Originally Posted by Tunagoblin
People thinks different about what "stable" means.
That is kind of the whole reason why this thread is here, isn't it?
And I believe this thread is valuable for who seeks every kind of stability. (Thanks to munaim1
Some wants 80% stability (to me. but enough for them.) and some wants 120% stability. (like me
So if we are to make all of those people who seek stability happy, what are we going to do?
We have 3 choices.
1. No, we can't accommodate all. We want extreme stability. Nothing but that.
Since this thread IS dedicated for stability.
We're sorry for people who think that 30% of their RAM tested and 2 hours of gaming are enough to make them think their systems are stable.
I mean if you think you only use 3GB of RAM so you test only 3GB....why did you get 8GB of ram in the first place???
We only want to see the results of well tested 120% stable systems.
2. It's OK to have a list of whatever people think it's stable.
1500mb or 6500mb or whatever.
After all, we are going to use our system however the way we want.
So what I think stable IS stable.
Just give us the results so we can compare them whether it gives us BSOD with 6GB of ram used or not.
3. We can make 2 different result listings.
One for the previous rule which is... yes, the stable systems.
And the other is for "extreme stability" seekers.
But there is one problem...
munaim1 has to do all this pain in the f^@kin' a@% work.
I believe stability is stability at the end of the day, therefore with the work that it's going to take to change it around to suit everyone's need will be a lot of hard work. This is the stable thread, that's means general stability and extreme stability, both are included. I like option 1 and 2 as that makes more sense, but option 3 is toooooooo crazy
and I really don't think it's necessary but we'll see how things turn out. As my gut feeling is saying option 1, option 2 sounds better for EVERYONE, that way we can get more results in without people thinking that this club is all about OCD therapy on stress testing.
Originally Posted by turrican9
huh... what?.. now my CPU suddenly needs Internal PLL Overvoltage Enabled at x50 to boot Windows...
I knew it was too good to be true that it booted Windows at x50 + multipliers, with it disabled. So must have been some kind of a bug.
The opposite happend to me once, I thought with PLL overvoltage was only needed for high multi's but that's not the case, it's needed for particular multi's generally the high ones, so if you need it for 50, you might not needed it for 51 and again need it for 54.
I tried multi 55 with pLL overvoltage and it didn't work, I disabled it and then it was fine, now it works regardless of it being on or off, however didn't test it on multi 56 though.
Originally Posted by turrican9
I guess so, since I stated my personal opinion, and that seems to fit with the #1 option
Edit: I'm not sure if testing with 1600MB VS 90% mem is easier on the CPU, as long as RAM is stable. I just always use 6000MB for my testing. So if anyone can confirm they did a 12 hours + run at 1600MB but failed when using more RAM, even though their RAM (Overclock or stock settings) was stable in itself, that would be great. Memcontroller is built-in the CPU, so it may be stressing Sandybridge more if using more RAM. It is said, Sandybridge often overclock better (Lower Vcore with lower ammounts of RAM, like 2x2GB makes the CPU needing lesser Vcore VS 2x4GB.
should start a poll on this subject?
That sounds like a very good idea, personally I think it will still pass regardless of RAM usage, but further testing will help.
Originally Posted by Cheeba-Ace
I was having trouble passing the 1792fft (15min test with 11000mb) with my vcore in bios set at 1.385v. Running prime cpu-z would read 1.384-1.391 It would pass the test every 2 out of four runs, sometimes only making it to the 12 min. mark. Making it to the 9min mark usually was not a problem.
In order to consistantly pass that FFT I had to increase LLC to 100% (extreme in bios), previously I had been running with LLC at 75% (ultra in bios). Had to lower vcore in bios to 1.370v (1.365v made it pass the 9 min. mark though), running prime cpu-z reads 1.384v sometimes up to 1.4v.
Is it ok to have LLC at 100%? When I set it at 75% it seems the only way to pass that fft is to set vcore in bios to about 1.395v and even then I failed one test at the 12min mark. I see that a 100% LLC setting raises the vcore under load but I've tried to compensate for that with a lower vcore bios setting. Also not sure when I should consider it stable, (although I don't think I need it 27hrs stable! wow turrican9
) since if I had only run the test for a couple of minutes I would have passed easily. Most of the times it seemed as if I got a bsod early on (2-4) min mark it was not enough voltage. Seems once I could get pass 9 mins it was just a matter of tuning.
edit: sorry forgot to change desktop rez, pic too damn small again
Leave LLC on ultra high, because at Extreme (100%) it will most likely cause voltage spikes whih is not good. Also what I recommend doing is, you have your stable settings right for which you used to submit to the club, try with exactly those settings but a custom blend with the RAM. We know that you can pass a 12hour standard blend test with 1600mb, 4.8ghz @ 1.342v, try doing the same for the custom blend and higher RAM.
I think it'll pass but you never know, even if you fail it, I think your rig should still be stable with whatever you use it for.
Edited by munaim1 - 7/28/11 at 9:25am