Anyone seen this behavior on LinX?

[Compddd]

If I set my VTT to +75mv and do a 5 pass run on LinX, it takes 20 minutes and my max GFLOPs are 52

If I set my VTT to +25mv and do a 5 pass run on LinX, it takes 28 minutes and my max GFLOps are 48

Any idea what is going on here?

 

[mastrdrver]

Yes, the +25mv more than likely are not enough and if you run longer than a 5 pass, it will error out eventually.

For me, 25 pass is minimum for LinX. 100+ pass is preferred cause I thought I had one stable once but errored out at about pas 75. Or....you could just run 25 and if pass, up the voltage a notch or so to add buffer and to also see if there is change in the number.

 

[Compddd]

I settled on +50mv

Just did a 50 pass run in LinX, no errors and every pass was 52 GFLOPs

 

[Compddd]

Ok this doesn't make any sense. How come I can pass a 50 LinX run with 52 GFLOPs each run, and then the next day when I run the same test, no settings changed, I'm only getting 45 GFLOPs and it's freezing to a black screen?

 

[RogerTheEngineer]

This is why these new quick tests aren't all they're cracked up to be. Yes, they will usually find a problem quicker than the old way (Memtest and Prime95), but they still won't guarantee complete stability unless you run them for a long time. Run it over night. Run it for days. Run it and prime at the same time. Run it with furmark or 3dMark looping. If you really want to know if you're completely stable, a few runs of IBT or LinX isn't enough.

 

[Compddd]

50 runs of LinX is quite a lot