overclock damage?

[jaga]

Hi,

I would like your ideas about what hardware is damaged - my cpu or motherboard? I overclocked my e2140 100% - corrupted vista 64. Turned it back to stock speeds, and even though it still booted, I was getting BSODs. Now I reinstalled and I still get a BSOD every couple of days. I have a ASUS P5K and OCZ 800mhz platinum rev.2 RAM.
I plan to change the processor for a e6850 or quad to get the power I need.
Any ideas?

Thanks

 

[MarcVenice]

Bsods are often ram related. Run memtest to check the ram ...

 

[jaga]

The RAM was good at fsb 333 - tested both with memtest +86 and vista memtest tool. didn't test at 400 - not that I remember anyway.

Thanks

 

[taltamir]

use memtest 86+ from http://www.memtest.org/
You should have 0 errors.

Reinstall windows. Test CPU for errors...

 

[soccerballtux]

Originally posted by: taltamir
use memtest 86+ from http://www.memtest.org/
You should have 0 errors.

Reinstall windows. Test CPU for errors...

Hi I'm neffing and psoting without reading ops.

 

[jaga]

how to test cpu for errors? Could the bios be corrupted. All the volts are set correctly for the RAM and the cpu volt is set at 1.325. I don't have a case fan.

 

[GundamF91]

you could've corrupted Vista. If you can boot into it, try to repair the OS.

 

[taltamir]

Is the repair function reliable? I Wold only trust a full clean install as far as a repair goes...

Anyways, you should probably take off the heatsink and clean and reapply paste and see if there is any burnt parts there... (flakes of thermal paste and dark spots, etc). Also look for cracking or flaking of the actual chip itself. If the chip seems to have sustained physical damage then you probably need to replace it. If only the thermal paste got fried then simply clean with isopropyl alcohol 90+% and a thick paper towel and reapply with arctic silver 5. Ofcourse, maybe its completely clean and unharmed. You can't tell unless you look.

As for testing it... I unfortunately don't know any program that tests for correctness... I wish I did.

PS. I noticed you said already that you reinstalled windows, but what I meant was that once you track down the problem (and possibly during) reinstall again... if you had memory errors or CPU errors during installation it sometimes doesn't BSOD, but instead leaves you with corrupted files that will cause lots of problems down the road... so reinstalling during and after the diagnosis process is recommended.

 

[jaga]

I spoke to microsoft support and they point to hardware. They suggest i should check the following: hard drive - bad sectors, cpu, RAM, hard drive controllers.I will check the CPU for signs of heat damage. I long stress test should also help. Thanks.

 

[jaga]

But what I want to know is what to replace. I overclocked the fsb - so all the components on the motherboard were also overclocked. Weren't they? That is the hd controller, the pci cards, the chipset, cpu, cache etc (see http://www.pcguide.com/opt/oc/risksRisksBus-c.html). Is this the case with the P5K - which has overclocking in mind - ? I noticed the pci-e clock went up to 150 - i didn't set it to this. Please help!

 

[BonzaiDuck]

I went through these troubles almost two years ago. I could run MEMTEST86 error-free at stock settings, but even half my previous over-clock would generate errors. I was candid with the memory-manufacturer, and they were indulgent in offering to replace my modules under the lifetime warranty, assured that I hadn't overvolted them.

The new modules did the same thing.

I replaced the motherboard, but found it too convenient at the current prices to replace the processor for more performance, and without bothering to test it. But it still seems more likely to me that I'd blown the memory controller on the motherboard.

It's always tricky isolating a failed component, and it often helps to have spare parts. so learning from that experience, my over-clocking barely stresses my motherboard beyond its spec; my memory is run within spec, and the only likely risk is isolated to the processor.

Even that -- is not much beyond spec -- as opposed to how I treated my motherboard as advent to the panic of two years ago.

 

[jaga]

Thanks for the Bonzai - It isn't clear whether the motherboard was stressed or not. I believe the P5K can go to 533mhz or there abouts. I don't think the memory was stressed - it was strapped to 800 mhz (stock). Having said that, at 333mhz, vista gave it a higher score, indicating that either the whole system was running faster, as hinted at earlier, or that the overclocked processor (333*8) boosted the memory performance. Which is correct? Of course the graphics card could be damaged too, but i don't think this is the problem.

 

[Denithor]

Corrupted OS is generally related to either:

1) Bad/failed memory or memory socket (good info goes, in, bad comes out, and OS writes that bad data onto the hard drive and viola! corrupted system files).

2) Bad sectors on hard drive (self explanatory).

The problem with a corrupted OS is that you really have to do at the minimum a repair install to fix the corruption, and that should only be done after you have fixed the root cause of the problem. If bad memory caused the problem and you try to repair the installation, the corruption problem is probably just going to get worse as more data will be passing through the bad memory and getting corrupted. I have seen this happen several times during very first OS setup on newly assembled computer system, generally either the install won't complete or if it does there are lots of stability problems and random crashes.

