E8400/8500 degradation myth possibly busted?

[mrcape]

If we believe what was said on this forum in many topics and on other numerous forums if you read around, +1.40v with high frequency is most of the times deadly, +1.35v with high overclocks would degrade over longer periods for many, no degradation reports I ever seen reported at 1.33v maximum for any speed. Probably, vdroop desactivation is playing a big role here with deadly voltage spikes. People should try to overclock at a maximum of 5% within intel specs: 1.36+5% = 1.43v as a max bios set without any vdroop disabled. For most boards, it'll give you a 1.33-1.35 real vcore after vdroop (my little finger is saying this 1.33-1.35v is not by case as it corresponds to most degradations feedbacks). The problem is that setting +1.35v with vdroop disabled, you'll exceed the 1.43v spikes at about thouthand times/sec

Unless we have more feedback or new steppings, that's what I'll do with my chip. If you have money to throw away, than you can of course bump 1.5v into them. Now, if people like living with some dreams that their CPU is not degraded while it needs a heat up period, they are free. But hey, a CPU needing heat up and think it's OS issue is really blindly denying the truth

Finally, the poll is a nice idea if many contribute with real objectiveness

We're on the same page man. Just a little poll data to nail it down and this is good enough for me to feel like I know what's going on 99.9%. And hopefully never need to discuss it again. lol

I think that there are going to be a lot of people dealing with this as more of these come on the market, so thanks to everyone who's chiming in. I bet the threads will help some guys who make the mistake of thinking these are just as durable as 65nm.

[jonny_ftm]

We're on the same page man. Just a little poll data to nail it down and this is good enough for me to feel like I know what's going on 99.9%. And hopefully never need to discuss it again. lol

I think that there are going to be a lot of people dealing with this as more of these come on the market, so thanks to everyone who's chiming in. I bet the threads will help some guys who make the mistake of thinking these are just as durable as 65nm.

It will be even more interesting when the Penryn arrives. It looks like they should be also prone to degradation. People used to overclock 65nm chips will have bad surprises, but indeed, the feedback database of degradation will increase. I don't think we'll be definately fixed before few months though as all this takes time

[PhilDoc]

I really like the idea of a poll and I think you've done a great job of setting it up, but it may have been a better idea to start a new thread in the Intel section. I think it would have gotten more traffic.

Edit: Too bad there's no good way to add VTT, PLL, and FSB into the mix as all could be part of the problem.

[WeldZilla]

I run to E8400's one air cooled on my P5KE-wifi Platform running with G.Skill 6400HZ mem powered by PC Power & Cooling Silencer 750 Quad (Crossfire Edition) It is coming up on 2 months. This processor has NEVER seen less that 1.488V and most of the time it is at 1.544 or higher. It clocks just as high as ever and runs as fast as ever. This machine is air cooled.

My other E8400 resides on my P5E again with G.Skill 6400HZ and powered by the Corsair HX620 psu This system also runs a raid0 array. It runs just as fast today as the day I put it together. I installed the raid0 right out of the box and loaded XP-pro at 500fsb running 4500mhz. Not so much as a hiccup.
I just clocked it up last night and beat my previous 1M pi by .002 and set a new high on PCmark05 so it has not degraded either. This proceesor really has been slapped around with the volts seeing 1.5765 and higher daily and it loves it. It is my little volt whore. I tell it I am gonna turn down the voltage and it say no, no, no.

These systems are so flexible that it is just amazing. I spent the day yesterday running prime 95 in the background well I finished ripping my entire cd collection to my system it never got above 62C all day.
Today I ran all day at 523fsb just to run at 4700+mhz simply cuz I can.
I played some COD 4. Watched Gary Moore concert. Great guitarist. Let it idle for 3 1/2 hours while my son and I went dirt riding and made a dump run.
Then just watched a "Girls Aloud" From Wembly Concert with the neighbor girl.
She wanted to see it on my system. I got to say I really enjoyed it. They are pretty damned good and they are not bad to look at either!
And now I am yakking here and will keep it at this speed the rest of the night. Why not? I idle at 28C at this setting.

If I start to see degradation I will speak up but, I have seen none on either machine.


