that after a period of time results in needing a small
vcore bump. After I/We figured this out, I now run near extreme (I just have
water cooling) vcore (i.e. 1.5v -1.6v), without degrading any further,
and I know it's not because I've been hammering it now with p95 for
about a month now, It is just rock solid

It's not all that great that it needs a small bump in vcore, but im still very
happy with it especially since it will not slip any further.

[mrcape]

Isn't it interesting that all three of us found we need exactly .05v? That's just enough info for me to feel we've pinned this.

Whether we call it burn in or slight degradation doesn't matter to me.

Thank you guys a million for posting your findings!

[fireice2]

Three similar findings from 3 different sources. Nice find guys!

[CrazyNutz]

Isn't it interesting that all three of us found we need exactly .05v? That's just enough info for me to feel we've pinned this.

Whether we call it burn in or slight degradation doesn't matter to me.

Thank you guys a million for posting your findings!

Yes it is. And I've read some articles about the burn-in theory, almost all state that cpu burn-in can make the cpu more stable at lower voltage,
and it can also do just the opposite, which seems to be the case with most
of our 45nm cpus. Anyhow 0.05mv is not that much so I say BFD! Im totally
ok with that. Im just happy that I can go about my business and pump some
hefty vcore in these things for some bad arse overclocking, and not
have to worry about crippling it any time soon

[Monstru]

That;s ok, but remember that good cooling and inteligent overclocking is the key. If you keep it at 1.5 for daily use I don't think it will last long

[CrazyNutz]

That;s ok, but remember that good cooling and inteligent overclocking is the key. If you keep it at 1.5 for daily use I don't think it will last long

Oh, I agree, I keep it @ 1.35 for daily use, I just crank it up sometimes to
do some benches, and it's on high-end water so temps are not much of a
issue for me.

[tranceaddict]

Update:

Initially I though I might have degraded the CPU by benching @1.43 for an hour or so. I was Prime stable at 4200MHz @ only 1.33V, and when I re-tested Prime stability after the benching session (roughly 2 weeks later) it took 1.35V to be stable. Naturally I assumed that the 1 hour benching session caused it to degrade. Well, it has been close to 2 months since I even touched Prime or my BIOS, and the Vcore has been at a constant 1.35V with load temps in the high 40s, low 50s. I ran Prime today and it is not stable at 4200MHz even at 1.37V. What is up with that?? I am really confused, why does my stability continue to decline? No variables changed in the last 2 months and an Overclock which was Prime stable for over 16hours now fails in between 10-20minutes!!

[CrazyNutz]

Update:

Initially I though I might have degraded the CPU by benching @1.43 for an hour or so. I was Prime stable at 4200MHz @ only 1.33V, and when I re-tested Prime stability after the benching session (roughly 2 weeks later) it took 1.35V to be stable. Naturally I assumed that the 1 hour benching session caused it to degrade. Well, it has been close to 2 months since I even touched Prime or my BIOS, and the Vcore has been at a constant 1.35V with load temps in the high 40s, low 50s. I ran Prime today and it is not stable at 4200MHz even at 1.37V. What is up with that?? I am really confused, why does my stability continue to decline? No variables changed in the last 2 months and an Overclock which was Prime stable for over 16hours now fails in between 10-20minutes!!

You prolly need to go with an extra 0.05mv over your initial 1.33, which puts
your new vcore requirements at 1.38. I think it will make it stable at that vcore, and it should not decline any further, based off my findings.

[TL1000S]

Nice thread/initiativ CrazyNut

I was a bit frustrated over my 1.gen E8400 after initially running stable 24/7 4.0Ghz (my goal) in a P5N-E SLI.
This required 1.47 (bios) which is 1.37 fully loaded OCCT with this "simple" motherboard.

Then suddenly later OCCT-session would fail (instant reboot) at previously verified settings (I have tried several combos clocks vs. volts to find "the golden middle")

Now I see I must feed my E8400 1.42 fully loaded for stable 4.0Ghz (which means 1.52 idle/bios).
And there it will be set.
If it "dies" it is a good exuse to buy a newer and hopefully better chip.

[tranceaddict]

You prolly need to go with an extra 0.05mv over your initial 1.33, which puts
your new vcore requirements at 1.38. I think it will make it stable at that vcore, and it should not decline any further, based off my findings.

It seems that you are right, I bumped it up a notch to (1.376V IDLE 1.368V LOAD) and I just passed two sessions of around 12 hours each. Hopefully it will not decline any further. Do you think that this voltage is acceptable for 24/7 use? My temps are fine and the load voltage it a tad below 1.37V.

