AMD® Phenom™ II Overclocking Thread

nelcon · 22 Mar 2010 at 18:27

Just running prime at 4ghz ht link 2880 and nb 2880.Lowered the volts slightly so will see how it goes.

Martini1991 · 22 Mar 2010 at 18:50

Those timings are immense.
I can only get 8-8-8-24 stable.

gareth170 · 22 Mar 2010 at 19:17

now im getting really ****** off with this system.

i ran prime again today to double check everything is still stable at 3.8ghz. but it failed after 6hours.. it's either the cpu is dieing or it's messing me around.

i will put everything to stock then run prime then work up AGAIN . :mad:

nelcon · 22 Mar 2010 at 19:17

Stopped prime running because forgot to reset time in bios. :rolleyes:

. Memory is Corsair Dominator GT (CMG4GX3M2A2000C8) after reading the forum thread on XS regarding best memory for AMD this looks to have come out top. Although they now seem to be binning the chips for this version CMT4GX3M2A1600C6.

mmj_uk · 22 Mar 2010 at 19:22

It could just be that the ambient temperature is higher.

To be honest winter is the worst time to be testing overclocks, it's a safe bet that any 'on the limit' overclocks done in winter will fall over as soon as it gets warmer.

gareth170 · 22 Mar 2010 at 19:26

the temps are the same still, 47-49c

Big.Wayne · 22 Mar 2010 at 19:45

Gareth: don't get hacked off as that never helps the trouble-shooting? . . . Has anything changed in your system since you last did those 8 hour prime runs?

mmj_uk: Hello, can I point you towards the welcome post please!

nelcon · 22 Mar 2010 at 19:47

Just run another couple of tests, running the ht and nb at 2880 does seem faster.I always run cinebench10 first just to test the overclock before running prime.

Muel · 22 Mar 2010 at 19:58

Blimey, if you can get that stable, it's time for a 1.55v suicide run at like 4.3ghz haha, if it'll run.

See if you can get superpi 1m below sorta 16.5 secs, fastest I got was 17.3secs at 4ghz, not bad for a C2 really though.

gareth170 · 22 Mar 2010 at 20:00

Big.Wayne. nothing as been changed. but 2 days ago i tryed for 3.9ghz which failed i posted that #1086, #1088. but i put all the setting back to how they was before..

it's doing the same as before...

Big.Wayne · 22 Mar 2010 at 20:05

nelcon: jeeps that fast! :eek:

. . . I always stability test before I bench otherwise one becomes "tempted" to post suicide sceenies!

Gareth: I think the IMC is perhaps more sensitive to high load temps than the CPU? . . . while 57°C is not dangerous I do think you have exceeded the sweet spot temp wise!

. . . Deneb always seems to do better while chilly . . . what is your cooling? . . . is it still the Tru-120?

nelcon · 22 Mar 2010 at 20:12

There you go

gareth170 · 22 Mar 2010 at 20:18

Big.Wayne. at 3.9ghz i even tryed puting the cpu-nb to stock which droped the temp to 50c but it still failed...

i accept 3.9ghz is out for now

but why isn't 3.8ghz stable now, i put all the setting back to how they was before and the temp are same as before

yes im using a Tru-120 with 2 fans

eddiew · 22 Mar 2010 at 21:13

Oh man I was about to do a new screenie with proper RAM timing and temps (left the side off my case, lost 5 degrees

), then at 9 hours 7 mins Prime95 got an error

That's harsh man... real harsh...

Big.Wayne · 22 Mar 2010 at 21:35

Unlucky eddiew!

. . . . still if it makes you feel any better I had my PC BSOD after 20 hours of Prime Blend the other day!

. . . what can you do! . . . brush yourself down, make some adjustments and dive back in to the madness!

eddiew · 22 Mar 2010 at 21:49

Lol, ouch! That's a hell of a long time to go before finally bombing on you

I'm not quite sure whether to tweak the ram voltage (currently 1.66, specced for 1.65) or the CPU-NB voltage... any guesses?

liamcrane · 22 Mar 2010 at 22:03

Does prime actually fail for a reason? im gonna prime my machine on stock settings for 12hrs and i bet it will fail even tho my system is stable, i had know probs what so ever when playing high end games such as crysis constant 30fps

eddiew · 22 Mar 2010 at 22:11

It fails when a number comes back wrong. It may well be that it's often inconsequential - particularly if it's a number that's going to screen in an FPS game where you get a pixel wrong in a single frame. Or maybe it's a number being written to file, and suddenly your 150 mb photoshop image is toast

I'll pump the ram to 1.7v, put the CPU-NB up a notch, and bring tRC to 28. Hopefully that will get it either way - I know I'm not miles out, or it would have BSOD'd within seconds as it was doing this morning

Oh, and lol... case off it primes at 55 degrees, and plays world of warcraft at 41. I wonder if my PSU is failing to draw off hot air from the top of the case...

gareth170 · 23 Mar 2010 at 10:50

well here goes i've the ram back down to 1333mhz 6-6-6-20 1t which it passed 10 hours on blend before. let's see if it passes today.

eddiew · 23 Mar 2010 at 11:12

Finally got the little blighter to run for 12 hours! No speed records here, but I'm pleased by the low voltage its taken.

Also much better temps than last time - opened up the case and turned off the radiator next to it. Least I know what to focus on if I want to push for more

Alas, I just cannot get the memory to do better than 9-9-9, even at a small overvolt, with the CPU/CPU-NB on stock speed at 1.375v each. Something just ain't happy, and it's probably the penalty of cheap "optimised for intel" RAM

Ah well, doubt there's any real world performance difference anyway.

Might have to re-check the stability of the slower p-states - that NB-CPU voltage they've got is good for slack memory timings and 2200 HT, but not sure how it'll hold up with the 9s...

Also geez nelcon, 4.3 is madness xD