AMD® Athlon™ II Overclocking Thread

HeX · 1 Jan 2011 at 14:43

I usually do about 20 loops of IBT/LinX, on about 20000 problem size.

If that all passes fine, then go for a long Prime 95 Blend run.

Just ordered myself an X3 450, so will see what I can get from it next week

Frozennova · 1 Jan 2011 at 17:45

Im still waiting for my 235e to be shipped from the seller on the bay. By looking at Hwbot 4ghz is fairly easy on them, its only going to be a chip to bench with. Im looking at getting some dry ice in a few weeks time to really push it.

Pulseammo · 2 Jan 2011 at 14:25

HeX said:
I usually do about 20 loops of IBT/LinX, on about 20000 problem size.

Thanks for the tips, good luck with your x3 450

Decided to do 20 loops of IBT and it failed at 1.375v. Increased voltage slowly to about 1.41v in the bios to achieve 20 loops:

IBT definitely seems to draw out an unstable overclock faster than prime blend for me - blend was able to last all night yet IBT can't last for 40 minutes!

Anyway, I think I'm stable now at 3.5ghz at 1.41vcore with my X4 630. Temps on the cores show as 50c in HW monitor, TMPIN0 goes between 50c and 55c.

I'm pretty pleased so far and not entirely worried about the temps since other apps only bring it to 45c - I may fiddle with my smart fan settings now.

HeX · 10 Jan 2011 at 20:18

Just got my X3 450, 4th core is a no go, boots but bluescreens on windows load.

Also having a mare trying to clock it, and i've a feeling it's to do with me having 8GB of memory.

At stock it's fine, it'll IBT/LinX/Prime no fuss.

But any OC and it will not error in IBT/LinX but the time and GFlops are all over the place.

I've been trying to get 3.5GHZ, 250FSB, 2000HT, 2000CPU-NB stable, i've tried everything from 1.375-1.45V on CPU, and 1.2-1.3V on the CPU-NB, and nada.

It's infuriating... :mad:

Going to be a faff this one it seems

Scon · 10 Jan 2011 at 20:26

@Bigwayne, if you're up for it a table of the fastest clocks from which forum member would be cool, with cpu screenshot only of course.

HeX · 10 Jan 2011 at 20:43

Well, 3.25GHz, 250FSB, 2000HT, 2000CPU-NB, works at stock volts...

EDIT -- No it doesn't it fails after 45mins... hmmmm... wonder if the 8GB is limiting the FSB I can hit... time to work up slowly...

joem1983 · 11 Jan 2011 at 18:07

Hi guys, been following this thread. Just finished building my new system and i think this thread probably persuaded me to go with the AMD so cheers.

How much load temp is too much?

running an Athlon II 640 at 3.6ghz and 1.5~ v core. using an Artic cooler 7 (not the best but... its cheap!)
Board is an ASUS m4a88t
Running around 90% load while rendering i was getting like 65 degrees. Think its too much perhaps.

Anyway, some pic

Pulseammo · 11 Jan 2011 at 18:16

@HeX That's sad to hear the X3 isn't going so well

. Did you give it a try with only 2 sticks of ram to see if it is that which is holding it back?

@joem Your cores are only at 55 degrees there, I'm not sure which to trust - the CPU or the Cores temperature but I've read that performance on the type of silicon AMD use degrades after 55c so you may wish to do a benchmark to see if that is the case.

AMD list the max temperature as 71 degrees so 65 degrees would be on the high side for my liking personally. Others may say different though. Have you tried getting to 3.6ghz without the 1.5vcore?

HeX · 11 Jan 2011 at 18:55

Ah ha, breakthrough, apparently my CPU (or mobo not sure which), is fine with my 8GB @ 1600MHz at stock, I did a 6hr Prime Blend, and a 20 loop LinX test no fuss, however to get it to do any FSB clocking at all I have to run my memory at 1333MHz instead.

I've just dropped the mem multi down to 1333MHz and high timings, bunged FSB to 250, kept HT and CPU-NB on stock for now (to rule them out), and ta da straight to 3.5GHz stable @ 1.375V (in BIOS, not tried to lower it yet just wanted to make sure it had enough juice).

So now I know for certain it was the memory holding it back, as I tried the exact same settings last night, but with the memory at 1600MHz and it wouldn't pass a single loop of LinX.

Now I have a starting point I can work the clock and other settings to something more impressive

Yay \o/

Pulseammo · 11 Jan 2011 at 19:54

@HeX Great news

Had you worked out the ratio wrongly perhaps and the memory was running faster than 1600mhz maybe?

Also, did you try a voltage increase to stabalise the unlocked core to stop it from BSODing, sometimes it just needs a wee bit of extra juice to get it happy.

HeX · 12 Jan 2011 at 08:34

Cor' this thing is driving me nuts

So, here's the deal.

STABLE:

3200MHz, 1.300V, 200FSB, 2000HT, 2000CPU-NB, Stock Volts, 1600MHz C9-9-9-24

3500MHz, 1.375V, 250FSB, 2000HT, 2000CPU-NB, Stock Volts, 1333MHz C9-9-9-24

Those are the only two configs I can seem to get stable. The first one being stock clocks and settings for all components.

The second gets me a 300MHz OC but lower mem speed, and is multiple LinX/IBT loops stable.

Using the 3500MHz clock I dropped the mem timings down to 7-7-7-20 31 (from 9-9-9-24 41), and that passed two full loops of Memtest86+ without problem.

So the mem is obviously stable, so you would assume so is the mem-controller.

However it won't pass LinX/IBT. It doesn't error but again the timing and GFlops are all over the place.

