The rig in my sig has run stable since July '05. I did extensive OC tests, and stability testing before turning it over to my son. Recently, he has complained of system lockups while playing BF2, CS:S and other games.
I ran the Memtest86 that is part of the DFI bios. I remembered that Test #5 was a good stressor. The test failed with one error every time after the 5th or 6th run. The system failed at the same register each time. I thought an OC config issue, or a faulty dimm. I switched the OC bios config to 'stock', and the test still failed.
Next, I started to swap dimms between the 4 sockets. I first swapped dimm1 and dimm2 sockets. I retested, and Memtest86 #5 failed at the same register as before. No help.
I next swapped dimm3 and dimm4 sockets. Upon retest, the failure reported a different memory register! It had to be a problem with dimm3 or dimm4. All I had to do is switch one more time, and I could identify the problem dimm.
I switched dimm3 (in the original dimm4 socket), with dimm1 (in the original dimm2 socket). Unfortunately, all runs of Test 5 PASSED! Not what I expected. I expected Test 5 to fail, either at the same register, or a new one.
Next thought was that just by reseating the dimms, I had cleared a resistance, and everything might be OK. I switched dimm3 and dimm 1 back to the original positions, retested, and it failed. Swapping dimm3 & dimm1 out of original position one more time, Memtest86 passed.
I don't necessarily want to rush out and replace the current RAM. The system is currently passing all Test5 runs, 50+ and going.
Questions:
1) Why would dimms be socket sensitive, especially considering it originally passed Memtest at system build time.
2) Is it likely that the 'bad' dimm is going to give failures again in the near future? (Duh!)
3) Is it possibly a 'quirk' in Memtest? The Memtest instructions suggest that Tests 5 & 8 can give strange results with AMD systems.
4) Suggestions?
Thanks!
I ran the Memtest86 that is part of the DFI bios. I remembered that Test #5 was a good stressor. The test failed with one error every time after the 5th or 6th run. The system failed at the same register each time. I thought an OC config issue, or a faulty dimm. I switched the OC bios config to 'stock', and the test still failed.
Next, I started to swap dimms between the 4 sockets. I first swapped dimm1 and dimm2 sockets. I retested, and Memtest86 #5 failed at the same register as before. No help.
I next swapped dimm3 and dimm4 sockets. Upon retest, the failure reported a different memory register! It had to be a problem with dimm3 or dimm4. All I had to do is switch one more time, and I could identify the problem dimm.
I switched dimm3 (in the original dimm4 socket), with dimm1 (in the original dimm2 socket). Unfortunately, all runs of Test 5 PASSED! Not what I expected. I expected Test 5 to fail, either at the same register, or a new one.
Next thought was that just by reseating the dimms, I had cleared a resistance, and everything might be OK. I switched dimm3 and dimm 1 back to the original positions, retested, and it failed. Swapping dimm3 & dimm1 out of original position one more time, Memtest86 passed.
I don't necessarily want to rush out and replace the current RAM. The system is currently passing all Test5 runs, 50+ and going.
Questions:
1) Why would dimms be socket sensitive, especially considering it originally passed Memtest at system build time.
2) Is it likely that the 'bad' dimm is going to give failures again in the near future? (Duh!)
3) Is it possibly a 'quirk' in Memtest? The Memtest instructions suggest that Tests 5 & 8 can give strange results with AMD systems.
4) Suggestions?
Thanks!
Facebook
Twitter