E5200 degrading?

[VirtualLarry]

My friend's rig has an IP35-E and an E5200. I built it for him. When we installed the E5200, we overclocked it to 3.75Ghz, at 1.425v (BIOS). Under load, it drops to 1.4v or a little under. It wasn't stable at any faster speeds with only that voltage, and I didn't want to go higher on voltage because I had heard that 1.4v was the safe max for 45nm chips.

Well, fast-forward to today. The guy from the electric company came to shut off his power, and actually did, for a few minutes, until my friend negotiated with him. Anyways, I was using his rig a little later in the day, and I noticed that it was no longer overclocked. So I restarted, went into BIOS, settings were still there, saved and exited, and went back into Windows XP. CoreTemp and CPU-Z then showed that it was overclocked. I had wanted to check the temps, so I ran Prime95 25.9 small FFTs. Within less than a minute, one of the cores errored out, at settings that had previously been stable when we originally set it up. I ran Prime95 again, thinking it was a fluke. Again, an error, this time on the other core.

So I went back into BIOS, and rather than crank up the voltage, I cranked the speed down. From 300FSB to 290FSB, 3.625Ghz.

This time, Prime95 seemed stable, at least for the few minutes that I tested it, but temps were horrible, 83C on both cores, with a TJmax of 100. So again, off the computer went, this time to get its insides blown out with some compressed air. I did that, and temps under Prime95 small FFT dropped to 67C on both cores. With a HyperTX2 cooler.

So tell me, is this an issue of the CPU degrading? Or the mobo degrading? Vcore under load reported the same values that it always has in CPU-Z, so it doesn't seem like the mobo is degrading. I don't think the dual-core draws that much power. So that leaves the CPU voltage, and lifespan of the chip. Does this mean that even 1.4v is NOT SAFE for 45nm dual-cores? If that's true, then a lot of people are going to be surprised, I think.

 

[swanysto]

I think Intel said that 1.36 was about the max safe voltage, but I could be wrong. Although what you have described is what happened to an older E4500 I had back in the day. I went up a little past 1.4 to see if I could get it stable, but it failed, and it never really ran the same again. It ran fine at stock after that but I would have problems at anything over 2.6(stock was 2.2).

Not sure if this is the case with yours, but it's possible.

 

[jjmIII]

I never felt either of my IP35E's held clocks well. I call mobo.

 

[Bill Brasky]

I've never heard of a power surge 'slowly' killing electronics, but maybe the psu or some power regulators were damaged when electricity went on/off. If the processor isn't getting the same quality of power is was before, that would explain why the overclocks failed but not the temps issue. Do you guys have any spare parts to swap out?

I'm guessing it's the PSU or mobo, but it's hard to tell.

 

[MyLeftNut]

I would say that it's the motherboard. It happened to me with my Asus P5B Deluxe and E6400. Once I switched to a new motherboard, the overclocks came back.

 

[E4300]

Reseat the CPU. Check for problem. I run an E4300 @ 1.46 over one year (EIST/C1E enabled) without issue. 1.40V is too much for most 45nm chips. 1.36 is the official Intel spec. The chip has probably degraded due to the high voltage. A bad decision for that extra 150MHz.

Most of these E5xxx chips top out at 3.4-3.5GHz with 1.36V in BIOS. A great chip can break 4.0GHz.

 

[Gillbot]

Originally posted by: jjmIII
I never felt either of my IP35E's held clocks well. I call mobo.

this ^

Unless you have another system to test the CPU in to verify, because i've never seen a CPU degrade. 99% of the time it's ram or mobo.

 

[Schmide]

From another thread. 4-phase motherboard degrading. I know it's another motherboard but meh

Gigabyte P35-DS3L, will it OC a Q9550?

Direct link to image

 

[nerp]

It's most likely the mobo. I have the same abit. It is a solid board at stock but I similarly lost my ability to OC over time. It's part of the reason I gave up overclocking (don't game or do anything CPU intensive really) and swapped it out for a rock solid intel board with no frills. Stability trumps 10% extra speed, IMHO.

 

[Kenmitch]

Originally posted by: E4300
Reseat the CPU. Check for problem. I run an E4300 @ 1.46 over one year (EIST/C1E enabled) without issue. 1.40V is too much for most 45nm chips. 1.36 is the official Intel spec. The chip has probably degraded due to the high voltage. A bad decision for that extra 150MHz.

