CPU & RAM burn-in guide

-->The final cut!<--
(Remember a 70's song of Pink Floyd).. :D

When I returned home I thought to check again MSF3, MSF4, MSF5 and MSF6=2630 MHz (see table above) using 8M SuperPi (cos this table filled running 8M).
I did not check MSF7 cos this is the final step.
New MSFs were 1-3 MHz higher than before.
That's why I decided to bumped Vcore to maximum value via BIOS although stretching step by step is better but needs time.
I started at 2670 MHz the burn-in process (Lowest Priority, ZZZ=0, Threads=1).
After 30 minutes (CPU Stretcher was running) I tried 1M SuperPi and CPU passed it.
Immediately I tried 2M SuperPi but failed after 4 loops...
"I will retry 2M later" I thought...
Damn failed again...

http://www.xtremesystems.org/forums/...chmentid=36965

Jim Morisson said "People are strange" LOL I say "overclockers are crazy" :D
After 4 hours burn-in CPU passed 4M!

http://www.xtremesystems.org/forums/...chmentid=36966

But this does not mean that my CPU is SuperPi 8M/16M/32M stable there.
Is getting stable and I hope after some hours better results for 2670 MHz; why not for 2680 MHz?? etc..
After this I increased speed to 2680 MHz.
I did not run SuperPi immediately but after half hour...
Damn I tried 5 times and CPU could not pass 512K test!

http://www.xtremesystems.org/forums/...chmentid=36967

"Never mind" I thought burn-in will stretch you!
YEAH!

http://www.xtremesystems.org/forums/...chmentid=36968

Notes
1) When I tried for first time Vcore7=1.663 Volts LOL my CPU could not POST!
I turned off the mainboard and then OK.
2) My mainboard has a program for on the fly overclocking.
This program increases Vcore also but I have noticed instabilities sometimes (abnormal restarts or blue screens)..
That's why I prefer overclocking via BIOS.
But I will retry on the fly Vcore increases after a couple of days...
3) Most instabilities are when I increase FSB.
That's why I will run my PC at semi stable frequencies for soft applications (word processing/programming) and for 100% usage 50 mVolts more!
4) Although table shows from Vcore6 --> Vcore7 gain was only 20 MHz (see % loss is high) Vcore7=1.663 Volts will be for permanent usage cos I want 2650 MHz (or more) stable now. :)

can a folding software (D2OL, F@H etc) be used to burn in your cpu?

x714n I don't run folding programs so I have no idea.
Some friends told me that running Prime95 for hours did not see any significant gains.
But don't confuse burn-in programs most of them don't include extra routines for checking any results --> if your CPU produces wrongs burn-in program will not stop.
Similar idea follows CPU Stretcher.

Damn when I returned home CPU crashed!
Never mind!
I retried again with same settings and CPU passed 8 hours burn-in at 2680 MHz.

Mainboard burn-in?
Well before this time consuming process my mainboard could not POST for FSBs > 289 MHz.
Yesterday I boot my mainboard at 8*292 and system could boot in Windows but SuperPi produced errors.
I don't think that CPU Stretcher helps mainboard's chipset cos I increased Vagp to a higher value.

MrSeanKon , try my burn in guide as a change , you only need Prime95 and Cpu Burnin for it. I have been running it for over a week. Earlier I wasnt stable at 2.5 ghz at 1.5+ V. Right now Im sitting here typing at 2.53 ghz @ 1.48 V while Priming for the past 5 hours.

Its worth giving it a shot. Also I didnt like the MSI utility while i had the board. I used to make my overclocks unstable. YMMV.

WOW!
Better results!
Thanks mate and burn AMD babe burn! :toast:

Finally I stopped burn-in process!
Why???
Cos Thanasis=a friend of mine here gave me his Winchester 3200+.
It is 1.5 months old.
As you can see burn-in seems does not work cos it is not :stick: new...

Finding MSFs based on data
Here we can see the importance of database.
The second 939 CPU is a Winchester 3200+.
System and settings were same like Venice 3200+ test.
Default Voltage for Winchesters is 1.400 Volts.
Thus I started with Vcore1=1.361 but I could start with Vcore2=1.413.