WZ

[dnottis]

I've seen this "degradation" with all the hardware I've used. Here is the thing. 1. Ambient temps change, this DOES affect stability from the memory getting warmer to the chipset to the PWM on the motherboard. 2. Reseating and applying fresh thermal compound can make a world of difference after a few months. 3. if you are changing or reseating your block or cooler - a bad mount can suddenly make a stable OC unstable. 4. Using AUTO for any voltage could be 1.2v on one boot up and 1.4 another. Again, introducing this phantom degradation.

I have never seen true degradation, but I have seen any and all of the above affect my stable overclocks. Temps rising have caused me to lose stability, I have had stable orthos runs for 24 hours. Then shutdown and power up the next day to have it fail in 30 seconds. I have had this happen with AMD 64 chips, conroe, my quad - you name it. I think 50% of the issue is that it wasn't stable in the first place the other 50% is just one of the issues I mentioned above. My E8400 has actually gotten better for me, using less and less voltage to achieve the same overclocks. The last remount I had was an issue after swapping back in my quad for some last min benchmarks. I was having issues with stability so I pulled the block, re applied my TIM and block and I've been having great luck with my E8400 right now. In fact I'm running orthos at 4.3 Ghz right now.

[mrcape]

I wonder how many of the poll takers noted minimum required voltage before going up over 1.4vcore. That's one thing missing in the poll.

[g1raffe]

Mine has definitely degraded, although I have run ~1,65V several times at 4.5Ghz+ when Benchmarking, and running 1.48V for several weeks.

[dnottis]

CPU: E8400
OC: 4302 Mhz
Stepping: 6 C0
Week: Q748
Code: A219
Vcore: 1.504v
Pack Date 01/29/08

Cooling: Water
Cooler brand/model: D-Tek Fuzion, Thermochill PA120.2

Motherboard: ASUS P5K-E Wifi
Bios Version: 1006
Operating System: XP-32 x86

RAM brand: Mushkin
RAM rated speed: 6400
RAM actual speed: 956
Latency: 5-5-5-15 2T

[Big Lar]

Havn't degraded at all here and I did vote. Tho I don't run high volts, of which are in my sig. Retested today with Orthos Blend and I get the exact same results as when i first got the chip. I do keep detailed results as well.

Larry

[dnottis]

Alright - I'm having issues with what you guys might call degredation. But before we say it's degredation let's take a step back.

Started low and working my way up Orthos never really failed. Got up past 4.3Ghz and 1.5v and it started to fail for me. So I loaded the 4.1Ghz profile in my P5K-E and decided to run Orthos again. Now it's failing in 1 hr.

I have a few issues that lead me to believe that the first run was a fluke.

1. At 4.050Ghz I need 1.40v (actual) to prime 10 hours. Somehow I primed 10 hours at 4.1Ghz at the same voltage, yet 4.2Ghz needs 1.45v (actual). Now when I came home to 4.1 stable @ 1.4v, something didn't seem right, it was too "well I'll be damned" 4050 and 4100 could be stable with the same voltage? Not really likely.

2. Once Orthos starts failing things head downhill. This has occured with alot of my rigs from AMD64 to C2D to my quad. Once orthos starts failing it will fail more often until I shut down and cold start. So I shutdown, hit the PSU rocker and let it sit a few mins. Otherwise if I just keep rebooting, changing settings and hammering it I find Orthos will be more prone to now fail. Shutting everything down, killing the power might reset something elsewhere that is holding power.

3. I have found sometimes the best thing to do is just reset the CMOS and start over. Those pesky AUTO settings are my top reason for believing people experience this "Orthos stable one minute, not the next". Anything set to AUTO could be detected and set differently on the next boot up. And if you don't know what it was when Orthos was fine and you don't set it manually you always face that possibility of it being different. There are settings like CPU PLL voltage that I have on AUTO and I can't find any good documentation for what it does.

Anyways, I'm trying to figure out what is causing Orthos to fail at ~1 hour now on me and I really don't believe in degredation. Not when you are water cooled and the temps are reasonable. I have a few settings in my BIOS set to AUTO and I'm trying to force settings to find which one it might be. Transaction booster, CPU PLL voltage are a few I'm guilty of still having on AUTO.