Same goes for the CPU-NB, at 2000MHz it will pass flawlessly, at 2.25Ghz or 2.5GHz it fails LinX/IBT. Even giving it extra CPU-NB from 1.25-1.35V does nothing to make it any better. In fact the higher CPU-NB volts seem to cause it to error quicker.

Hmmmmmmm...

If I could get 3.5 and 1333 at CAS7 stable i'd be happy! 2500CPU-NB would be nice, but I can live without!

SeraphX · 12 Jan 2011 at 09:00

joem1983 said:
Hi guys, been following this thread. Just finished building my new system and i think this thread probably persuaded me to go with the AMD so cheers.

How much load temp is too much?

running an Athlon II 640 at 3.6ghz and 1.5~ v core. using an Artic cooler 7 (not the best but... its cheap!)
Board is an ASUS m4a88t
Running around 90% load while rendering i was getting like 65 degrees. Think its too much perhaps.

Anyway, some pic

SNIP

i'm pretty sure the loads temps should be under 70c
upgrade the cooler too, i swaped my Freezer 7 pro for a titan fenrir and it made loads of difference,
Idles at 30 now, and full loads at around 57-60c

also try bringing your Vcore down a little more, you should be able to get 3.6Ghz at 1.475 or there abouts
The Freezer 7 pro really shows it's simplicity when ya turn the volts up

joem1983 · 12 Jan 2011 at 17:44

Yeah, I've dropped the voltage to 1.43

Seems to be just over 50c under load.

I need to run some stability tests but the video edit rendering wasnt pushing 90% and I dont have any games to play on this yet.

The arctic is just a make do. Hope to get some fancy Corsair liquid cpu cooler then I can push for 3.7+. But then again 3.6 will do. will leave silliness for when i get a 6 core

HeX · 12 Jan 2011 at 19:47

This is the kind of weirdness I'm getting.

Hour of Prime Blend no fuss.

2 full loops of Memtest86+, no problem.

3mins of LinX/IBT... random time/gflops, but no error... :confused:

I'm at a loss...

Thejay · 12 Jan 2011 at 19:55

Heres mine.... I do normally let it run for much longer to test stability and it is 100% stable I just ran it to heat things up to take the screenshot, really wish I could change the cpu multiplyer rather than keep upping the bus speed so much

Using DDR2 memory and Titan Fenrir cooler.....

Uriel · 12 Jan 2011 at 20:19

HeX said:
This is the kind of weirdness I'm getting.

Hour of Prime Blend no fuss.

2 full loops of Memtest86+, no problem.

3mins of LinX/IBT... random time/gflops, but no error...

I'm at a loss...

If it's any consolation, HeX, my unlocked Athlon X2 5200+ just made it through 24 hour Prime Blend with a 940 odd MHz overclock so I was confident that it was stable (http://forums.overclockers.co.uk/showthread.php?t=18225299&page=2).

IBT brought it down in minutes. It just tests different aspects of the CPU. In my case pushing up vCore a notch seems to have stabilised it. Just made it through 20x IBT on standard. Testing 20x on high now.

What are your temperatures like by the way?

HeX · 12 Jan 2011 at 20:40

Uriel said:
If it's any consolation, HeX, my unlocked Athlon X2 5200+ just made it through 24 hour Prime Blend with a 940 odd MHz overclock so I was confident that it was stable (http://forums.overclockers.co.uk/showthread.php?t=18225299&page=2).

IBT brought it down in minutes. It just tests different aspects of the CPU. In my case pushing up vCore a notch seems to have stabilised it. Just made it through 20x IBT on standard. Testing 20x on high now.

What are your temperatures like by the way?

Thing that's stressing me out is I don't know why it's failing... the only thing I can assume is the problem is the CPU is having trouble with me having 8GB of memory and the IMC is getting stressed.

But if that were the case surely I should get Memtest errors... which I don't.

Prime Small FFT's will run all day so the CPU clock is obviously OK, and Prime Blend has done at least an hour.

It's rock solid at 1333 C9, but at 1333 C7 it fails.

It's prime blend stable at C7 with 1.425V. But won't pass IBT/LinX even with 1.5V.

It's prime blend stable at C7 at stock CPU-NB volts. But won't pass IBT/LinX even with 1.35V on the CPU-NB.

It's the most stubborn CPU i've ever had! :mad:

Plus if the memory controller on it was a bit weak surely it shouldn't be rock solid at 1600MHz C9 on less than stock volts with stock CPU-NB...

It just doesn't make any sense!!! *weeps*

Temps are also great, even when I shoved 1.5V through it just to rule voltage out it was still ~50C. With 1.425 it's ~42C load.

Uriel · 12 Jan 2011 at 20:45

I've had CPUs that fail IBT in seconds due to the IMC not coping but run Memtest indefinitely.

I've heard that OCCT is good for picking up IMC errors.

joem1983 · 13 Jan 2011 at 21:54

Im confused as to temps that i'm seeing under load. Where should i be looking

HW monitor reports CPU at 60 but core temp is 51.

Coretemp reports it at 51

OCCT reports it at 51.

If im being optimistic and believe the lower temps i could push for a slightly higher OC.

thanks guys!

VFM Addict · 13 Jan 2011 at 22:08

HeX said:
...........
It's the most stubborn CPU i've ever had!

My sympathies are with you, Hex. I have a late Athlon II x3 445 which is the same stepping as most 450's (i.e. NADHC AD) and its a PIG.

Mine's VID is 1.4v and it needs 1.475v to be stable at 3.5Ghz. I decided to stick with 3.4Ghz at 1.45v just so my fans don't ramp. Temps not really problem but I hate even the slightest fan noise if it can be avoided. TBH the difference 'tween 3.4 and 3.5 will never be noticeable outside of b'marks. Shame though as all my past AMD cpu's were gem oc'ers.