Most of these E5xxx chips top out at 3.4-3.5GHz with 1.36V in BIOS. A great chip can break 4.0GHz.

I'd agree with this statement. The xtra mhz not even gonna be felt anyways. Most likely the chip degraded due to voltage and heat as you stated the heatsink was dusty. Who knows how hot she was running and for how long before the power being turned off.

I guess unless your buddy kept an eye on the temps all the time the chip could have even been run above the Tjmax temp as you stated the heatsink was dusty which would explain the drop in core temp when you blew it out.

It is possible that the motherboard could have been damaged also with the OC but unless you take the chip out and test it in another board we'll never know.

On another note I have one of the great chips e5200 that will do 4.16 @ 1.348 vcore but I chose to run her at 3.33ghz @ 1.108 vcore so I get a boost without any xtra heat as she runs cooler overclocked and undervolted than stock speed and vcore

 

[VirtualLarry]

Originally posted by: Kenmitch
On another note I have one of the great chips e5200 that will do 4.16 @ 1.348 vcore but I chose to run her at 3.33ghz @ 1.108 vcore so I get a boost without any xtra heat as she runs cooler overclocked and undervolted than stock speed and vcore

Prime stable? That's amazing.

Btw, if the mobo were degrading, wouldn't vcore droop more under load than it had been? That doesn't seem to be happening.

 

[Kenmitch]

Originally posted by: VirtualLarry

Originally posted by: Kenmitch
On another note I have one of the great chips e5200 that will do 4.16 @ 1.348 vcore but I chose to run her at 3.33ghz @ 1.108 vcore so I get a boost without any xtra heat as she runs cooler overclocked and undervolted than stock speed and vcore

Prime stable? That's amazing.

Btw, if the mobo were degrading, wouldn't vcore droop more under load than it had been? That doesn't seem to be happening.

100% stable. I was running prime on her and forgot all about it.....Needless to say after 27 plus hours she was still humming along so what did I do....Slammed Intel Burn Test on top of the prime at max mem available for 5 runs of 5 which all passed so I'll call her stable

Could be problem with NB or SB maybe....Maybe compound is cooked and not doing it's job anymore. Wouldn't hurt to check it out I guess.

 

[E4300]

IP35-E can easily put out 120W with active cooling at NB and PWM. That said, 45nm E5xxx shouldn't draw more than 95W, even at 1.4Vcore. Good air flow to NB and PWM is a must for any type of high Vcore overclocking.

The IP35-E/E5200 combo sold last year is still humming at 4.1GHz with 1.36Vcore. Peak prime temp is 52C @ 84F ambient.

Current IP35-E/E7400 combo is still active at 4.03GHz with 1.29Vcore. Peak Prime load temp is 54C at 86F ambient. Always enable C1E/EIST to extend CPU life.

Oxidation/contact resistance at the CPU pins can occur if the CPU is exposed to high temp (+70C) over longer period of time. Reseating the CPU may alleviate this problem. Inspect the back side of the NB and PWM for yellowish discoloration. This is a sign that these regions were subjected to excessive heat.

IP35-E was designed with a robust output section that should take a decent 45nm quad up to 3.7GHz (1.3Vcore). Unfortunately, the NB and PWM heat sinks did not receive the same treatment. Therefore, one must provide active cooling to these regions when applying a large dose of Vcore to the CPU.

 

[Lemon law]

The other thing to point out is that Kenmitch probably has that 1 in a thousand really great chip. And we might be able to say its mere luck, but then again, I have followed his threads and Kenmitch as a careful and responsible overclocker has done much to make his own luck. And has certainly shown the wisdom to blend speed with stability and safety.

 

[Kenmitch]

Originally posted by: Lemon law
The other thing to point out is that Kenmitch probably has that 1 in a thousand really great chip. And we might be able to say its mere luck, but then again, I have followed his threads and Kenmitch as a careful and responsible overclocker has done much to make his own luck. And has certainly shown the wisdom to blend speed with stability and safety.

Thanks

Guess mine would be a e5200s if intel made such a chip.

I guess I'd chalk mine up to intels need to fullfill the quota on e5200's and a great chunk of silicone. I'm sure many more chips are humming at stock speed in OEM systems without the chance of running to the true potential of the chip.