First advantage
It is not necessary to measure Vcores at full load cos I have filled my golden table!
This means that 1.400Volts in BIOS --> 1.413 Volts multimeter shows and so on. ;)

In general new steppings of CPUs achieve higher speeds.
But I started 8M SuperPi (like Venice 3200+) at same MSF --> 2310 MHz.
Winchester failed this test and passed 8M at 2300 MHz.
I think that I was very lucky! I found easily the first MSF --> MSF1=2310 MHz.
I increased voltage to Vcore2=1.413 MHz.
My golden table shows that Venice 2310 --> 2390 MHz.
"Let's try SuperPi 8M at 2380 MHz" I thought.
YEAH Winchester passed 8M and failed at 2390 MHz --> MSF2=2380 MHz!
I tried with Vcore3=1.463 Volts.
Venice 3200+ gained 80 MHz again (2390 --> 2470 MHz).
I tried at 2380 + 80 = 2460 MHz --> SuperPi 8M crashed immediately.
Reboot 2450 MHz --> crashed again (after 7 loops).
Reboot 2440 MHz --> SuperPi 8M crashed after 18 loops (required loops are 22).
Thus ;) it is not necessary (save time) to check at 2430 MHz SuperPi --> MSF3 = 2430 MHz.
With Vcore4 Venice 3200+ gained 50 MHz (2470 --> 2520 MHz).
Although Winchester behaved a bit different at previous step with respect to Venice I started at 2480 MHz.
Wow 8M passed!
Reboot --> 2490 MHz --> 8M passed again!
Reboot --> 2500 MHz (multiplier = 10X) --> SuperPi 8M failed ---> MSF4=2490 MHz.
With Vcore5=1.562 Volts Venice 3200+ gained 70 MHz (2520 --> 2590 MHz).
I tried at 2490 + 70 = 2560 MHz SuperPi 8M --> failed after 20 loops!
;) ---> MSF5 = 2550 MHz! ;) without check ;)
I increased voltage to Vcore6=1.610 Volts.
Venice's gain was 40 MHz (2590 --> 2630 MHz).
I tried at 2550 + 40 = 2590 MHz --> 8M crashed immediately...
2580 MHz --> crashed....
Damn 2570 MHz crashed again after 17 loops --> MSF6=2560 MHz (I did not test 8M again).
Finally Venice gain was 20 MHz with Vcore7=1.663Volts.
I tried Winchester at 2580 MHz --> passed.
Reboot 2590 MHz --> SuperPi 8M passed! Wow!
Reboot 2600 MHz --> crashed --> MSF7=2590 MHz.

Here is my golden table:

http://www.xtremesystems.org/forums/...chmentid=37083

I spent 1.5 hour only and I know anything for my second sample. ;)

Mainboard burn-in?? What's on??
As you can see in first page I posted that I could not boot my PC with FSBs > 289 MHz.
Here mainboard's chipset ran at 320 MHz FSB:

http://www.xtremesystems.org/forums/...chmentid=37084

Multiplier was 8X and for stability RAM/FSB was 166/200.
My mainboard settings in BIOS are simple it is has not tricky BIOS.
In addition I do not have any modded BIOS versions...
I tried two editions but they did not work very well with Vcore and I upgraded with MSI's latest edition...
Anyway this is --> cold boot issue --> Guidemania --> Top o/c guide --> Part 1.

Winchester vs Venice
Let's summarize MSFs and compute exponents:

http://www.xtremesystems.org/forums/...chmentid=37085

Both processors have n > 0.5 --> that's good for a quick approach..
In the beggining from Vcore1-->Vcore2 n was same (OK 0.908 is near to 0.912).
Winchester for next steps had almost same exponent (0.761, 0.764 and 0.749) but for Vcore6 n is 0.637 quite different than 0.749.
Input my data in O/c analyzer find exponents and see how is the % loss.
Venice had big loss from Vcore3 --> Vcore4.
For other steps both processors behaved normally.
Although my Venice sample wins Winchester you saw how much useful is data collection.
Both processors have similar core --> therefore any processor with similar behavior will have the same (or similar) o/c.
To be more specific:
Assume a San Diego 3700+ user.
It is not very different core than my samples.
Default Vcore for San Diego 3700+ is 1.350 - 1.400 MHz and runs at 2.2GHz.
So what are my recommendations to any 939 owner? (NewCastel/Winchester/Venice/San Diego --> good aircooling/watercooling not subzero cooling)...
1) He/she should find the first MSF with default Vcore (or the closest Vcore).
Normally all 939 models with default speeds <= 2.2GHz should achieve 2.4GHz but my samples ran up to 2.3GHz.
Other users have achieved higher speeds with default Vcore without subzero cooling for here is a San Diego 3700+ at 3GHz watercooled
If he/she has a 2.4GHz CPU (for example a San Diego 4000+ or a NewCastle 3800+) then first MSF should be around 2.6GHz (not subzero cooling).
2) Vcore2/Vcore1=1.038 and both processors gained 80 MHz.
Therefore when user increases Vcore near to 4% (from default value) MSF2 - MSF1 is around 80 MHz.
But this rule based on two samples --> if his/her sample behaves different then MSF2 will be different.
3) When he/she found MSF2 then things are easier.
He/she will follow BAN method :D and General CPU speed estimator (using O/c Analyzer programs) if his/her sample behaves different than my Venice+Winchester.
More tutorials about BAN method --> read O/c Bible's thread here at XtremeSystems, PcPer forums, ABXzone and OCAU.
Otherwise he/she will follow my tables and so on.
Visit OcForums and read section "O/c analysis in depth".


