Overclocking Phenom II x4 970

[matthew_nicho2]

Hi all,

I have been trying to OC my 970 for a couple of days now and have had trouble getting it stable. I have looked at guides and how to's for help but have had no success. I have tried many combinations of CPU-NB, NB and CPU voltages and speeds but none have been stable. I am aiming for 4Ghz (3.5Ghz stock). I got fed up with all the failure so I set the multi up to 4Ghz with stock HT and NB clocks and volts and put the CPU volt up to 1.5V (1.520V in CPUID) I have run it for 2 hours and has been stable, which is an hour longer that the previous failures.

I will run again overnight to make sure its good but is 1.520V too high? Temps maxed at 57° and 59° using speedfan (core and core2).

I have over-clocked most the CPU's I have owned and the fails usually end with a BSOD or a freeze BUT with this one I get an error in prime95 where one core will fail and the rest carry on (usually 3 or 4). the error has always been the same FATAL ERROR: Rounding was 0.5, expected less than 0.4.

Any advice would be much appreciated, also a side question, I left quiet and cool on but it doesn't seem to clock down when idle. is that normal?

specs:

CPU : Phenom II x4 970
Heatsink : Noctua C-14
Motherboard : GA-870A-USB3
RAM : 8GB (2x4) Gskill RipJaws
PSU : Ocz moodExtreme 700W
Windows 7 64bit

 

[matthew_nicho2]

Forget about it, I think I have a less than capable chip, couldn't get 3.8Ghz stable! Going to have to admit defeat and save up for a better chip for the future.

 

[Mt Power]

I had that cpu. it over clocked to 4.2 ghz on 1.456v. 58c high with prime95 for 9 hours. it ran beautifully for a year, then didnt like the overclock anymore. any time i applied any voltage to it it just jumped up in heat. i was only able to oc it to 3.8ghz, as thats all it would go to with no voltage increase.

i still have the cpu, works fine except for the no oc. its my new backup...

 

[kiezz]

Cool n' quiet should be disabled it conflicts the overclock

 

[matthew_nicho2]

I may give it one more go with C&Q off... but I still think there is something dodgy with my chip. In CUPZ and the bios the stock voltage is 1.475V?!? I have seen people get stable at 4.0-4.3Ghz with this voltage, yet mine needs that for stock 3.5Ghz.

I will try 3.8Ghz at my stock volts and see what happens, the next step, +0.025V, will bring it up to 1.5V .

 

[kiezz]

What bus frequency and multiplier are using

 

[matthew_nicho2]

every things stock (200fsb 2000HTT/NB) and the cpu multi is 19x, testing now (4 hours in). Ive read that the NB and CPUNB needs to be faster with more volts if I ever do reach 4+Ghz, but im not sure.

 

[matthew_nicho2]

I dont believe it, its just gone to sleep! Well thats almost 5 hours without error, that will have to do... temps reached 50 so a little room left for more. disabling C&Q seems to have helped, cheers kiezz.

Will try 3.9Ghz tomorrow night, cant do it during the day because I need to use it. And this time I will make damn sure it wont go to sleep

 

[matthew_nicho2]

thought id put it to 3.9Ghz for a bit and error FATAL ERROR: Resulting sum was 2452610657575.667, expected: 2452611535361.938 within 7 minutes.

By past experience with overclocking I usually get a BSOD or a freeze but I get these errors. Would this a bus issue where I need to increase NB and CPUNB?

Will try 3.9Ghz with more volts tomorrow night.

 

[kiezz]

I'd try using a lower multiplier say 260 x 15 for 3.9ghz just from my experience with phenoms they overclock better on the bus frequency than the multiplier I actually run mine on 300 x 13 with 1.51v

 

[matthew_nicho2]

Ok, I tried 240 x 16 @ 1.472V and primed for over 9 hours no problem. I also put the NB up to 2400 @ 1.25V and it does seem to be faster than 3.8 with just the multi changed. I stuck with 240 because my ram will still be at full speed.

Will boost the multi again tonight and see what happens, cheers kiezz. Will post what I manage to get up to.

 

[matthew_nicho2]

Best answer selected by matthew_nicho2.

 

[kiezz]

cool im glad it worked out for you

 

[matthew_nicho2]

Well 3.9 failed at 1.5V, seems like too much of a voltage increase for 100Mhz. 3.8 will do.