Back to the subject at hand....If it was me I'd remove the NB and SB heatsinks and replace the thermal compound and maybe try the 333fsb and drop multiplier. When I was doing my testing it seemed like the fsb didn't change the chips need for vcore by much and figured running on the 333fsb would have less stress and most likely heat to the other components in my system.

 

[error8]

I had the same thing happened to me, on my MSI P35-Neo 2 FR and an E7200. It ran for months at 4 ghz at 1.46 V (ups) then it started to degrade, until I had issues keeping it stable at 3.8 ghz at over 1.4 V. It might have been the voltage, it might have been the cpu, but surely, then motherboard wasn't the culprit. It was kept as cool as possible. I do believe that these Intel 45 nm chips, have a degrading issue after all, although some might disagree.

 

[Kenmitch]

Originally posted by: error8
I had the same thing happened to me, on my MSI P35-Neo 2 FR and an E7200. It ran for months at 4 ghz at 1.46 V (ups) then it started to degrade, until I had issues keeping it stable at 3.8 ghz at over 1.4 V. It might have been the voltage, it might have been the cpu, but surely, then motherboard wasn't the culprit. It was kept as cool as possible. I do believe that these Intel 45 nm chips, have a degrading issue after all, although some might disagree.

I'd be pointing the finger at the vcore in your case :Q

You should have just admitted you didn't get " The Chip " and took a comfertable OC and called it a day.

 

[Comdrpopnfresh]

If it were the mobo fading, you might not have more vdroop, but could have more ripple. From my own experiences, the cpu doesn't degrade- the other components it is reliant on do. Unless you've been blasting it with voltage for everyday usage. It could be a few things, or a combo. Maybe the computer has been moved/jostled, and the heatsink isn't fitting as snug or the TIM has oxidized. If the mobo doesn't have solid caps, you could simply be getting poorer + more unsteady voltage to the cpu. Same for the power supply- the caps could be deteriorating- from excessive draw or heat over time. The best way to know is to place the processor in another system that is comparable/better than your friend's initially was. I've had to ease chips into an overclock too- changing at steady increments and adjusting voltage accordingly, even when I knew the processor was capable.
As some people mentioned- the powerline could be unclean- a ups w/ line leveling/cleaning might help. But I'd say that's the least likely- unless the computer is on the same branch as a microwave or appliance w/ a compressor and is ongoingly getting brownouts

 

[VirtualLarry]

The power in my friend's house isn't that good. Lots of dead PSUs and dead light bulbs in that house. One of the guys living there seems to think that the voltage is around 130v.

 

[Schmide]

US AC voltage at 130v is great. A PSU would be super efficient at that level. It gets crappy when it drops below 110v

 

[Idontcare]

Originally posted by: VirtualLarry
The power in my friend's house isn't that good. Lots of dead PSUs and dead light bulbs in that house. One of the guys living there seems to think that the voltage is around 130v.

I can fathom myriads of ways the voltage ends up under 120V, line sag is line sag, but how does one go about accidentally ending up with higher voltage?

The source would have to be the step-down transformer outside the house, and since it operates on a hard-wired step-down ratio it would be rather odd for that ratio to be off. So maybe the line feeding the transformer is hotter than it is supposed to be? (not impossible, but usually they regulate those things don't they?)

But yeah, as Schmide is saying, high(er) voltage is better (to a point) for household appliances versus lower voltage. Lower voltage means the device ends up pulling more amps to operate at the same power-consumption, and the amps are what generate heat and result in things burning up.

Only reason our houses use low(er) voltage than the lines crossing the city is to reduce the chances of us electrocuting ourselves in our homes (because just enough of us are idiots that we'd find a way to fry ourselves). Europe and Japan, among others, use higher voltage for a good reason. And some of our appliances (electric clothes dryers for one) simply pull too many amps at 120V so they have to run higher voltage anyways.

 

[LCD123]

3.75GHz is a huge 50% OC from 2.5GHz stock. I read somewhere that those CPUs are Intel's bottom speed bins and most won't do above 3.5GHz at 1.4v If you need over 4GHz, get an e8600, they do up to 4.5GHz which is a 50% OC from 3GHz!

 

[MyLeftNut]

Speaking of UPS's, how well do they actually regulate the power going to the PC? I was stress testing my i7 920 with prime yesterday when some idiot flipped the breaker switch. The UPS didn't hold even 2 secs for me to stop it and shut down properly. Instead, the system just powered off completely. The UPS has a 600W capacity whereas my PSU is 560W.