I just wanted to pipe in with this as I don't believe the chip has degredated at all, it's just one of those things. Being obsessive with my CPUs I have often primed for hours successfully then just because I have nothing else to do, re-run prime the next day or a week later while I'm at work and find that it failed. This has gone back even to my A64's so it's nothing new for me. I think you guys believing in degredation after a short time as a few weeks are over-reacting though.

[STEvil]

gotta agree dnottis, never seen a chip degrade either.

[cadaveca]

Well, I've been playing 45nm on Maximus formula SE. NOt one setting on auto, definate degredation.


But, I noticed a while back that on occasion, FSB volts will be set to 2.0v(instead of 1.2v).

This bios bug has led me to change settings in bios, save and reboot, check settings are ok, reboot again, and verify again. Now, I check bios settings every time I turn the pc on("Discard changes" lets windows load without soft boot).

A few times tho, FSB volts are @ 2.0v again. Not too sure what's up with that...but i think this is the cause for my dying chip.

Anyone using auto settings...well...I must frown upon that...it's NOT PROPER OVERCLOCKING.

For example, @ 333FSB auto settings give 1.65v to NB, even though 1.2v is stable for 333mhz. Um...can you say overkill? If you think you have degredation...and are using auto settings...you've killed the chip yourself, in your lax clocking!

[mrcape]

Yeah I've made it a point from the beginning to keep every voltage setting as low as possible. On the DFI (for me at least), it needs the bare minimum setting for vtt, nb, pll. The only thing I've been leaving on auto are some sub timings and fine clock delay

[dnottis]

Anyone using auto settings...well...I must frown upon that...it's NOT PROPER OVERCLOCKING.

For example, @ 333FSB auto settings give 1.65v to NB, even though 1.2v is stable for 333mhz. Um...can you say overkill? If you think you have degredation...and are using auto settings...you've killed the chip yourself, in your lax clocking!

Thats not true. I had Transaction Booster on AUTO. So when I started having issues I've now disabled it and I'm orthos stable again at 4.1 Ghz @ 1.40v. It wasn't degredation... just a setting as I suspected. Having something like Transaction Booster or Static Read on AUTO isn't doing anything but affecting stability occaisionally. Having voltages on AUTO is a huge no, no.

[cadaveca]

Ok, i'll give you that, however, without knowing what the setting does that you leave on auto, you've left yourself open to all sorts of unknown issues.

I mean really, when confronted with a "new setting", I simply start at stock, and play with it until I see exactly what it does.

Transaction booster, for example...what a stupid name for a memory setting! But I knew what it was from day one...i've been begging for adjustments to this and TRFC for almost a couple of years now(pls feel free to google my user name and TRD/trfc)...when only DFI gave you access to it.

BUt as you say, without knowing what it does...you don't know how you are overclocking. Of course, my statement in my last merely applies to votlages, but clocking blindly on auto in any setting is not a good idea, IMHO.

When I say my cpu has degraded, it's degraded. My fault? I don't think so, but possible. BUt definately a real problem for users with the same config as me.

[dnottis]

Ok, i'll give you that, however, without knowing what the setting does that you leave on auto, you've left yourself open to all sorts of unknown issues.

I mean really, when confronted with a "new setting", I simply start at stock, and play with it until I see exactly what it does.

Transaction booster, for example...what a stupid name for a memory setting! But I knew what it was from day one...i've been begging for adjustments to this and TRFC for almost a couple of years now(pls feel free to google my user name and TRD/trfc)...when only DFI gave you access to it.

BUt as you say, without knowing what it does...you don't know how you are overclocking. Of course, my statement in my last merely applies to votlages, but clocking blindly on auto in any setting is not a good idea, IMHO.

When I say my cpu has degraded, it's degraded. My fault? I don't think so, but possible. BUt definately a real problem for users with the same config as me.

I dont recall you posting specifics - how has it degraded?

