Temperamental Processor

M

morfinx

[H]ard|Gawd
Joined
Feb 18, 2005
Messages
1,105
Got my Opty 165 a few days ago and overclocked to 2.7Ghz @ 1.35V (stock voltage). Orthos 8+ hrs stable plus several 6+ hrs Guild Wars sessions with no problems. Well, this morning I woke up, and found the computer locked up. No big deal, just a fluke I thought. I restarted, it didn't even finish loading startup programs in XP before it crashed and restarted itself. I restarted a few more times, and it always crashes and restarts within a couple of minutes of booting into Windows.

So I thought, maybe my memory's going bad. I fire up Memtest86, memory's fine. PSU crapping out? Not so much, 5V and 12V rails are stable. Perhaps temp's too hight? Nope, CPU is idling in the 20's. Nothing changed except for the temperature went down overnight and I didn't have heat on, so the processor was running a few degrees cooler. I crank Vcore up to 1.4V, launched Orthos, and went about my business. 4 hrs later, everything is stable. I thought, hmmmm, strange. I turned Vcore back down to 1.35V, launched Orthos again. And so far it's been 30 mins, no errors and no crashes.

What a temperamental processor! :p
 
Ok this is getting strange. The computer rebooted last night at 1.35V, so I upped the Vcore back to 1.4V and everything seemed fine. But then, the computer rebooted itself again this morning, with the CPU @ 1.4V. I'm running Orthos right now and will let it run for 8 hrs (so far it's been 1.5 hrs, no errors)

Well I'm stumped. Any ideas?
 
When your system restarts itself and goes into windoze, does it report that it has "recovered from a serious error". If it does, get the error codes and google it.

You mentioned adjusting the vcore... what about voltages for RAM? What are they set at? Is that Mushkin in your system PC-3500? (I always thought Level 2, rev 2 were BH5, but that may have only been for the PC3500s)

How long did you memtest your ram for? and what tests?

Your CPU idles in the 20's??? With air cooling? Hell, I'm running a naked Opty 170 under water and I idle at 32c. What's your ambient temp? What are you using to measure temps?

What kinda power supply are you running... and how did you measure those volts (software, BIOS or Multimeter)?

Normally, when a PC reboots itself without a BSOD, I'll suspect the power supply first. If the PS checks out... then my attention turns to RAM.
 
NordicRX8 said:
When your system restarts itself and goes into windoze, does it report that it has "recovered from a serious error". If it does, get the error codes and google it.
Click to expand...
It did one of the times, but most of the time it does not. I will look it up if it happens again.

You mentioned adjusting the vcore... what about voltages for RAM? What are they set at? Is that Mushkin in your system PC-3500? (I always thought Level 2, rev 2 were BH5, but that may have only been for the PC3500s)
Click to expand...
RAM is at 2.8V which is the max recommended for the Mushkin's according to the manufacturer. It's PC3200.

How long did you memtest your ram for? and what tests?
Click to expand...
Two passes in Memtest86. I've had the same RAM running at the same speed for the last two years, so I didn't do any extensive testing. I will run Memtest86 overnight tonight.

Your CPU idles in the 20's??? With air cooling? Hell, I'm running a naked Opty 170 under water and I idle at 32c. What's your ambient temp? What are you using to measure temps?
Click to expand...
Temps are from SpeedFan and BIOS. Case ambient temp is 26-28C.

What kinda power supply are you running... and how did you measure those volts (software, BIOS or Multimeter)?
Click to expand...
Enlight 420W. I'm measuring with a digital multimeter. Idle voltage stays at exactly 12V, load voltage goes down to 11.77V, which is far above the minimum 11.4V specified by the +/- 5% for ATX. I'm starting to think the PSU might be the weak point, I've had the thing for 5 years, it's served me well but might be time for something new. I'm looking at Seasonic and PCP&C.