My Winchester was banana
I increased more Vcore and for 2610 MHz needs 1.769 Volts! (SuperPi 8M).
Here are the results:

http://www.xtremesystems.org/forums/...chmentid=37086

% loss is very high for Vcores > 1.663=Vcore7.
Venice 3200+ had higher exponents for Vcores > Vcore7 but these voltages is suicide. :D
That's why I stopped burn-in process after 18 hours with CPU Stretcher; I did not see any difference for MSF7.
Never mind!
Until now it passes 32M at 2570 MHz (MSF7 - 2*multiplier) for 2590 MHz + 32M needs more Volts :D hope to stabilize it there with lower voltage cos I don't expect speeds > 2.6GHz.

Continued: :D
I took back my Venice 3200+!
Finally after a week burn-in it is 32M stable at 2670 MHz.

http://www.xtremesystems.org/forums/...chmentid=37375

Conclusion
As you can see until until this moment the gain is:
Before burn-in 8M stable at 2650 MHz, after burn-in 32M stable at 2670MHz.
Vcore is the same of course.
But for 100% safe I run my chip 2650 MHz regardless it is 32M stable +20 MHz higher.
After this process I will run my CPU at 2670 MHz for 6-8 per week just to keep my gains. :)

That is a lot of work =)

Most work for me is not when I use this program but when I prepare guides/manuals/correct my grammar/programming C# --> v1.43 will be around 55.000 lines source code...
This program requires patience (see FAQ).
OK for some users is useful; others don't need it...
Any user who visits our site can vote for O/c Bible! :)
If it is difficult or not! :toast:

hey, I am gonna try this sometime this week, but I have ha my chip for awhile now should I even bother to do it? Also I notice that I need more voltage to get the CPU stable for a given overclock, example I could be stable at 1.35volts 240x10 (3200+ venice) now I need 1.4volts similarly I was stable at 1.45 volts with 250x10 now I need 1.5volts.

Very useful info FE4R! :toast:
Consider this:
I was on holidays for July/August and my previous A64 754 socket passed 32M with 1.744 Volts at 2550 MHz.
It ZZZed :D LOL a new verb :D for 40 days and when I came back it could not pass 32M at same speed.
Ambient temperature was same and cooling...
:stick: "I will reburn you" I thought and after 3 days (3 * 12 hours burn-in) CPU became stable again at 2550 MHz.
Xm I think to rename this program to CPU doctor! :D

News!
Temperatures here are lower!
Now room is at 25C not 31 like previous days!
2680 MHz 32M stable!

http://www.xtremesystems.org/forums/...chmentid=37716

I think AMD gain estimator works fine for this 939 although I did not performed special measurements.
More info see here AMD gain estimator analysis

Why is the OC Bible ofline :( , I would like to have a look at since I,m having problems with voltage and burning in.

Nice guide dude.

very well laid out and great pics :up:

Quote:

---

Originally Posted by Hugo van Dijk

Why is the OC Bible offline :(

---

OK you can download it now our site works.

Quote:

---

I am having problems with voltage and burning in.

---

Please read instructions carefully.
I assume that you have found your CPU MSF with a stability program of your choice (Prime95, OCCT, SuperPi 32M etc)...
Also I have to warn you again don't use this program if your PSU is not qualified.

Quote:

---

Originally Posted by MrSeanKon

OK you can download it now our site works.