P.S. Did you switch over to BIOS 1012 on your P5K-E or are you still on 1006? I am wondering if I should flash over to the newest BIOS?

P.S.S. Now that think of it, you were also right that these chips did not degrade due to voltage but rather a sort of break in period. I got it stable at 1.33V initially and 2 weeks later it needed 1.35V. I attributed that increase to a brief 45min or so session of benching at 1.43V. However, the chip probably would have lost that stabilty whether I ran that 1.43V short session or not. Since then stabilty decreased to where I need 1.376V and the voltage was never touched above 1.35V. Moreover, on other forums I remember comming across posts were people had their chips at 1.3V or so and never higher, and a little later then needed more volts to get the stabilty they once had.

[tranceaddict]

Isn't it interesting that all three of us found we need exactly .05v? That's just enough info for me to feel we've pinned this.

Whether we call it burn in or slight degradation doesn't matter to me.

Thank you guys a million for posting your findings!

Make that 4 guys!

[CrazyNutz]

It seems that you are right, I bumped it up a notch to (1.376V IDLE 1.368V LOAD) and I just passed two sessions of around 12 hours each. Hopefully it will not decline any further. Do you think that this voltage is acceptable for 24/7 use? My temps are fine and the load voltage it a tad below 1.37V..

Excellent, I'm glad to hear you are stable again As for the voltage I
personally think it's fine for 24/7 use, as long as your temps are safe.
I have been running my vcore @ 1.37v and freq @ 4050mhz for about
a week, and it's just fine, I also have had it in the 1.45-1.50v range to
beat it up with some benches(just a few hours), and It has not done
any damage.

P.S. Did you switch over to BIOS 1012 on your P5K-E or are you still on 1006? I am wondering if I should flash over to the newest BIOS?
..

Yes I did, and I like it. I had that issue where sometimes the pcie 16x slot
would get stuck at 1x, and after flashing 1012 I have not had it happen
once, so maybe that is actually fixed. I do recommend it.

P.S.S. Now that think of it, you were also right that these chips did not degrade due to voltage but rather a sort of break in period. I got it stable at 1.33V initially and 2 weeks later it needed 1.35V. I attributed that increase to a brief 45min or so session of benching at 1.43V. However, the chip probably would have lost that stabilty whether I ran that 1.43V short session or not. Since then stabilty decreased to where I need 1.376V and the voltage was never touched above 1.35V. Moreover, on other forums I remember comming across posts were people had their chips at 1.3V or so and never higher, and a little later then needed more volts to get the stabilty they once had.

Yep, It does seem very apparent. The second e8400 I bought I kept it @
1.32v and It eventually needed more volts, so it's obvious voltage was not
causing this. The good news here is once it completes burn-in it slips
no further.

[chris.y2k.r1]

Capacitors.... Cheap Chinese capacitors. I purchase GOOD 1000Mf 250V replacements for some of my PSU units. Most of them have 2 large capacitors that affect the performance the most. On larger wattage PSUs, the capacitors are usually rated fairly high so 1000Mf is a good replacement but certainly not a 1 size fits all. The over all difference in my CPU performance and system stability is night and day. If you are so inclined, go to digikey and replace every capacitor you find (in your PSU) with a decent one. As a rule of thumb, for replacements, same voltage (Meaning if it's a 250v - get 250v) but you can go roughly 50% either way (+/-) with the Mf rating.

I simply do this as a rule of thumb anymore. Obviously, it voids my warranty but I've found that you won't need to worry about it crapping out if you do it anyway. Besides, I've sent one back even after I replaced them because it had so many crappy parts (Bad Diodes, Capacitors, Resistors... etc.,) and they didn't notice. In reality, what I gave them back was still better than the crap they gave me. It had way better parts

[pentium777]

Count me in whether people want to call it degradation or just not stable for the first week or two weeks either way I'm in the club.

I never exceeded 1.45v (bios) 1.40 (CPU-Z&Everest) I was able to do 4Ghz with 1.3v then 4.4Ghz with 1.45v stable (Small FFT for hours) after that I tweaked my ram etc and backed down to 1.41v @ 4.3Ghz, few days later I found I needed to back down to 4.28Ghz, then 4.25Ghz. Now I am failing again. Also CPU is running hotter than it used to.

Now I'm going to be testing 4.2Ghz @ 1.4v hopefully as temps are 47C idle, 77C load! too hot... However my laptop with Core2Duo 2.4Ghz T7700 stock will run 80C using Orthos and it is 55C idle, so they must be designed for these temps.