Overclocking AMD II x4 Phenom 945

[BeepBeep2]

Not sure about that. From your link, if I click on the link for v.2.0, copy and paste the URL for v.2.0 page into a document and then click on it it takes me to the page for v.2.0 not v. 2.1. Perhaps I am misunderstanding what you are saying. At any rate, it's important that OP be directed to the right information relative to the product he is using in case he needs to download files.

What I'm saying is that if you click on the board (no revision number listed) on the Gigabyte website, the page displayed will be for the latest revision board. Same thing from google.
The Product ID changes in the URL when you click the revision number you need, but the only way to get those URLs is to go to the board page and click the revision number you need.

Anyway, it was posted before the OP came back with the revision number and I can't see through the internet so no big deal.

 

[trents]

Yeah, no big deal. I guess we beat that one to death, huh?

 

[BeepBeep2]

Yeah, no big deal. I guess we beat that one to death, huh?

A little, will try not to do that again.

Baaaaccckkkk on topic now...

 

[3dwa]

Something went wrong, now pc doesn't load at all, repeating 1 long beep what to do now?


Ok, I think something was wrong with volts, bios reseted, seems good for now

 

[soulwatch5]

There should be a way to clear your CMOS on the motherboard. Usually its placing the jumper in a specific position and leaving it there a couple of seconds before returning the jumper. If worse comes to worse u can always remove the battery on the motherboard and make sure there isn't any power going to the mobo. Even doing this though you have to have the battery removed for a couple of minutes (If i remember they say like 15 or so) because even without the battery, capacitors on the motherboard can hold a charge. Doing this should reset bios to defaults and reverse your oc.

 

[3dwa]

Removed the battery for a while, seems good now.

 

[3dwa]

So I found out my max fbs is 260, so I chose 240 for beginning. Turned up CPU voltage by 0.15 to 1.45v. When doing prime test first and 3rd cores get error in first minute: rounding was 0.5, expected less than 0.4. What's the problem?

Now reseted all changes in ram bios, seems good for first 5 minutes in prime95

 

[trents]

The thing we need for you to do not is to post back with pics of HWMonitor which reflects temps during your last run and pics of CPU-z tabs: CPU, Memory and SPD. In other words, we need to see what your CPU socket and core temps were under load and what your voltages and frequencies are for the CPU and the ram. A pid is still worth a 1000 words. Perhaps we can spot the likely reason or reasons likely for your instability.

 

[trents]

The other thing you need to do now is create a "Sig" so that details of your system travel with every post you make. This is forum etiquite and very helpful to those helping you, especially as the thread gets long and the infromation you posted early on gets lost in the pile. To create a Sig, click on Quick Links and Edit Signature at the top of the page. You can refer to mine if you want guidance as to what kind of info we like to see.

 

[trents]

If all you've changed is the fsb and the CPU vcore it is likely your ram frequency has gotten too high and maybe also your HT Link and NB (aka, CPUNB) frequencies have gotten too high as well. Remember, all those things climb with the FSB.

The strategy I recommend is:

1. to start the ram at 666 instead of 800, the HT Link at 1800 (9x), the fsb at 205mhz and leave the CPU vcore at stock to begin with. Let's see how far you can gert on stock vcore.
2. Run a 20 minute Prime05 blend test with HWMonitor open on the desktop.
3. If you pass the Primetes, jack up the fsb to 210 and retest.
4. Repeat this until you fail the Prijme test and then add .025 vcore and retest with Prime95. Add more vcore if necessary to pass the test.
5. Repeat steps 1-4 until you cannot pass the test anymore or until your core temps begin to exceed 55c. 55c is approximately when we notice temp induced instability.
6. At that point, post back with pics of HWMonitor and CPU-z tabs: CPU, Memory and SPD from your last failed test and we will advise further.

Adjust the HT Link frequency as necessary to keep it between 1800-2000. BeepBeep2 and I disagree as to the need to keep the HT Link frequency at stock or lower on the Deneb CPUs but it is still my recommendation.

When you seem to have hit the wall and tentatively foun the max overclock for the CPU then you will need to do a longer Prime95 blend test - several hours - to confirm stabilit. The 20 minute Prime test is time efficient during the initial overclock phase and guarantees that you are close to stable but is not enough for confirmation.

This method is rather tedious but it's safe and has the advantage of helping you zero in on the cause of instability because you are not making a lot of change at one time.

 

[BeepBeep2]

Rounding error is memory subsystem, either CPUNB or DRAM