I would be highly surprised if your cpu was damaged by the OC or was in any way related to this corrupted OS problem.

As mentioned before, run the memtest86 program (burn a bootable CD from the .iso you can download from their site) to check for memory errors first. If that comes back clean use the disk tools provided by your HDD manufacturer to check for drive errors or bad sectors (you can also use the chkdsk function in Windows for this test).

 

[myocardia]

Originally posted by: jaga
Thanks for the Bonzai - It isn't clear whether the motherboard was stressed or not. I believe the P5K can go to 533mhz or there abouts. I don't think the memory was stressed - it was strapped to 800 mhz (stock).

If you had your RAM @ 800 Mhz, with a 200 Mhz FSB, that's using the 4x memory multiplier. So, change your FSB to 333, and your RAM is running @ 333x4= 1333 Mhz. If you also had your vdimm set to Auto, you've fried your RAM.

 

[jhick]

I havn't overclocked my computer yet, but this makes me glad I will be following the guides outlined in the sticky

 

[jaga]

Hi Myocardia. Thank you. Can you explain why leaving the vdimm to auto is bad - in the case you gave? I remember setting the fsb strap to whatever was nec. to leave the ram at 800mhz. The asus has a nice control for this. The first time i tried 400, and it didn't boot because the vdimm was set too low. I then set it to 2.1 v (max) and I got these nasty vista errors. Then 333 fsb was tried, again the ram was at 800 using the strap control, and the 2.1v, but by then I think the damage was done. I have tested the ram (5 cycles of memtest +86 - test 5) and tested the drive. The machine is still BSODing. Now I have reset the bios to its defaults and so far so good.

 

[Regalk]

Couple of things newbie overclockers should need to know
In the old days you had to set your PCI bus to 33 and AGP bus to 66 or close to that. The PCI bus was critical since that controlled ATA controller i.e.HDs CDroms, pci cards. AGP of course controlled Video.
Voltage was not too much of a problem since auto did not do much other than default vcore - you had to either mod or manually set in bios or through a dipswitch.

Currently most Mobos (from tier 1 manufacturers) not sure about (Foxconn, ECS and that lot) locks the PCI at 33 even though some still have this in their bios (Asus's A8N-E comes to mind). They also have the PCI-e which can be set from 100 to 150. 100 is the recommended speeds or 101 (over 110 is not recommended). I have read somewhere that the PCI-e has some kind of link to the PCI bus but I doubt it. Also some motherboards have the SATA linked with the PCI-e bus or independent e.g. JMicron controllers.
AUTO vcore these days is actually dangerous since as the OC speed rises the voltage rises as well. Ditto for DDR volts. Therfore you could actually fry your cpu or DDRs. I really do not understand why manufacturers did this. The problem is twofold - both the speeds increase and voltages at the same time well beyond the specs! If you did not test those limits and went too far beyond the specs and your cooling is stock or average = bad.
SO WHEN OVERCLOCKING YOU DO NOT WANT TO OVERCLOCK ANYTHING OTHER THAN CPU and DDR(2) TO A LIMITED EXTENT within an acceptable range of voltages (for DDR2s you should not go beyond 2.2v definitely not AUTO)

Bottom line is unless you know your bios and have done your research DO NOT OVERCLOCK! It took some of us years to get to this point but some you want to do this in 2 days! A disaster waiting to happen.

 

[myocardia]

Originally posted by: jaga
Hi Myocardia. Thank you. Can you explain why leaving the vdimm to auto is bad - in the case you gave? I remember setting the fsb strap to whatever was nec. to leave the ram at 800mhz. The asus has a nice control for this. The first time i tried 400, and it didn't boot because the vdimm was set too low. I then set it to 2.1 v (max) and I got these nasty vista errors. Then 333 fsb was tried, again the ram was at 800 using the strap control, and the 2.1v, but by then I think the damage was done. I have tested the ram (5 cycles of memtest +86 - test 5) and tested the drive. The machine is still BSODing. Now I have reset the bios to its defaults and so far so good.

