Sign in with
Sign up | Sign in
Your question
Solved

Questions about Prime 95

Last response: in Overclocking
Share
a b K Overclocking
a c 152 à CPUs
November 6, 2011 3:57:55 AM

Ok I'm a little confused with Prime 95. I was fooling around with it tonight to help figure it out before I went and started to really overclock my cpu. The problem is I don't get how you know when its done doing its test. To start with I just wanted to do something simple not go nuts so I set the stress time for only one minute. 1 minute comes and goes and I expect to see a pop up saying test complete but after one minute I looked at the cpu status bar it's still running at 100% maxed out. Also for overclocking the cpu what is the best test to use small FFT's, in place FFT's, blend or custom? If you go custom what do you set the FFT's to? Then you have advanced setting like ECM and P-1 what is all that and how important is that? This is the first time I'm using Prime 95 so I just want to learn the program so when I do decide to really overclock I know what I'm doing, thats why I just wanted to start off very small first.

More about : questions prime

a b K Overclocking
a b à CPUs
November 6, 2011 5:58:27 AM

First there are lots of different releases of P95 hanging around. You can download the latest P95 (and other useful OCing and stability testing utilities) here:

http://www.newegg.com/Product/Product.aspx?Item=N82E168...

I use the "Blend" test. Once started, it will run until you click "Stop Test".

After a new build, I'll use it to test heat control and stability. I run the test for about 30 min, and then again combined with Furmark Torture Test for 5 to 10 minutes. Temps are monitored using CPUID's Hardware Monitor. After stability testing, I'll finish loading the HD and then let P95 run overnight as a final test.

If P95 fails (eg, Rounding Error), it is usually because of a memory issue.

For overclocking I will use a mix of 10 minute and longer runs depending on how the OCing session is going. Once I think I've found an OC I'm comfortable with I may run P95 for 30 min to make sure heat buildup is handled, then use the PC, then run P95 overnight.

Hope this helps.
m
0
l
a c 197 K Overclocking
a c 172 à CPUs
November 6, 2011 12:48:09 PM

I do things a little differently. I start off with short runs of small fft's to check load temps. Then when I'm ready for final testing, I start with a 24 hour run of small fft's to test the CPU. When that passes, I finish with a 24 hour blend test to test the memory.

But then, I am a little anal about stability.
m
0
l
Related resources
a c 158 K Overclocking
a c 218 à CPUs
November 6, 2011 2:22:58 PM

rds1220 said:
Ok I'm a little confused with Prime 95. I was fooling around with it tonight to help figure it out before I went and started to really overclock my cpu. The problem is I don't get how you know when its done doing its test. To start with I just wanted to do something simple not go nuts so I set the stress time for only one minute. 1 minute comes and goes and I expect to see a pop up saying test complete but after one minute I looked at the cpu status bar it's still running at 100% maxed out.


You seem better suited to OCCT .... 1 minute doesn't test anything nor allow the heat to reach a stable setting. OCCT runs 60 minute tests during which it monitors temps and voltage and displays graphs of them at the end for your analysis.

http://www.ocbase.com/perestroika_en/index.php?Download

It's only the highest stable OC I get w/ OCCT that I test w/ Prime 95 .... overnight

rds1220 said:
Also for overclocking the cpu what is the best test to use small FFT's, in place FFT's, blend or custom? If you go custom what do you set the FFT's to? Then you have advanced setting like ECM and P-1 what is all that and how important is that? This is the first time I'm using Prime 95 so I just want to learn the program so when I do decide to really overclock I know what I'm doing, thats why I just wanted to start off very small first.


I run one test @ blend for memory testing and small FFT's for the CPU

m
0
l
a b K Overclocking
a c 152 à CPUs
November 6, 2011 4:09:55 PM

Jack,
I was just fooling around with it to learn the software I know 1 minute won't really prove anything, I wasn't trying to. I didn't even have the CPU overclocked at all. I just downloaded Prime 95 then HWmonitor to just learn how they work together when overclocking. Right now I'm not ready to overclock yet I'm still reading and learning about how to do it, it will probably be awhile plus I'm waiting to do my water loop before I overclock. You guys brought up some good points that have led to a few things I just thought of:

1) between each time you raise the BCLK how long should I run Prime 95 for to get the best idea of if I'm getting stable temps?

2) If something is not stable how will I know, where will I get an error message and what will it say.

m
0
l

Best solution

a b K Overclocking
a b à CPUs
November 6, 2011 4:51:57 PM

Assuming you're talking about your i5 2500k, you might not want to play with BCLK - just the multiplier, and then the voltages, etc, necessary to gain stability. BCLK gains available are tiny and the resulting changes are consequential.

Failed overclocks can show up in a lot of ways - here are *some*: failure to post or boot into Windows; BSOD (or restart, depending on your Windows settings) while booting or while in Windows; freezing, restarting, or shutting down in use.

This is why you use standard test programs like Prime95, Furmark, et al to test stability and not, eg, a game. If they ran on the system at stock, they should run after an OC. If they do not, the OC is at fault, and not some hidden bug or incompatability between a game and your configuration.

How long to run P95? The safest answer for me to give is "as long as possible". The practical answer is different lol. I run P95 long enough to ensure I know the temperature at which the CPU becomes stable, perhaps 10 min. That normally is as far as you will practically want to go in between OCs.

Since that's not long enough to really test stability, you may later find something you thought was stable, but isn't. You then either adjust parameters or take a step back in OC. This is "normal".

Once I reach a temperature above which I do not want to operate on a 24/7 basis, I'll treat the PC as if it were a new build, finishing with an overnight P95 run. That's still no guarantee - as JSC said, he doesn't call it stable until its passed the 24-hour test.

PS: Even after deciding I have a stable OC, I will keep CPUID's Hardware Monitor running for many days, and check it periodically and before shutting down. That way I can see any unexpected heat spikes.
Share
a b K Overclocking
a c 152 à CPUs
November 6, 2011 5:11:32 PM

O ok. I was reading this thread created by Mrface here on the forums and I was under the impression that the BCLK was how you mutiplied the core clock to increase frequency. I read that AMD still uses the old FSB to increase the multiplier but the new Intels did away with that and now use the BCLK to increase the CPU frequency.
m
0
l
a b K Overclocking
a c 152 à CPUs
November 6, 2011 5:11:52 PM

Best answer selected by rds1220.
m
0
l
!