[cadaveca]

I was one of the first. Initially I blamed cpu due to how weird it behaves...second is working fine First noticed degredation after a few days, posted on this over a month ago now...maybe 6 weeks.

anyway:

Spyrus, inital 3.6ghz boot voltage(not stable) was 1.1175v. It now requires 1.225 for boot, and 1.275 for stability. So @ 3.6ghz the degredation is far greater. It really makes no sense to me...other than I got a bunk chip. But that happens...in fact, I'm off to buy 2 more e8400's as soon as the local store opens.

I've tested cpu in maximus formula, and MSI p35 Neo2FiR, as well as P5K vanilla. same voltage required on all three boards.

http://www.xtremesystems.org/forums/...postcount=1120

[dnottis]

I was one of the first. Initially I blamed cpu due to how weird it behaves...second is working fine First noticed degredation after a few days, posted on this over a month ago now...maybe 6 weeks.

anyway:

http://www.xtremesystems.org/forums/...postcount=1120

I'm hanging onto my 3.6Ghz B3 (24/7) just in case.

[mrcape]

I also have two that have degraded and have been following this since Jan 21.

The 748 was thrashed beyond help. The 745 needs only .05v extra to pass prime at all of it's clocks, from stock to 4.3ghz. Three or four weeks after sustaining damage, it still needs the same exact amount. No further sliding, no temp sensors out of whackor locked etc., like I had on the 748.

[cadaveca]

I'm hanging onto my 3.6Ghz B3 (24/7) just in case.

I don't think you really need to worry that much...i think my board pushing 2.0v through the chip's FSB link did it. Back then I wasn't aware my board was doing this so much...running prime for 12hrs or so @ 2.0v FSB would more than likely cause some damage, i think.

[mrcape]

My first one was also my mistake with high vcore, PLL & FSB term voltage.

[CrazyNutz]

I don't think you really need to worry that much...i think my board pushing 2.0v through the chip's FSB link did it. Back then I wasn't aware my board was doing this so much...running prime for 12hrs or so @ 2.0v FSB would more than likely cause some damage, i think.

Man 2.0v on the FSB term that's bad news! It's a wonder that it would even
work at all like that since the FSB term is much higher than the vcore. How
did you notice it was at 2.0v? Did you measure the Vtt on the mobo with
a DMM?

[mrcape]

How about a bios designed for 65nm, but able to use 45nm chips? I would assume could be dangerous, unless the bios can identify the chip is 45nm.

Especially the way GTL reference tables play with FSB term. I set gtls manually to 59/59 for best results with my preferred clock. Not sure about leaving gtls at auto and how the bios would set them.

[cadaveca]

Man 2.0v on the FSB term that's bad news! It's a wonder that it would even
work at all like that since the FSB term is much higher than the vcore. How
did you notice it was at 2.0v? Did you measure the Vtt on the mobo with
a DMM?

Yes, DMM, I'm not one to trust bios readings or software. I measure vDIMM from inside the DIMM slot too.

Not too sure why this board does this, but maybe 1 out of 10 boots, vFSB is @ 2.0v setting in bios, reading by DMM 1.93v-1.97v.

[M.Beier]

I havent read the entire thread... But I'm very concerned of this matter, as Im studying and cant afford equipment all the time - and lets just say, I've really really put an effort into making contact with Intel, but without success...

To my knowledge, based on information from OCXtreme, this started back with the 65nm C2D / C2Q, but only at very high voltages... Kevin, OPB (onepagebook) has tested this, and he says its due vPLL, not vCore... Cant remember his figures with 65nm, but 45nm, the one of importance for us - its not recommended to go past 1.7vPLL for dualcores, and 1.6vPLL for quadcores, if one doesnt do this, the vCore shouldnt have any impact...

- Before some of you get suspicous, no, I have not tested this myself, I havent even configurated a 45nm setup yet... 45nm was released very late in Denmark, I only got one a few weeks before Cebit, and directly after Cebit, I went for a marketing competition in France, returned this saturday,... Now my E8400 is up for sale, and Im going for Xeon....
- I intend to give it as high vCore as it will scale with on my watercooling setup....
For benchmarks, I've only done 1.8v - mobo didnt allow higher FSB for us to benefit from higher vCore...