Suspected faulty psu

[surfjunkie]

So I recently began my journey into overclocking. I have a 955BE in a M4-79XTDEvo and 4G Gskill Ripjaw 1600. I have a Noctua NHU12P-SE2 cooler in Antec 900 Case. Right now my specs are as follows:

FSB: 243mhz
CPU: 3645mhz
VCore: 1.4125
NB: 2637mhz
NB Volt: 1.2
HT: 2187mhz
GSKill Ripjaw 1600 Ram @ 1620mhz
XFX 6850 @ 875/1145mhz

PSU: Antec Basic 550W (I think that was the name)

I can run prime95 for 2hrs with no problem but temps get up to about 58C (41C at idle), my room temp is 27C@51% humidity. This is not a large overclock, so I was expecting cooler temps from the Noctua. I can run multiple passes with memtest with no errors. I can run furmark with no artifacts for over an hour. When I play some Bad Company 2, I crash in about 25 minutes. If I run P95 and Furmark together one of my cores will cease completing tests in P95. I have also tried just using multipliers to overclock. That leads me to believe the PSU is faulty. I did not think this humble system would draw that much wattage. Also, PC Probe reports my Vcore @ 1.45 at idle and it drops to 1.41 at load. CPUZ will show 1.43 at idle and drop to 1.41 at idle. Overdrive always shows 1.41.

In the future I plan to add one more 6850. Should I opt for a 650W or 700-750W? There is about a $30 difference between a 650 and 750 and budget is a huge factor right now. Thanks for your input.

 

[Slugbait]

Yeah, I think you're jumping to conclusions. First thing to try is to pull the HSF, clean it up, re-apply the thermal, and pay close attention to remount of the HSF.

 

[VirtualLarry]

Yeah, that doesn't immediately scream "PSU issue" to me either.

 

[bryanl]

I would first suspect the lowest quality device in the computer, generally the one with the gaudiest appearance, made from inferior components of undeterminable origin, and with a silly name that suggests xtreme (as opposed to extreme) action. The power supply is not the problem unless its voltages become too low, and you should measure them with a meter. However merely removing the case side panel to access the measurement points may make the computer run cool enough to prevent failures.

 

[surfjunkie]

sorry for not responding...I was expecting to get an email when there was a response.

I will reinstall the HSF and make sure they were no mistakes

 

[surfjunkie]

I reinstalled HSF. P95 is running blend. Ambient temp is 28*C/47% Humididty. Cpu is running 55*C and MB is 30*C. My 3.3v reading on PcProbe is 3.48 which is above 5% if that matters.

There are three things I don't understand:

1. Why are the voltage discrepancies there, Overdrive, cpuz and PCProbe all used to read identical.

2. Why does VCore drop 50mv when the cpu is loaded.

3. Why doesn't the pc lock-up when just the cpu is loaded. The case has good airflow as MB temp usually only reads 1*C above ambient.

It seems that when the PSU is really loaded with both vga and cpu that there becomes a problem. I really thank you guys for your input as this appears to be a little over my troubleshooting knowledge.

Edit: 45 minutes running prime blend with 27 minutes of running furmark. CPU temp never went over 55*C and VGA never went over 71*C. No crashes - I suppose I am looking for a BC2 problem. Don't even have a memory dump to look at.

 

[PreferLinux]

I reinstalled HSF. P95 is running blend. Ambient temp is 28*C/47% Humididty. Cpu is running 55*C and MB is 30*C. My 3.3v reading on PcProbe is 3.48 which is above 5% if that matters.

There are three things I don't understand:

1. Why are the voltage discrepancies there, Overdrive, cpuz and PCProbe all used to read identical.

2. Why does VCore drop 50mv when the cpu is loaded.

3. Why doesn't the pc lock-up when just the cpu is loaded. The case has good airflow as MB temp usually only reads 1*C above ambient.

It seems that when the PSU is really loaded with both vga and cpu that there becomes a problem. I really thank you guys for your input as this appears to be a little over my troubleshooting knowledge.

Measure the voltages with a multimeter. Your PSU should be fine. Try with no overclock at all.

1. Doesn't matter. They must use different methods of measuring the voltage.

2. There is some amount of resistance in the power delivery circuitry, so there will be a higher voltage drop over that resistance with a higher current (which is caused by the higher load), which will mean less voltage for the CPU.

3. It could be because of the PSU, although that would probably cause reboots without any errors whatsoever. It could be drivers, I guess. Could it be the RAM? (Extra RAM being used than normal as the video card uses plenty if it can for texture caches, and P95 will use plenty in blend mode too.)

 

[Ghiedo27]

It sounds to me like your IMC is flaky due to heat. I'd drop the volts down until your cooler can keep the cpu under 50c at full load and then see how high of an OC you can get.

Until you know what temperature and voltage the memory controller is stable at, I'd keep the memory itself as close to stock as possible (so stick to multiplier OC only). The video card may as well be at stock as well. With the second one you're planning on putting in there you're going to be CPU bottle necked anyways, imo.

 

[surfjunkie]

I have dropped everything back to stock speeds. Load tested for about 20 minutes and my temps were at 48*C. I did not have time to test BC2 out (maybe tonite). When using the stock hsf and playing bc2, temps were around 55*C as well. I am somewhat disappointed in the NH-U12, I was expecting a little better than just 7*C drop over stock hsf.

 

[surfjunkie]

well no dice...played for 12 minutes before bc2 froze and reboot.

EDIT: After 20 minutes with prime and furmark running the NB heatsink is about 43*C checked with infared thermometer.