Normally, when a PC reboots itself without a BSOD, I'll suspect the power supply first. If the PS checks out... then my attention turns to RAM.
Click to expand...
My thoughts are the same. Just wanted to get a second opinion.
 
morfinx said:
Two passes in Memtest86. I've had the same RAM running at the same speed for the last two years, so I didn't do any extensive testing. I will run Memtest86 overnight tonight.
Click to expand...
Try running just test # 5 for at least 20 passes... then try test # 8 for the same amount. Every stick of ram I've had error out on me, happened during test # 5. Some may not believe in "burning in" RAM, but I used test #5 & #8 to burn in my Mushkin PC3500 (BH5), and after the burn in, was able to raise FSB while keeping my 2-2-2-5 settings.


morfinx said:
Temps are from SpeedFan and BIOS. Case ambient temp is 26-28C.
Click to expand...

wow...your CPU temp (20c) is BELOW ambient??? something is awry. Try CORETEMP :D

morfinx said:
I'm looking at Seasonic and PCP&C.
Click to expand...

Good choices!
 
NordicRX8 said:
Try running just test # 5 for at least 20 passes... then try test # 8 for the same amount. Every stick of ram I've had error out on me, happened during test # 5. Some may not believe in "burning in" RAM, but I used test #5 & #8 to burn in my Mushkin PC3500 (BH5), and after the burn in, was able to raise FSB while keeping my 2-2-2-5 settings.
Click to expand...
I will try it and report back.


wow...your CPU temp (20c) is BELOW ambient??? something is awry. Try CORETEMP :D
Click to expand...
No no, I said in the 20's, I didn't say below ambient ;) It usually idles at about 1-2C above ambiet.

It's been 8 hrs in Orthos and it's still stable. The crashes seems so random. If it is the PSU, how can it remain stable for multiple 8+ hr Orthos runs and multiple hr heavy 3D gaming sessions?
 
Well, 30 passes with test 5 and 25 passes with test 8 in Memtest86, both no errors.
 
morfinx said:
Well, 30 passes with test 5 and 25 passes with test 8 in Memtest86, both no errors.
Click to expand...

Good to hear!

morfinx said:
It's been 8 hrs in Orthos and it's still stable. The crashes seems so random. If it is the PSU, how can it remain stable for multiple 8+ hr Orthos runs and multiple hr heavy 3D gaming sessions?
Click to expand...

That does boggle the mind... try to cature one of those error codes if the system crashes again. You may want to make sure you have a good backup :D

When was the last time Windows was installed (from scratch)? Is it patched to current levels? Have you scanned your system for virii and malware?

I still think its hardware related, but until you can produce a stop error code, it'll be tougher to diagnose. I've always hated the replace and retest procedure versus the diagnose then repair procedure (the same difference between a "grease monkey" and an "automotive technician"), unless of course you have 'known good" computer components laying around, so it won't cost you anything out of pocket.

I just thought of another program you can try... it's called OCCT. Do a google on it to find it. It may not test dual cores, but what it does do... is stress the CPU, and measures (and outputs log files) the voltage signals (for dips and spikes) during stress testing. Take a look at the log files it outputs to see if you have excessive ripple in any of your voltage lines. I had an Antec SL-350 that worked fine for 2 years and suddenly became the culprit of random reboots. OCCT showed excessive 12v ripple and after replacing the Antec the reboots disappeared... the OCZ that replaced the Antec (higher wattage as well), showed steady power across all voltage rails.
 
I'm starting to think it's the PSU more and more. It crashed again this morning while I was asleep. According to SpeedFan log, it was at 5:47am. It did give a BSOD this time, BCCODE 9C. I looked it up, and it seems like a CPU related issue, which can certain be a power related problem:

http://support.microsoft.com/?kbid=329284

According to SpeedFan log, Vcore seems to fluctuate a bit, which might be the problem. I find it very interesting that crashes seem to ONLY happen at idle or near-idle, which is counterintuitive. I have had absolutely no crashes under full load situation.

Windows install was recent, within the month, fully patched. Spyware and Virus check is done every night.

It sounds like OCCT does the same thing as Orthos + SpeedFan logging.