Actually, besides what Regalk already said, the reason is in my last post. There is no way to lock RAM to any certain speed, with any motherboard. RAM speed always rises, no matter what, when the FSB rises. BTW, most likely the only thing you need to do to correct the problems you're having now is to "hard reset" your BIOS. To do this, take these steps, in order:

1. Unplug your power supply
2. Hold in your case's power button for 60-90 seconds
3. Take your computer somewhere without carpet, like the kitchen
4. Remove your motherboard's battery
5. Move the CMOS reset jumper from the normal position, to the shorted position
6. Wait at least 20 minutes, 30 is better
7. Reverse all of the steps, starting with #5, and working your way back up to #1

 

[jaga]

This is good stuff - the last two posts especially. Currently with auto settings and stock speeds, I am noticing a stable system so far - been on for 24 hours - but the motherboard temp is about 46deg (pcprobe) and I had an alarm on the vcore voltage which went to 1.63v. I will try the hard reset and post the results. Also, the auto settings give the ram slacker settings than spec i.e. 5,5,5,15 rather than 4,4,4,15 (http://www.ocztechnology.com/p...0_platinum_revision_2)

 

[SX2012]

overclocking increases the rate of electromigration along all of the circuits of a CPU. While an overclock may seem stable at first, usually they eventually become unstable sometimes even if you return to stock speeds. Whether its 6 days 6 months or 6 years its never predictable. Also Overvoltage can damage motherboard compontents as well.

An unstable overclock can damage windows if it crashes during bootup.

Also while some of you may or not be aware. A Stock CPU at 2.0GHz takes could take 150 watts at load, this same CPU overclocked just 500MHz could take 250 watts because of all the power leakage. So one could argue just from this case point that overclocking is seriously counter-productive. From the cost of electricity and extra cooling hardware you can practically just upgrade the cpu to something faster.

Some poeple have great success with overclocking while others dont. I remember a friend who overclocked his x2 3800+ to 2.7 GHz. It was stable for about 2 months straight be then it started acting strange and he had to bring it down to 2.5 GHz. Eventually he go tired of dicking with settings and timings and memtests that he just said to hell with it and set it back to stock.

 

[jaga]

Thats just it. I don't have time for this. IT is a time consuming and expensive hobby. I should have looked before I leaped...

 

[billy2003]

overclocking. my thoughts is ...if you are going from celeron 300a to 450, its good...or pinmod the k7 to 2.2 ghz is also stable...pushing the opteron to 2.5 is also stable..to me if i overcock my fsb and memory to the next level or spec if you will ...then its stable..anything in between or over is not...for most people better to get stock, avoid these time consuming hobby...right now there is no need to overclock anymore..not like in the old days...now cpu is fast enough already. get the fastest cpu you can afford and be happy.

 

[SX2012]

the only point in overclocking now is if you have the fastest CPU money can buy and the cash and time to blow on risky water cooling and sub-ambient cooling equipment. Or if its a hobby you like. Some people love to tweak settings day in and out.

for example, You can buy a budget Chip with a lot of overclocking headroom. But you can end up spending 50+ bucks for a good aftermarket heatsink and 150+ dollars on a good watercooling setup. Its just a pain in the ass i have gone through it all the equipment sits in a box now. Its easier to buy a new chip and sell the old one.

Everytime i saw someones system lock up they would try to blame it on anything but the overclocking(it was a fluke, its a problem with the game). Some overclockers are in huge denial about this stuff. It just makes troubleshooting weird computer activity a big hastle when you have an overclock on top of everything.

Ohh and I used to overclock my pentium 120 and my k6-2 400. Those were the good ol days. I tried to overclock my x2 4200+ but it didnt really do anything since the videocard is what matters now.

 

[jaga]

myocardio - I followed your instructions - not fully - and now the m/b has reduced 5deg in temp. I left the battery off and the cmos clear for 10 mins. Why the big procedure?

 

[Acanthus]

Originally posted by: SX2012
overclocking increases the rate of electromigration along all of the circuits of a CPU. While an overclock may seem stable at first, usually they eventually become unstable sometimes even if you return to stock speeds. Whether its 6 days 6 months or 6 years its never predictable. Also Overvoltage can damage motherboard compontents as well.

An unstable overclock can damage windows if it crashes during bootup.

Also while some of you may or not be aware. A Stock CPU at 2.0GHz takes could take 150 watts at load, this same CPU overclocked just 500MHz could take 250 watts because of all the power leakage. So one could argue just from this case point that overclocking is seriously counter-productive. From the cost of electricity and extra cooling hardware you can practically just upgrade the cpu to something faster.

Some poeple have great success with overclocking while others dont. I remember a friend who overclocked his x2 3800+ to 2.7 GHz. It was stable for about 2 months straight be then it started acting strange and he had to bring it down to 2.5 GHz. Eventually he go tired of dicking with settings and timings and memtests that he just said to hell with it and set it back to stock.

BS.

Im faster than the fastest money can buy, and im in for less than half the price of the CPU by itself. Power consumption is 30% over stock measured from the wall. You grossly exaggerated everything in your post.

Electromigration happens on *ALL* CPUs, if you stay with the CPU manufacturers stated "safe range" (which is in their whitepapers) you will not significantly reduce the life of the CPU as long as you are also in the safe range for thermals.