G620 fails LinX

[CU]

I plan to test it in my main system also. I should really just test my main system anyway, since I never did.

 

[CU]

Haven't tested the individual sticks yet, but I have a question about HCI memtest. I ran it on my main system (In my Sig.) by running 4 instances testing 2048meg each. The test ran for about 5 hours. One instance had 500% coverage, one had 300%, one had 75%, and one had 0%. Am I doing something wrong? Would/Should they all show the same or over 200% to prove the ram is good? Also none of the instances reported any errors, but the coverage percentage just seem like an error to me.

 

[sm625]

Bad RAM can work fine in another system. It can also work better if you speed it up or tighten the timings. Yes you read that right! It seems there are no rules with RAM. Each stick could work fine in any given slot, only to fail when you run both of them in dual channel mode. Further, adding two more sticks could cause the original two to work fine. Ugh.

 

[CU]

Bad RAM can work fine in another system. It can also work better if you speed it up or tighten the timings. Yes you read that right! It seems there are no rules with RAM. Each stick could work fine in any given slot, only to fail when you run both of them in dual channel mode. Further, adding two more sticks could cause the original two to work fine. Ugh.

I didn't test the bad ram from the new system in my main system. I just test the ram that has always been in my main system just to test it. My question is on how to use HCI memtest and what should the results be?

 

[john3850]

I was under the impersion that HCI memtest was to be used to locate errors in the cpu cache.
I use memtest 86 for just the system memory
Since 2500K uses ecc in the cache I am not sure if it worth using HCI memtest anymore.