Design a test for specific Err-Bits

[gmatt]

I found that my system is giving me errors at exactly 2 memory locations with the same error bits 00010000. The 2 locations are 00098456f14 and 000983daf14.

I tried writing a test that would test a pattern I know that those two locations have failed previously, however it seems pretty random, there is no guarantee that they will fail that pattern again.

Is it possible to design a test that will find these 2 errors consistently even if I start swappng out memories? I don't want to have to wait 48hrs + everytime I do a swap for a thorough test ( I have 8GB of ram, it takes about 24hours to get 1 error when passing all 8GBs, since its only 2 addresses that seemed to be screwed up.)

I want to be able to do this so I can improve the speed of the diagnosis. Otherwise the diagnosis is pretty hopeless, as my hardware warranty will probably expire before I finish it.

[ArShui]

I found that my system is giving me errors at exactly 2 memory locations with the same error bits 00010000. The 2 locations are 00098456f14 and 000983daf14.

I tried writing a test that would test a pattern I know that those two locations have failed previously, however it seems pretty random, there is no guarantee that they will fail that pattern again.

Is it possible to design a test that will find these 2 errors consistently even if I start swappng out memories? I don't want to have to wait 48hrs + everytime I do a swap for a thorough test ( I have 8GB of ram, it takes about 24hours to get 1 error when passing all 8GBs, since its only 2 addresses that seemed to be screwed up.)

I want to be able to do this so I can improve the speed of the diagnosis. Otherwise the diagnosis is pretty hopeless, as my hardware warranty will probably expire before I finish it.

If you know the memory, location that caused the error. You can press c and then 5 to see the slot that caused the error. and then try testing that slot only (removing the ram inserted at other slots)

[Wichetael]

What happens if you constrict the test range to a small range that includes those two addresses and run only the test in which you got the error? You might have the errors again quickly that way.