Athlon 64 X2 5600+ Failing Torture Test

[Cruelty]

Increase the Vram to 2.2V - it will cause you no harm - believe me. Also experiment with different memory slots, and most of all read the asus forums.
As for temperatures, IMO 66C is too hot for dual-core CPU, especially not overclocked, and (I know I am repeating myself) what you have described does not mean that the memory is failing, beacuse CPU has an integrated memory controller, it could still be related to CPU...

[paulunderwood]

From BIOS, what are the temperature readings for the CPU and for the mainboard?

[lbthe3rd]

The CPU temp with a slight load (Firefox and Itunes) is around 42-45 C, and the Mobo temp with same load is about 32-34 C.

After setting the Ram timing to 5-5-5-15 2T I went back and set the Voltage to 2.0v and changed the clock speed to 667MHz.

I booted up ran PC wizard 2007 and Prime95. The system began to pass the torture test. First the self test at 1024 FFT then at 8 FFT and so on. My mind a bit eased I minimized Prime95 and opened up Azureus to begin downloading torrents of some audio.

Within about five minute the system didn't fail, it insted just shut off and the rebooted. It didn't BSOD it just shut off. I unfortunately didn't get a glimpse at the temperatures before it happened because I wasn't expecting it to do so. This happened around 12:30 AM. I was tired so I waited for the reboot, opened Prime95 started running the torture test (on blend which is what I have always ran it on if that is important). Then I reopened Azureus and went to bed. When I awoke this morning the computer was a t the welcome screen. The same thing must have happened, because I got one of those "Your system has recovered from a serious error, would you like to send an error report?" dialog boxes.

By decreasing the Memory Clock speed what am I doing? And secondly how does DDR2 work. I mean should the clock speed be 400MHz (like 400 x 2 = 800) for DDR2 800 speed ram, or am I way off base with that assumption?

Thanks as always,
Louis

[paulunderwood]

Try the memory at 1.9v and set to DDR-667.

Setting it to DDR-667 is slowing it down from it's DDR-800 specification.

Have you installed the windows drivers for the motherboard's chipset? These should be on a CD that came bundled with the mainboard. (Just pop in the CD.)

[rx7350]

In the absence of an obvious cause that is quickly identified/corrected, as seems to be the case here, I suggest reducing the number of system components to the absolute minimum, and see if it will just boot to the BIOS. This means only a mobo, cpu (with fan), a power supply, and one stick of memory. Eliminate the hard-drive, additonal memory sticks, video cards, etc. If the minimum config will boot into BIOS, and stay there for an hour or so, then add the hard-drive (only), and boot to the OS, and see if it stays stable for an hour or so. Then add the additional memory stick, and repeat. And so on, until it fails. The last component you added at the time it fails is the culprit.

Of course, this is simplistic, and only checks for a 'hard-fail' due to a defective component. But, like experienced troubleshooters say, 'if you can't find what's wrong, start proving what's right, and then what's left is wrong'. And of course, this won't (probably) expose a load-related problem.

My guess is it has to do with the RAM - bad stick (or two), bad slot on the mobo, the sticks are in the wrong slot, etc. Or, possibly the cooling solution. I agree with CRUELTY - 66C is too high for a fully loaded stock-frequency dual-core cpu.

[paulunderwood]

With respect to CPU temperature. Using a high quality thermal compound (such as Artic Silver) will help. Applying the compound to the correct thinness, after thorough cleaning with a specialized fluid. A better heatsink and fan would help.

Try the ram in another slot.

[lbthe3rd]

I reduced the voltage to 1.9v and made sure the ram was in the correct mobo slots for running dual channel memory.

Ran torture test. CPU temps were running around 63 C. I don't know if that is still too hot. I am using Arctic Silver 5 thermal compound and the stock CPU fan, as to not invalidate the processor warranty. I have an ARCTIC COOLING Freezer 64 Pro 92mm CPU fan that is new if you think this would be better.

Anyway test ran for about an hour and a half, and was running when I left my house for work. when I got home about 10:30 PM it had restarted again.

I will try with the fan upgrade.

As to address rx7350's post, I had already reduced my system to mobo, power supply, hard drive, and processor, to rule out any faulty PCI or other add on components. I also ran prime95 from an Ubuntu live disc to rule out a software problem. The problem I am having is more than certainly between the memory the processor and the power supply (and their status, i.e. Temperature.)

My question is What is a good temp for my CPU? Can I monitor Ram temperature? and Is there anyway to rule out a faulty power supply, and if so what would be the signs (i.e. low voltage, high voltage etc)?

[paulunderwood]

Let us assume is either the PSU, the CPU or RAM.

Ram: try one stick of ram in single channel mode. Then try the other one in (single channel mode.)

CPU: try to get the temperature down. Try setting "Cool n Quiet" in BIOS to on. After, also try this with "CPU Q-Fan" to "performance". Let us know if this reduces the CPU temperature.

PSU: get it tested by an expert to see if it has enough amps. Also look in BIOS and tell us the readings for:
Vcore voltage? -- this should be between 1.3v and 1.35v
3.3v Voltage?
5v Voltage?
12v Voltage?

[Cruelty]

Originally Posted by lbthe3rd

I reduced the voltage to 1.9v and made sure the ram was in the correct mobo slots for running dual channel memory.

Have you tried changing the location of the RAM? I also have an ASUS board and what is written in manual is not always right...

Ran torture test. CPU temps were running around 63 C. I don't know if that is still too hot. I am using Arctic Silver 5 thermal compound and the stock CPU fan, as to not invalidate the processor warranty. I have an ARCTIC COOLING Freezer 64 Pro 92mm CPU fan that is new if you think this would be better.

using aftermarket cooler will invalidate your CPU warranty? what kind of warranty is that? Of course you should switch to Freezer64 if you have it

My question is What is a good temp for my CPU?

I would say that it should be below 60C

Can I monitor Ram temperature?

no you can't unless you have some digital thermometer

and Is there anyway to rule out a faulty power supply, and if so what would be the signs (i.e. low voltage, high voltage etc)?

you can eliminate PSU only if you tried another one and it worked fine, apart from that you may take a look at the voltages in idle and during stress as paulunderwood suggested to see if there are some major fluctuations.
BTW: do you have the latest BIOS version?

[lbthe3rd]

Thank you everyone for your help.

I ran the torture test for almost 17 hours and passed with flying colors. From what I gathered, this problem originated as a ram settings problem in the bios. Once we straightened that out and the test could run continuously, it became an overheating issue. After reapplying the arctic silver 5 and switching to the Freezer 64 my CPU temps dropped by 15 to 20 C at full load. I maxed out at around 52 C.

So I ended up with a better board than I started with, and a stable PC.

My final settings are:

Mem clock speed - 667 MHz
Mem timing - 5-5-5-15 at 2T
Mem voltage - 1.90v

And an after market CPU cooler.

I have but one question, why doesn't my Ram run 100% stable at 800 MHz if that is the advertised mem clock speed?


Any further posts or or info is greatly appreciated. Hopefully my problem will help others. If for any reason I start having problems reoccur, I will put up a post immediately.

Thanks much to all who helped,
Louis

[paulunderwood]

I have but one question, why doesn't my Ram run 100% stable at 800 MHz if that is the advertised mem clock speed?

Have your tried it set to DDR 800? If it is unstable at 1.9v try it at 2.0v.