Got my new cooler, unstable nonetheless.

[skpwn]

I've done this and my PC booted into Windows just now with a Multiplier of 20x.
Thus resulting in 4GHz.
How should I continue to increase the Multiplier and voltages?
Every .5x 1 step in the CPUNB upwards?

Every time you change the settings and boot into windows you need to prime for at least 20 minutes. If you make a 20 minute run, you can guess that it is fairly stable, but you will not know if it is 100% stable until you've done a 2 hour + run.

So now that you're at 4 GHZ, you need to run Prime for 20 minutes and post the results.

 

[akatsui]

I'd start with a 20 minute prime95 blend test to see if your 4ghz is stable for at least that long before trying to go any higher.

I was planning on doing this with OCCT [linpack] as I've read that P95 is less... powerful?

 

[gnusounduave]

I was planning on doing this with OCCT [linpack] as I've read that P95 is less... powerful?

Seems that the consensus around here is P95. IIRC OCCT was built off of P95,to an extent, so I don't know if P95 is "less powerful" but I'll defer that to somebody else that could clarify better than I can.

 

[akatsui]

Seems that the consensus around here is P95. IIRC OCCT was built off of P95,to an extent, so I don't know if P95 is "less powerful" but I'll defer that to somebody else that could clarify better than I can.

Alright thanks, It just seems proper logic to me that when software is built off the same base it would be better, if there's a dedicated team behind it, like OCCT.
Anyway, here are the results!

It was a 20 minute run using 90% of my RAM and a Linpack burnin run, as you can see, Temperatures do not reach 50c, there's room for a few more MHz's.
My target is 4.2GHz.

 

[gnusounduave]

That's just giving us temps and voltages. What we are looking for is how does it handle the mersenne prime numbers and are the cores stable with the current voltages you are using.

 

[akatsui]

That's just giving us temps and voltages. What we are looking for is how does it handle the mersenne prime numbers and are the cores stable with the current voltages you are using.

Oooh, you wanted a P95 log in particular, sure, let me run it for 20 minutes, I'll post it when I get it.

 

[Bosco]

Check your psu voltage if its randomly freezing up and no bsod.

 

[gnusounduave]

Oooh, you wanted a P95 log in particular, sure, let me run it for 20 minutes, I'll post it when I get it.

That's what we're lookin for

 

[Bosco]

Ah, you younguns with yer fancy tests....I just monitor voltages as I'm gaming... Hit the windows key on me keyboard and look at my taskbar.

 

[akatsui]

That's what we're lookin for

Haha, I love that Gif and will make it my avatar.
EDIT: GIF's Don't play as avatar.. time to screen it!
Anyhow, OCCT Linpack and IntelBurnInTest also based on Linpack work perfectly fine, but Prime95 crashes my system about 15minutes in.
Temperatures did not rise above 45c.

[Wed Apr 18 11:16:04 2012]
Self-test 800K passed!
Self-test 800K passed!
Self-test 800K passed!
Self-test 800K passed!
Self-test 800K passed!
Self-test 800K passed!

There is no error log either.

Prime95 doesn't exactly crash, it freezes up my system, I have to restart it, there are no BSOD's either.

 

[gnusounduave]

Anyhow, OCCT Linpack and IntelBurnInTest also based on Linpack work perfectly fine, but Prime95 crashes my system about 15minutes in.
Temperatures did not rise above 45c.

There is no error log either.

Prime95 doesn't exactly crash, it freezes up my system, I have to restart it, there are no BSOD's either.

Give it a bit more vcore then try for another 20 minute run. Try it at 1.4vcore and move up in very small amounts untill you can at least get 20 minutes.

 

[skpwn]

Ah, you younguns with yer fancy tests....I just monitor voltages as I'm gaming... Hit the windows key on me keyboard and look at my taskbar.

Monitoring voltages != stable cores.

Give it a bit more vcore then try for another 20 minute run. Try it at 1.4vcore and move up in very small amounts untill you can at least get 20 minutes.

Do this. If Prime is locking up, something isn't stavle. You're gonna need more vcore!

 

[akatsui]

That's just giving us temps and voltages. What we are looking for is how does it handle the mersenne prime numbers and are the cores stable with the current voltages you are using.

Where can I find those numbers you ask for?
Running P95 for about 12 minutes now, still stable, good temps, etc.
I set my vCore to 1,4 as you guys said.

Do this. If Prime is locking up, something isn't stavle. You're gonna need more vcore!

Alright, so anytime I upgrade my multiplier by .5x and it becomes unstable, I take a step up on vcore untill it becomes stable again?

 

[skpwn]

Alright, so anytime I upgrade my multiplier by .5x and it becomes unstable, I take a step up on vcore untill it becomes stable again?

You have two choices:

1) Pick a vcore you are comfortable with and start upping the multiplier, while running prime, until you find the most stable build.

2) Stutter step your way up (more time consuming). You will go to 4.1ghz, set your vcore to whatever, then prime it. If it doesn't blend for at least 20 minutes, then you'll need to go set your vcore one step up, until you find your stable vcore at 4.1. Now you go for 4.2, give the vcore a nudge, and try again.

Every time you up the multiplier, you are more than likely going to have to up the vcore as well. For example, my FX-6200 will run 4.1 at 1.38 vcore, but needs 1.40 vcore for 4.2ghz. 4.4 ghz is all the way up to 1.48 vcore...notice how large the jump is? (Note: these vcores will probably not work for you, please experiment).

Some times, 20 minutes isn't enough to see if you're truly stable. You could be an hour and a half into a blend and have a core die...so that tells you that you need more vcore for the chip to be stable.

 

[gnusounduave]

Where can I find those numbers you ask for?
Running P95 for about 12 minutes now, still stable, good temps, etc.
I set my vCore to 1,4 as you guys said.

You'll know when a core fails on P95. You can either attach a screen shot or simply say "after 10 minutes core 6 died" or whatever.

Alright, so anytime I upgrade my multiplier by .5x and it becomes unstable, I take a step up on vcore untill it becomes stable again?

You are correct, sir.

You have two choices:

1) Pick a vcore you are comfortable with and start upping the multiplier, while running prime, until you find the most stable build.

That's solid advice right there. If you don't want to go over, say 1.48 vcore, set your vcore to that and then slowly start upping the multiplier until you reach equilibrium between multiplier and vcore.

 

[akatsui]

Some times, 20 minutes isn't enough to see if you're truly stable. You could be an hour and a half into a blend and have a core die...so that tells you that you need more vcore for the chip to be stable.

Well, that's the thing, 4GHz on 1,4v didn't work, I upped it 1 step again.
Also, in P95, core's don't die out, my system just freezes up.
And also, it only happens on Prime95, not on any other burn in test...

 

[gnusounduave]

Prime isn't a burn in test per se, as it's testing for stability. However if your system is hanging up then indeed add more voltage. Go 1.41 and then 1.42 as i'm sure you'll find something that's good for you around there.

 

[akatsui]

Prime isn't a burn in test per se, as it's testing for stability. However if your system is hanging up then indeed add more voltage. Go 1.41 and then 1.42 as i'm sure you'll find something that's good for you around there.

Are there other stability tests comparable to P95? Sometimes it even froze on 3,7GHz with 1,3875... while it's been stable on 3,9GHz with that voltage since my new cooler.
temps weren't outrageously high with my old cooler.
How about SuperPi? or HyperPi? what's the difference even...

 

[gnusounduave]

I can't answer that question. I'd have to defer.

 

[akatsui]

I can't answer that question. I'd have to defer.

Well... Time to run P95 again...