In-Place vs Blend, 1 pass 1 fail

Hi, I've been using Prime95 for a while and this is the first time this problem has come up.
I ran the torture "In-Place large FFTs" and it ran for a few hours. I then ran the "Blend" test, and it failed at test 1. It happened a few times so I ran the In-place test again right after the Blend failure and during the same Test1 calculation (4000 Luc-Leh M19922945 using 1024K...) the In-place passed. Is there any reason this might happen besides the "hardware failure" error from the Blend test? Thanks!

PS: I tried different CPU OC settings to try to fix the error, even ran at stock speed/voltage, error still came up. No other changes.

[Mini-Geek]

Originally Posted by Unregistered

Hi, I've been using Prime95 for a while and this is the first time this problem has come up.
I ran the torture "In-Place large FFTs" and it ran for a few hours. I then ran the "Blend" test, and it failed at test 1. It happened a few times so I ran the In-place test again right after the Blend failure and during the same Test1 calculation (4000 Luc-Leh M19922945 using 1024K...) the In-place passed. Is there any reason this might happen besides the "hardware failure" error from the Blend test? Thanks!

PS: I tried different CPU OC settings to try to fix the error, even ran at stock speed/voltage, error still came up. No other changes.

Considering the main difference is the amount of RAM used, I'd say you have some problem with your RAM. You could try lowering the RAM's OC or changing its settings or something.

Hi,
Thanks, I just tried that, but it didn't seem to work. The memory isnt OCed, its set at stock. The blend test is also failing at the same point, test 1 of the 4000 iterations. Weird.

Hey, thanks, I think you're right, but it wasn't the settings, I think it was a bad RAM stick. After troubleshooting with memtest it looks like I have one good and one bad stick. Thanks for your help! Now time for RMA...