I have a Seasonic S12 Energy+ 550W on the way, hopefull that'll fix what ails me :p
 
Hmmm... found something telling while examining SpeedFan logs. When the cores are fully loaded, the Vcore regulation (for whatever reason) is much tighter. With Vcore set to 1.4V in the BIOS, when idle, Vcore swings from 1.36 - 1.41V, with spikes up to 1.44V and dips down to 1.32V. When loaded, Vcore never once went below 1.39V or above 1.41V. Orthos stable for 11 hrs so far, will let it run for 24 hrs.

Now I'm more and more positive it's the PSU.
 
Another update:

Orthos stable for 24 hrs at 1.4V. Backed it down to 1.35V, and has been Orthos stable for 8.5 hrs so far. I'm thinking definitely a power related problem now. Thoughts, Nordic?
 
UGH!! I installed the Seasonic, voltages look a lot better. Absolutely no drop at all during full load. Then, it crashed overnight AGAIN, same stop code. This is the same setup that under full load of Orthos stable for over 24 hrs. Hell, I've been running Orthos nonstop for the last few days while waiting for the new PSU, no crash at all! Why does it only crash under idle conditions??? Nordic, or anyone else, I'd appreciate any help!
 
morfinx, are the lines stable at idle? I believe you said that was where you were experiencing swings with your prior PSU. My S12 600's (I have two) are stable throughout.
 
tvh55 said:
morfinx, are the lines stable at idle? I believe you said that was where you were experiencing swings with your prior PSU. My S12 600's (I have two) are stable throughout.
Click to expand...
Yes all rails are rock solid at idle. However, core voltage fluctuates a lot more when idle (as much as +/- 0.08V), but when under load it's almost perfectly still. That's the only potential link I can think of now. Very strange behavior.

What I think is happening is, while it's been tested to be completely stable at 1.35V, it may not be when Vcore drops below that. And possibly the crash is caused by voltage dropping below a certain threshold, even if just for a moment.
 
Could be something to do with the voltage regulators on your motherboard rather than your powersupply. That's my guess
 
the a8n-e isnt the best OCer and your 165 does only have a 9x.... have you updated your bios to the newest one


also have you upped your chipset voltage? and what is your HT speed at?

FSB of 300 on a A8N-E might be your problem... i will see if i can find some other OCs on the board..


A8N-E Bios version 1005
1.Support new CPUs. Please refer to our website at: http://support.asus.com/cpusupport/cpusupport.aspx
2.Fixed "ASUS Update" can't update BIOS when using Dual Core CPU.
3.Fixed NOS can't work after enabled NOS by AI-Booster.
4.Add more DRAM Items in the Bios.
5.Fixed it fails to boot when Enabled NOS function.
6.Fixed ATI X600, X800 VGA Card sometimes fails to enter OS.
7.Support Single Core CPU cool&Quiet function under Windows 2000.
8.Set the lowest frequency of Dual Core CPU with Cool&Quiet to 1000MHz.
9.Fixed sometimes CPU Vcore may be abnormal.
 
Adidas4275 said:
the a8n-e isnt the best OCer and your 165 does only have a 9x.... have you updated your bios to the newest one


also have you upped your chipset voltage? and what is your HT speed at?

FSB of 300 on a A8N-E might be your problem... i will see if i can find some other OCs on the board..


A8N-E Bios version 1005
1.Support new CPUs. Please refer to our website at: http://support.asus.com/cpusupport/cpusupport.aspx
2.Fixed "ASUS Update" can't update BIOS when using Dual Core CPU.
3.Fixed NOS can't work after enabled NOS by AI-Booster.
4.Add more DRAM Items in the Bios.
5.Fixed it fails to boot when Enabled NOS function.
6.Fixed ATI X600, X800 VGA Card sometimes fails to enter OS.
7.Support Single Core CPU cool&Quiet function under Windows 2000.
8.Set the lowest frequency of Dual Core CPU with Cool&Quiet to 1000MHz.
9.Fixed sometimes CPU Vcore may be abnormal.
Click to expand...