---

Thanks MrSeankon :woot:

Regards mate!
LOL I think my Venice starts loving Vcore! :D
That's why I bumped Vcore for permanent usage to maximum value = 1.816V.
Yeah it passed 32M at 2750 MHz!
Few days ago max MSF (32M stable) was 2730.

http://www.xtremesystems.org/forums/...chmentid=37849

Note:
Actual temperature is 32C (measured with thermistor placing probe on heatspreader).

Hey, I am burning in my CPU at 2400Mhz vcore at 1.35v (via bios, and I do not have a multimeter). As I said before it was primestable at 1.35volts for hours, now its not anymore.

I am trying to get it primestable again, what settings do u suggest? I have been doing the burn in at Thread prioroty set to above normal, threads set to, zzz set to 10, vocre set to 1.35 (read from bios) and the mhz at 2420Mhz.

EDIT
well, I cant pass prime at 1.35v not a big deal as the stock vcore is 1.4v. but I was shocked to see I can now again do 2.5Ghz when I set the bios to 1.45v (I need a multimeter) Passed prime95 for 41mins untill I stopped to play game. I gonna do CPU burn in everyday for the next couple of days/weeks with vcore in bios set to 1.35 and clocked to 2400Mhz to see what happens, thanks man :).

FE4R you should give me more info:
1) What's your mainboard?
2) What's your cooling?
3) What Vcore you are thinking for permanent usage?

You did OK cos in the past your CPU passed Prime95 at 2.4GHz with Vcore=1.35.
But don't use ZZZ at this moment cos you are running near to previous MSF I mean 2420 - 2400 (the previous MSF) is not great difference....
Also be careful with priority don't set Above Normal or High Priority right now...
Find first how far you are CPU Stretcher stable (ZZZ=0, Normal Priority, Threads=1) and then if you want to go higher set ZZZ=20 - 50 and Lowest Priority...
Run this program at least for 6 hours per day but as you saw my recommendations don't run it for many hours!
You must relax your CPU for some hours! :)

Well.. i didn't follow your guide, but the burning-in seems to work.

I have a Venice 3000+ on a DFI Ultra-D NF4 (on air) and on stock volts (1.4v in BIOS) i could only get to 2556Mhz Prime Stable without the burn-in. I used Toast 2*4 hours (with 2574Mhz stock v) on separate days and now i can do 2574Mhz Prime Stable. Also, i tried 2601Mhz before burn-in and it wouldn't prime, now i can prime for 30 mins. Doesn't seem much but it seems to be working.

I must add i don't trust SuperPI anymore because i can do 2601Mhz 32mbstable, but it doesn't prime properly, nor does it play games very well.

Sean, I'm looking for a max v.core of 1.55v in BIOS, With my info above, what kinda things do you recommend to improve my speeds/burn-in's?

Quote:

---

Originally Posted by KingYubbo

Well.. i didn't follow your guide

---

As you can read in previous posts we discussed for alternative methods --> 29th post above.

Quote:

---

Originally Posted by KingYubbo

but the burning-in seems to work.

---

Yes but any user should be patient and careful.

Quote:

---

Originally Posted by KingYubbo

I used Toast 2*4 hours

---

It's a burn-in program also.
CPU stretcher is a soft burn-in program and for my PC Toast is harder but running CPU stretcher at Highest Priority and two threads regardless cos I don't have AMD X2 all seem frozen.

Quote:

---

Originally Posted by KingYubbo

i don't trust SuperPI anymore because i can do 2601Mhz 32mbstable, but it doesn't prime properly, nor does it play games very well.

---

SuperPi 32M is good stability program for RAM checking and for AMD CPUs but Prime+OCCT are more sensitive for CPUs.
To go further my AMD runs video compression at 2760 MHz for 6 hours but I don't prefer a semi-stable stable there.
So cos my favourite applications (video + Maths) are not heavy such as SuperPi 32M this stability program is enough for me.
I don't play games also.

