Unknown psu or motherboard issue.

[painterh]

Hello good people. I have a problem and I'm not sure where to put the blame. I'll give a brief description of the problem after I tell you my specs. My system is homebuilt almost 2 years ago. It is as follows:
Case-- Coolermaster Centurion 5
PSU-- Antec NEOhe 500W (modular cabling) More on that in a moment
Cpu-- Opteron 170 w/Thermaltake "Big Typhoon" cooler
MB-- Asus A8N-SLI Deluxe
Ram-- 1 gig Corsair Twinx XMS dual channel kit (2 x 512) 2-3-3-6 1T timings
Video- Asus Geforce 7600gs 512megs (passive cooling)
Floppy/cardreader-- dvd/rw drive-- 2 Seagate 160 Sata hard drives


My system started having power issues some time ago. The first psu I put in when I built it was an Antec TruPower 500W which started having issues and had to be RMA'd back to Antec. They were very good about it and replaced the defective one with a NEOhe 500W (seems the Trupower 500 was discontinued). Things went ok for a while untill I started having issues starting the computer up. If it was shut down I had to press the start button up to five or six times before it would finally start, (all the while the fans and light would be coming on and then going off. Sometimes if I tried too many times to start up it would reset the cmos and I would have to go back into bios and set up all over again.

Again I called Antec about it and went over the symptoms, switching the power and reset switches on the motherboard but made no difference. Again they were very good about replacing it. Now I am on my third Antec psu, and it did indeed solve the startup problems. Now, however, after only a week or so since installation, my computer has developed some new power related problems. I have had this system overclocked to 2.8ghz stable for quite some time, and last night after doing some tweaking I ran prime95 for about 8 hours. (No errors!) I stopped the test this morning and figured all was well. Then I tried to run Super_pi multiple times and could not get a correct checksum. I tried several small changes in voltage (vdimm, vcore and ram timings) none made any difference in super_pi, (still incorrect checksum). I even set the bios back to default settings and still could not get a correct checksum. I ran memtest 86 for several complete runs and had no errors.

Now this evening, I noticed (using speedfan) that my temps were randomly spiking and then going back to normal. I wasn't even running any programs at the time. I checked using cpuz and found my vcore was fluctuating from a low of 1.39v all the way up to 1.5 on occasion. I don't want to think about another RMA, but I will if I must. Does anyone know if this could be a bios issue or some hardware issue on the board. I keep all my mb drivers up to date and bios is latest version, but it seems unlikely that 3 PSUs from the same manufacturer would all turn out to be lemons. I f anyone has any Ideas, I would be most grateful for the input. Thanks in advance.

 

[QuietIce]

Just a nay-sayer opinion that you may not want to hear. It's possible one of the PSU issues damaged a MOSFET on the board. That MAY not be the case, but it's possible. Let's get some more information first before saying the sky is falling.

What are you using to measure vCore, the BIOS or CPU-Z?
If you've used both, are they both reading more or less the same?

What are the BIOS 3.3v, 5.0v, and 12.0v readings?

Do you happen to have a multi-meter ...?

 

[painterh]

Thanks for the reply. In Windows I used cpuz and bios, in Linux I used xsensors. I reflashed the bios last night with a fresh downloaded copy of same bios , and while it may not have solved the problem, I think it might have moderated the spiking a little. The 12 volt readings seem to hold right at 11.97v, and the 3.3 seems to be right at 3.25. I don't understand how I could run Prime95 for 8 hours and still get a constant "incorrect checksum" using super_pi.
It doesn't seem to matter if I run 64k calc or 1 million, I always get invalid checksum. I even tried rolling back my overclock to default settings in bios but same result in super_pi. I will probably have to open the case and see if I can see anything noticeable on the MB (maybe capacitor bulging or something. I'm going to be very irritated if it turns out to be some collateral damage from one of Antec's PSUs. It seems to be running great right now, even though it won't pass super_pi.

 

[QuietIce]

Odd that a BIOS update would correct a fluctuating vCore. How bad is it now with a rock-solid load, like 2xPrime95? I tend to have a 0.03-0.04v drop under full load compared to idle, so that's normal, but it shouldn't wobble up & down much when it's under a constant load like that.

It IS odd that SPi would error out when Prime95 doesn't. Does the error happen on the same loop every time for a given calc ...?

 

[maxfly]

i agree with Quiet,it definately sounds like the psu issues you had may have damaged your mb.another easy suggestion would be to check for bulging or bad caps and any small burn marks on the mb pcb.do you happen to have a mb that you can test your gear with?

 

[RJARRRPCGP]

The HDD may have a bad sector and SuperPi may be located in one of them.

An oddball error problem like that makes me suspect bad sectors.

 

[painterh]

In response to QuietIce, super_pi runs all the way to checksum on every test. It doesn't ever give me an error during a run, It just gives me an incorrect checksum when I try to validate the time and checksum on super_pi mod website. I don't have another setup to test any components. My only other computer is my wifes (the one in my sig) and it is off limits. (that xp mobile has sure been a great little cpu). How can I run 2 copies of prime 95 simultaneously? When ever I try to open a second copy, it just reverts to one. I will try checking my hd for errors. I have a utility disk from Seagate for just such diagnostics. I sure hope I can narrow it down. Thank you folks for all the help so far. It's good when an old fart like me can learn from smart young minds. I'll keep you posted.

 

[QuietIce]

The HDD may have a bad sector and SuperPi may be located in one of them.

An oddball error problem like that makes me suspect bad sectors.

Another distinct possibility! That's why I was asking when the error occurred in SPi. If it's the same place each time I'd be tempted to re-install SuperPi.

Edit:

It's good when an old fart like me can learn from smart young minds. I'll keep you posted.

Why, THANK YOU!!!

It's been a loooong time since anybody has referred to ANY part of my body as "young" ...!

 

[painterh]

You're very welcome. Ever since I passed (gulp) 55, most folks seem young (except for the ones I see when I drop off my wife at work where she helps care for seniors in assisted living). I am still wondering how to run 2 instances of prime95 at same time to make sure I test both cores simultaneously?

 

[painterh]

I was just thinking about the bad sector idea, and I am thinking that's probably unlikely. I have run superpi from Windows (one hd) and also in Windows install on Virtualbox within Linux (different hd). Same result, different hd.

 

[QuietIce]

To run 2xPrime95 copy & paste your current P95 folder. I rename the folders Prime95_0 & Prime95_1 but you can use whatever you want. Then open the folders and start Prime95 being sure to select affinity 0 for the _0 folder and affinity 1 for the _1 folder. That'll let you run two instances at the same time - one on each core.

Note: The affinity setting is under the Advanced drop-down menu in Prime95 and it will keep that setting when you exit the program so you should only have to do it the first time. After that the program from the P95_0 folder will run on core 0 and the program from the P95_1 folder will run on core 1. I like to put shortcuts to them on the desktop or in the Start Menu until I'm done OC'ing, then I erase the shortcuts ...