BIOS is the newest (1013), HT is at 3x (900Mhz). Chipset voltage is stock. I don't think it's a FSB issue. As I said, I basically ran Orthos in the background for 3-4 days straight without a single failure. Crashes only happens at idle.


Could be something to do with the voltage regulators on your motherboard rather than your powersupply. That's my guess
Click to expand...
That's my suspicion as well. The mobo is less than a month old, and I have 3 more days for RMA if I choose to. I'm kind of limited as to which mobo. I need a S939 NF4 Ultra mobo with 3 usable PCI slots.
 
Adidas4275 said:
well id return the mobo and get this


http://www.newegg.com/Product/Product.asp?Item=N82E16813136163

3 pci slots NF4 and will OC well...i justy built one not to long ago and it it is stable to about 340fsb
Click to expand...
I have looked at it when I was picking out a mobo. I'm weary of the low ratings this board received. 35% of reviewers gave it a 1 or 2, that's pretty significant.

Also, I have a double slot video card, which will render one of the PCI slots unusable.
 
Tried various voltages, the thing is just not stable at IDLE. I have never heard of a problem like this before. Seems like a lot of other members are running the same board with no problems. I'm starting to wonder if it's just my board. Looks like S939 boards are going out of stock quickly on NewEgg, A8N-E isn't even in stock anymore.
 
are you sure it's idle overnight/early AM? when do your virus checks and spyware scans run?

i only ask b/c mine are overnight.
 
zero_vertical said:
are you sure it's idle overnight/early AM? when do your virus checks and spyware scans run?

i only ask b/c mine are overnight.
Click to expand...
My spyware/virus scanners do run in the early AM. However, the computer still crashes during day time if left alone for a couple of hours.
 
zero_vertical said:
call a priest!
Click to expand...
More like an exorcist!

Interesting revelation: I dropped the FSB down to 289Mhz from 300Mhz, which results in 2.6Ghz CPU speed. Preliminary testing shows it idles stable for 3 hrs, will have to confirm overnight. So it does look like the board is holding me back. Although the manifestation of it is rather strange (stable at full load but not at idle). Assuming the current theory is confirmed by overnight testing, I have several options:


  • 1. Keep it at 2.6Ghz and call it a day
  • 2. Keep it at 2.7Ghz and keep Orthos running in the background (with lowest thread priority) 24x7 to maintain stability
  • 3. Return the board, and pay twice as much ($150) and get DFI LANPARTY UT nF4 SLI-DR Expert, which is the only reasonably good S939 OC board with 3 usable PCI slots, even though I won't be using the SLI feature. Keeping in mind I have a double slot video card. With that I can hope to hit 2.8+

So what do the gentlmen of the board (and ladies, but probably not) think?
 
morfinx, I'm reasonably certain I've seen others with that board in their signatures who were reporting 300+ FSB's without problems. I'd check the forums and try to contact them. Maybe they're using a different Bios or settings than you are.

Good luck:)
 
tvh55 said:
morfinx, I'm reasonably certain I've seen others with that board in their signatures who were reporting 300+ FSB's without problems. I'd check the forums and try to contact them. Maybe they're using a different Bios or settings than you are.

Good luck:)
Click to expand...
Believe me I have looked, no one with this motherboard seems to have gone past 300FSB. There are only so many settings, and I've been overclocking for 10 years (I feel old :(). If you read through this entire thread, the motherboard seems to be the only limiting factor.
 
morfinx said:
  • 1. Keep it at 2.6Ghz and call it a day
  • 2. Keep it at 2.7Ghz and keep Orthos running in the background (with lowest thread priority) 24x7 to maintain stability
  • 3. Return the board, and pay twice as much ($150) and get DFI LANPARTY UT nF4 SLI-DR Expert, which is the only reasonably good S939 OC board with 3 usable PCI slots, even though I won't be using the SLI feature. Keeping in mind I have a double slot video card. With that I can hope to hit 2.8+

So what do the gentlmen of the board (and ladies, but probably not) think?
Click to expand...

Anyone? Overnight stable at 2.6, preliminary testing shows stable at 2.65 as well.
 
You must log in or register to reply here.
Back
Top