Quote Originally Posted by Leeghoofd View Post
@n7 plz fill in the template and post your settings plz...
I'm heading out to work in a bit & am running HCI right now, so i'll fill them out next reboot, tonight hopefully.

I think i may have figured out the trick to get 8 GB "stable" with HCI Memtest though, & it's not making me happy at all.


I always try to test as much of my 8 GB as possible with HCI, & Windows limits me to usually around 1450 MB per instance, which usually means i run 4x 1400 MB, etc.

I remember asking you to run HCI with your 8 GB, which you did, with 4x 1000 MB, since you said that Windows wouldn't let you allocate more per instance.

Here's the ed up thing.

When i run 4x 1400 MB HCI Memtest, it will fail @ DDR2-1050 5-5-5-15, PL 10, tRFC 60, & every single subtiming manually loosened by a couple numbers.

But then i remember your screenshot with only 4 GB being stressed.

I run 4x 1000 MB, & it makes it past 100%.
Then i try DDR2-1080 5-5-5-15, & it makes it past 100%.

The thing that annoys me about this is that it almost seems that if you stress only 4 GB of you 8 GB, suddenly for me, i'm getting "stable" results where when i stress 5600 MB of 8 GB, it's not stable.

jVIDEA, exactly what config of HCI Memtest are you running?
How many instances, & set to how much RAM?

I'm starting to wonder just how "stable" people's HCI Memtest results are if they're only stressing half the RAM.

From what i've seen in my own testing, it was basically impossible to get 8 GB stable in HCI on my DFI.

It seems the same issue on my P5Q except that now i've discovered cheating & only stressing 4 GBs worth "passes".

I'll post screenshots, & the template after work, but i expect that what i ran P95 Blend for all night "stable" 7 hrs. is going to fail sometime during the day in 4x 1400 MB of HCI.