how to test stability of cool n quiet?

[shamans]

how does one test stability of cool n quiet? I need a prime95 or something similar to use x% of cpu instead of using 50% or 100% ;(.

Anyone know of any?

 

[opmike]

I'm not quite sure I understand your question. I don't see how the 100% (or very close to it) programs like Prime 95, Super Pi, CPU-Burn, Rosetta, etc. would be ineffecient at testing stability

Are you looking for some sort of selective CPU useage percentage?

 

[myocardia]

Want to test the stability of your CnQ? Start web browsing. Open up about 10 or 15 IE windows. That's usually enough to show problems, if any exist.

 

[shamans]

Originally posted by: opmike
I'm not quite sure I understand your question. I don't see how the 100% (or very close to it) programs like Prime 95, Super Pi, CPU-Burn, Rosetta, etc. would be ineffecient at testing stability

Are you looking for some sort of selective CPU useage percentage?

Yes, that's sort of what I'm looking for. Amd's cool n quiet raises and lowers multiplier and vcore depending on how much of the cpu u use. So I need a program that does something like "limit up to x# of calculations per second", not a x% of cpu as I said before...

It would be nice if it tested each pair of multiplier/vcore states that it uses.

Originally posted by: myocardia
Want to test the stability of your CnQ? Start web browsing. Open up about 10 or 15 IE windows. That's usually enough to show problems, if any exist.

Not a reliable enough test...are you telling me that no crash implies reliable?

 

[myocardia]

Originally posted by: shamans

Originally posted by: myocardia
Want to test the stability of your CnQ? Start web browsing. Open up about 10 or 15 IE windows. That's usually enough to show problems, if any exist.

Not a reliable enough test...are you telling me that no crash implies reliable?

No, I'm telling you that, at least with this system, you'll usually get a memory dump, if you aren't giving the cpu enough voltage, while CnQ is running.

 

[Unkno]

if you want complete stability, record the voltage and the multiplier of the cpu when CnQ is running. Then turn off CnQ, and set the voltage and multiplier that you recorded down earlier. Now, stress your system with prime95 for roughly 16 hours (24 hours for peace of mind). Finally, if it passed prime95, set it back to the original voltage, multiplier, and turn on CnQ, then do what myocardia said to test that your system is able to scale up/down the multipler and the voltage without any problems. Btw, if you want to, you can try myocardia's method first to save the time from testing.

 

[shamans]

Originally posted by: Unkno
if you want complete stability, record the voltage and the multiplier of the cpu when CnQ is running. Then turn off CnQ, and set the voltage and multiplier that you recorded down earlier. Now, stress your system with prime95 for roughly 16 hours (24 hours for peace of mind). Finally, if it passed prime95, set it back to the original voltage, multiplier, and turn on CnQ, then do what myocardia said to test that your system is able to scale up/down the multipler and the voltage without any problems. Btw, if you want to, you can try myocardia's method first to save the time from testing.

hmm, well opening/closing 15 to 20 ie windows does fluctuate the cpu requirements...lol

And I'm too lazy to record every CnQ state ;( lol and test every single one of them. I think there's like.....4 to 5....Someone must've thought of making a proggie to test this

 

[krotchy]

The whole point of CnQ is that as soon as anything is running that requires any processing power, it raises your vcore and multiplier. Programming something that keeps it from jumping up but still verifying your calculations would be very tricky, and not all the necessary. If your stable with dual prime running CNQ, your gonna be stable no matter what. Since anytime something intensive is happening, your running at your high multiplier and vcore, not your low.

 

[shamans]

Originally posted by: krotchy
The whole point of CnQ is that as soon as anything is running that requires any processing power, it raises your vcore and multiplier. Programming something that keeps it from jumping up but still verifying your calculations would be very tricky, and not all the necessary. If your stable with dual prime running CNQ, your gonna be stable no matter what. Since anytime something intensive is happening, your running at your high multiplier and vcore, not your low.

Yah, I was doing some tests with this and noticed that there are 2 permanent states. When I mean permanent, I mean states that CnQ switches to with any longetivity.

On my x2 3800 (with 240 fsb), it was...
5x multiplier, 1.1v
10x multiplier (stock), 1.35v (stock)

There were some interim states but they were just for switching between the 2 permanent states. I tried running multiple instances of classic media player (sort of like running multiple quicktime players) and noticed that I couldn't get it to stick on an interim state. It'd just switch between the 2 permanent states by using interim states...

It looks difficult to test the interim states (And see if they're stable) besides starting/stopping prime (assuming that doing this tests CnQ interim states in the very beginning as CnQ ramps up my cpu). There were a lot of interim states, hard to test them all by hand. (sort of like a bruteforce approach).

But the permanent states are very testable (since there's only 2 of them).

As far as I can tell, my system seems completely stable from my tests.

 

[avi85]

use RMclock and set only one P-state transition and set it to the voltage you want to test it with, run prime 95 and tada!

BTW I would run Prime95 for only like half an hour or less cause in a real world situation if it actually does end up using 100% of the lowered frequency it will bump up the voltage and multiplier to the normal one

 

[Unkno]

Originally posted by: krotchy
If your stable with dual prime running CNQ, your gonna be stable no matter what.

WRONG, if a system is stable running CnQ, that means the system is stable only at max performance. That doesn't mean it is stable at 5x multiplier at 1.1V. Remember, a system can still crash at idle, it doesn't have to be at load.


to shamans: when i posted about testing the system, i meant the settings during idle (kinda impossible to test every state). So, according to your settings, turn off CnQ and keep your fsb at 240, set multiplier to 5x, and set the voltage to 1.1. This test the majority of the setup and the chances another multiplier (say 9x) crashing is slim.

 

[GuitarDaddy]

Manually set the minimum multi and volatage used by CnQ in bios, then test with Prime,memtest,etc...


/thread

 

[Unkno]

Originally posted by: GuitarDaddy
Manually set the minimum multi and volatage used by CnQ in bios, then test with Prime,memtest,etc...


/thread

exactly what i said earlier....

 

[Pederv]

If my system is on the edge of not working with CnQ, the system hangs during a soft reboot. That's how I know to give it a little more juice. My BIOS supports increasing the CPU voltage by %, so my voltage gets a little bump when running in low power mode.