The Phenom II List of Overclocks

[Oldguy932]

Alright, so here's the thread that will have the overclocks of all Phenom II's by who has the highest overclock, and later on maybe by the 5 letter stepping at the beginning if we get lots of different ones like we did with rev F3 X2's.

To be included in the ranking, you need to post the following information.
CPU Model Used
CPU Full Stepping
CPU Frequency
CPU Vcore
NB Frequency
HT Link Frequency
Memory Frequency
Memory Configuration (2x1gb, 2x2gb, etc)
Memory Ganged/Unganged
Mainboard Model
Cooling Type
Windows Version and x86 or x64

You will also need screenshots of the following.
Prime Stable 2 hours minimum for the list of stable clocks or OCCT 1 hour high priority, none for list of suicides.
CPU-Z main screen and memory screen

Not necessary but still appreciated
CPU-Z Validation

Get Overclocking Guys.

Prime Stable:


Suicide:

[leoy]

Prime stable and 8 hours is useless

[merkk1]

With what the power set ups are like on alot of the mother boards prime for 8 hours mite kill the mother board . Some people have lost there mother
board to long prime runs already so they may not wanted to do that just a thought .

[BertM]

according to some, certoun heavy games results in crashes while prime seemed stable.
for me, 3 hours of gaming and a 30minute prime run is a stability test.
prime95 says torture test, and thats what it is, a torture. personaly, i will never run it 8 hours.
also its unknown how long the phenom survives on high oc's.
when i get my phenom, i will not put more on it then 1.45 volt. i will have to see how much mhz that will be.

[Astennu]

With what the power set ups are like on alot of the mother boards prime for 8 hours mite kill the mother board . Some people have lost there mother
board to long prime runs already so they may not wanted to do that just a thought .

In my opinion the system is not stable then. And they have gone to far with voltages/clocks. I have never killed a board with running prime for more than 8 hours. But i'm not clocking with high v-cores anymore. Because the performance per watt gets screwed like hell if you raise the v-core.40 watt exta for only 100 MHz. Its not worth it imo.

But i have done that in the past getting the last MHz out of chips to get the highest 3dmark 01 score i wonder how much power that prometia system with a 2.8 GHz Barton was using back then.

[crazydiamond]

im with them guys, i have only run prime >1 hr maybe a handful of times and never >2 hrs,, too many horror stories lol

[Oldguy932]

I guess doing prime for 2 hours would catch most errors, maybe people could switch it to a custom test that switches fft length after every 10 minutes instead of 15 so they still get all the tests in the 2 hours. What do you guys think about that?

[Brother Esau]

im with them guys, i have only run prime >1 hr maybe a handful of times and never >2 hrs,, too many horror stories lol

You guys are fresh on the Boat aint ya? We at one point would not consider a Overclock Valid unless a Full 8HR Prime 95 Run was done immediately Followed by 3DMARK 05 -03 then you could post your overclock achievement

OK......Then........1HR CPU High Priority of OCCT

[Oldguy932]

I'm fine as long as we can agree on something. I originally posted the 8 hours prime because that's what it was like last year and the year before(meaning 07 and 06). I've never managed to run the intel burn test on my 9600, but it doesn't crash even when I have it run wcg for days on end. I'm also fine with OCCT, last time I ran that program was when it was only for single cores.

[BertM]

as maintainer of the list you can demand more proof of stability and only add to the list like you see fit.

[Brother Esau]

as maintainer of the list you can demand more proof of stability and only add to the list like you see fit.

I agree! This so called new standard of what people deem as "Stable" is utter Bull Sheet!


There is absolutely no way of gauging what we can expect as a norm for 'Stable" Clock Speed per Stepping if we do not set a standard for the word Stable.


@OldGuy..........It's imperative that you institute strict guidelines on this and enforce it. Will be the only way to a concise and accurate Data Base for others to follow!

