Hi,
Looking for advice on what to test/try next. I have an Asus P5N32-SLI Premium Wifi/App board with Core 2 Duo E6750 and 6Gb Crucial Ballistix (2x1GB, 2x2GB) ram which is now horribly unstable.
I bought the system back in 2007 with 2x1Gb crucial ballistix. I was running it at 1333 FSB (motherboard overclock - rated max is 1033) for quite a while and everything was stable. Then I upgraded, adding 2x2GB ballistix. The trouble was the first pair wanted 2.2V at 4-4-4-12 and the 2nd pair wanted 2.0V. I had to lower the timings of both to 5-5-5-18 at 1.8V to get it running/stable. Some time after that it would not POST on a reboot, it would report a memory error (constant/long beep). Google tells me many people have had this same problem, often with a new board. So I must have been lucky with my initial setup.
Recently, 3 months ago, it started to blue screen, and the frequency of these started to increase. The minidumps didn't show any sort of pattern, they were all over the place in all sorts of drivers etc. Which made me suspect a memory, cpu or motherboard problem.
I started by testing the memory at 5-5-5-18 (1.8V) a stick at a time in memtest86+. All sticks tested ok individually, in pairs, and all 4 together for at least 1 pass. So, I started testing with Everest and Prime95. Both cause errors after 1-2 minutes, sometimes as soon as 15-30 sec. Occasionally the computer would blue screen during testing. The interesting thing is that prime95 will run stable for 45 mins on small ffts, which according to google indicates the cpu/cache itself is ok. large ffts involve the motherboard memory controller and memory, this and blend torture testing fail.
I used HWMonitor to watch temps and voltages during testing and these seem good/stable. I have also tried 2 other power supplies, and without my 8800GTS (using a new passively cooled ATI card instead), to rule the power supply out as the problem.
I am running the latest BIOS and drivers for all components.
I tried downclocking the cpu and running the mobo at 1033 fsb, setting the ram to looser/lower timings and voltages.. it doesn't seem to make any difference. However, I did notice the old 1x1gb ram by itself was slightly more stable, and lasted longer in prime95 than all 4 sticks together, or the newer 2gb sticks by themselves. I think the newer sticks are poorly made, one has a bent heatsink which is lifing off the ram .. it has always been that way .. I should have RMA'd it really.
I have seen suggestions that the northbridge cooling is insufficient, and this is what causes the POST long beep memory error etc. I wonder if my cooling has come away from the board slowly over time. It has 2 heatsicks on the chipset chips, and cooling pipes. There are also 2 small fans for these hestsinks, however 1 will not fit as my cpu cooler over hangs it.. the cooler has a plastic shield which appears to be directing air onto that heatsink.. and one thing I haven't tried is putting the other mini fan on the other chip .. I don't even know which the other chip is northbridge or southbridge, so that might be my first possible next test/step.
I have since bought a new i5 system to replace it (I wanted to spend as little as possible and still manage a bit of performance .. managed a 3.8Ghz overclock on the cpu so I'm happy with that) but I now want to put the old components back together for my inlaws. So, I've posted a WANTED on the sales forum for a LGA 775 mobo to test with, but I also wondered if there is anything else I can do/test to attempt to rule out the cpu or memory. If I can be 100% certain the motherboard is to blame for this then I could justify buying a new board and going from there. But, the last thing I want is to buy a new board and find the cpu or ram was the problem.
So, any thoughts?
Looking for advice on what to test/try next. I have an Asus P5N32-SLI Premium Wifi/App board with Core 2 Duo E6750 and 6Gb Crucial Ballistix (2x1GB, 2x2GB) ram which is now horribly unstable.
I bought the system back in 2007 with 2x1Gb crucial ballistix. I was running it at 1333 FSB (motherboard overclock - rated max is 1033) for quite a while and everything was stable. Then I upgraded, adding 2x2GB ballistix. The trouble was the first pair wanted 2.2V at 4-4-4-12 and the 2nd pair wanted 2.0V. I had to lower the timings of both to 5-5-5-18 at 1.8V to get it running/stable. Some time after that it would not POST on a reboot, it would report a memory error (constant/long beep). Google tells me many people have had this same problem, often with a new board. So I must have been lucky with my initial setup.
Recently, 3 months ago, it started to blue screen, and the frequency of these started to increase. The minidumps didn't show any sort of pattern, they were all over the place in all sorts of drivers etc. Which made me suspect a memory, cpu or motherboard problem.
I started by testing the memory at 5-5-5-18 (1.8V) a stick at a time in memtest86+. All sticks tested ok individually, in pairs, and all 4 together for at least 1 pass. So, I started testing with Everest and Prime95. Both cause errors after 1-2 minutes, sometimes as soon as 15-30 sec. Occasionally the computer would blue screen during testing. The interesting thing is that prime95 will run stable for 45 mins on small ffts, which according to google indicates the cpu/cache itself is ok. large ffts involve the motherboard memory controller and memory, this and blend torture testing fail.
I used HWMonitor to watch temps and voltages during testing and these seem good/stable. I have also tried 2 other power supplies, and without my 8800GTS (using a new passively cooled ATI card instead), to rule the power supply out as the problem.
I am running the latest BIOS and drivers for all components.
I tried downclocking the cpu and running the mobo at 1033 fsb, setting the ram to looser/lower timings and voltages.. it doesn't seem to make any difference. However, I did notice the old 1x1gb ram by itself was slightly more stable, and lasted longer in prime95 than all 4 sticks together, or the newer 2gb sticks by themselves. I think the newer sticks are poorly made, one has a bent heatsink which is lifing off the ram .. it has always been that way .. I should have RMA'd it really.
I have seen suggestions that the northbridge cooling is insufficient, and this is what causes the POST long beep memory error etc. I wonder if my cooling has come away from the board slowly over time. It has 2 heatsicks on the chipset chips, and cooling pipes. There are also 2 small fans for these hestsinks, however 1 will not fit as my cpu cooler over hangs it.. the cooler has a plastic shield which appears to be directing air onto that heatsink.. and one thing I haven't tried is putting the other mini fan on the other chip .. I don't even know which the other chip is northbridge or southbridge, so that might be my first possible next test/step.
I have since bought a new i5 system to replace it (I wanted to spend as little as possible and still manage a bit of performance .. managed a 3.8Ghz overclock on the cpu so I'm happy with that) but I now want to put the old components back together for my inlaws. So, I've posted a WANTED on the sales forum for a LGA 775 mobo to test with, but I also wondered if there is anything else I can do/test to attempt to rule out the cpu or memory. If I can be 100% certain the motherboard is to blame for this then I could justify buying a new board and going from there. But, the last thing I want is to buy a new board and find the cpu or ram was the problem.
So, any thoughts?