Quote:

---

Originally Posted by KingYubbo

I'm looking for a max v.core of 1.55v in BIOS, With my info above, what kinda things do you recommend

---

Xm cos you passed 32M at 2.6GHz try Toast as you know at same speed for a couple of hours.
If you choose CPU Stretcher start at 2620 or 2630 MHz with 1.55 in BIOS.
ZZZ=0, Lowest Priority, one thread.
If you OK continue higher and so on until final crash.
Then at CPU Stretcher's MSF burn it for 2 * 8 hours with Normal Priority, ZZZ=0, one thread and retry Prime95 if you gain something. :)

Have searched for CPU stretcher a lot..but without results...Can you give the direct DL link please?...I`m burning now with dual CPU BURN-IN...but unfortunately withouth results...My comp is rock stable at 2,6ghz 1.552V but can`t get any further...have burned it 4 days for 7-8 hours daily but at different settings ( 2,65ghz @ 1.6V, 2,7ghz @ 1,63V and 1,64V)...Super_PI pass 32MB withouth problems but SP2004 crashs after 30-40 min...any suggestions?

Regards,
Primoz

Quote:

---

Originally Posted by Primoz78

Have searched for CPU stretcher a lot..but without results...Can you give the direct DL link please?

---

Click on the first link and you will see the page and all links.

Also on the left of your monitor exist some buttons --> click on button "Forums" and you will see our forum.
Then in English section (it is a double forum Greek + English) you can see my first guides and the latest announcements for O/c Bible it is not necessary to bother anyone here or at other supported thread with announcements "I found this bug", "This is the tip for this program" or "You have to run this cos this program hangs" etc...
There things are different as you can understand. :)
Until this moment exist two categories in English we are under construction in the future we will add more.
Of course I made a special category in English only for O/c Bible questions LOL although I am Greek I will not post anything in Greek there cos it is too boring (I get confused switching between two languages).

Quote:

---

Originally Posted by Primoz78

My comp is rock stable at 2,6ghz 1.552V but can`t get any further

---

OK mate what's your stability program?
Keep in mind that MSFs differ if you test your system with different stability programs for example SuperPi 32M is 2620 MHz stable, Prime95 at 2.6GHz etc (hypothecal values)...
So tell us timings Vdimm, FSB etc....
Maybe your RAM is not stable cos Prime95 (blends FFT stretches CPU+RAM).
That's why I prefer S&M for fast FPU test cos it stretches only CPU.
After Mapping method I seek my system limits more.
Also play with HT ratio...
HT speed should be <= 1GHz.

Quote:

---

Originally Posted by Primoz78

have burned it 4 days for 7-8 hours daily but at different settings ( 2,65ghz @ 1.6V, 2,7ghz @ 1,63V and 1,64V

---

Xm I don't follow this idea.
When I burn my CPUs I don't change Vcore as you can see above.
CPU was 8M SuperPi stable at 2650 MHz before burn-in.
Vcore was 1.663 Volts.
I burned it as you read above but Vcore was 1.663 Volts also and finally passed 2680 MHz 32M (the first improvement).
After this :D cos temps are OK now here I bumped to 1.816 :slobber: and I retry burn-in and now runs 2750 MHz 32M.
To inform you more OCCT's MSF is 2720 MHz (Vcore=1.816).

Hope to help you.
Sorry I cannot reply soon. :)

MrSeanKon thx for all the explanations and help...will report later how far I will get with OCing...I have the HT x3 and I never excede 1000mhz!

Regadrs,
Primoz

What would you recommend for burning in the new opteron cpus ? I've got a 146 here stock voltage and stock vcore = 31° idle Temp on water. So it's running very hot, so what exactly should i try ?

Quote:

---

Originally Posted by MrSeanKon

Click on the first link and you will see the page and all links.

OK mate what's your stability program?
Keep in mind that MSFs differ if you test your system with different stability programs for example SuperPi 32M is 2620 MHz stable, Prime95 at 2.6GHz etc (hypothecal values)...
So tell us timings Vdimm, FSB etc....
Maybe your RAM is not stable cos Prime95 (blends FFT stretches CPU+RAM).
That's why I prefer S&M for fast FPU test cos it stretches only CPU.
After Mapping method I seek my system limits more.
Also play with HT ratio...
HT speed should be <= 1GHz.


Hope to help you.
Sorry I cannot reply soon. :)

---

Have tried the S&M...damn it get my CPU under water to 50°C and my PWMIC at 65°C :shocked: This APP is running really hot my CPU...never reachd 50°C before and for PWMIC never get over 55°C...even with dual Prime or SP2004 at 1.65V...and 2,7ghz (ram at 166 divider 245 2-2-2-5@ 3,3V). If It pass at normal settings for CPU and RAM time and full load CPU can I say it is 100% stable...and super_PI 32MB passed...? :stick:

Regards,
Primoz