[roofsniper]

well the reason why i say we shouldn't have a prime requirement is because many people are too lazy. if you looked at the old thread in all of the info listed in the spreadsheet one column would be for how long it was primed. it was blank if no one primed it and if they did it would be listed. if people see that it is completely blank then they would know that theres a possibility that its not stable. i think 2 hours is fine tho like i said prime usually picks up errors for me pretty fast if its not stable.

[Daveburt714]

I'm fine as long as we can agree on something. I originally posted the 8 hours prime because that's what it was like last year and the year before(meaning 07 and 06). I've never managed to run the intel burn test on my 9600, but it doesn't crash even when I have it run wcg for days on end. I'm also fine with OCCT, last time I ran that program was when it was only for single cores.

Thats wierd that you had trouble with Linpack Oldguy932. I never tried it with my 9600 (Maybe somethng to do with TLB, but I doubt it). It has worked fine on both my 9850 and 9950. In fact it has caught errors on a 5 pass run that showed stable on a 1 hr P95 run! There is a new version on that link BTW.

Maybe the best solution would be to just have a stability SS column, let people use the stress program of their own choice, and let people decide for themselves whether they consider that stable enough or not.

Whatever you decide looks like you may have to start a new thread for actuall results... Sorry if I contributed to the clutter...

[Oldguy932]

I guess cpu-z validation shouldn't be completely necessary, I only ever did it when I was doing stuff on hwbot. It also shouldn't be too hard to get, all you need is the click of a button to save a file, then upload it to their server to get it validated.

This thread does seem to pose as much more of a discussion on how to go about keeping the list, so at least it does have some use at the moment. It wouldn't be too hard to use this thread for everything since I'll be linking to people's posts and not posting all the information given in the table. Not everyone has their screen resolution at 1680x1050 or higher to read like 20 different columns without scrolling.

Edit: Just thought about something. Would you guys mind if I required like an Everest Screenshot with the cache speeds and everything for the suicide shots? Or do you think a screenshot would be sufficient? I was just thinking about that because we don't need anyone coming in and being like "There's no ****ing way you got that...".

[Daveburt714]

You guys are fresh on the Boat aint ya? We at one point would not consider a Overclock Valid unless a Full 8HR Prime 95 Run was done immediately Followed by 3DMARK 05 -03 then you could post your overclock achievement

Not trying to be disrespectful BroE, but I've seen an awful lot of Phenom boards die from running Prime for hours on end and it is usually the power sections of the boards that pop (IE: burnt Fets).

Prime95 is what killed my MSI DKA790GX, well that and a high CPUVid. The Asus 79-T has a very robust power section (8+2) and may be able to handle it but I for one won't be doing any 8hr prime runs.

Maybe I've been on the boat long enough to know better...

[roofsniper]

yea it might not be 100% stable if you don't run it for at least 8 hours but if you run it for at least two it should be stable for everything else since there aren't aren't programs that run 100% cpu usage besides stressing ones. i still think we should just have a column for how long it was primed and if people didn't prime at all we will all be able to see that.

[hstuehmeyer2000]

Not trying to be disrespectful BroE, but I've seen an awful lot of Phenom boards die from running Prime for hours on end and it is usually the power sections of the boards that pop (IE: burnt Fets).

Prime95 is what killed my MSI DKA790GX, well that and a high CPUVid. The Asus 79-T has a very robust power section (8+2) and may be able to handle it but I for one won't be doing any 8hr prime runs.

Maybe I've been on the boat long enough to know better...

This is exactly why I do not run prime any more for more than 10-15min. The hardware we use is not made to take this kind of beating and there is no reason for it anyway. If it is really unstable, you will know as soon as you do something demanding anyway.

[roofsniper]

This is exactly why I do not run prime any more for more than 10-15min. The hardware we use is not made to take this kind of beating and there is no reason for it anyway. If it is really unstable, you will know as soon as you do something demanding anyway.

ehhhh 10-15 min is not that long i run it a minimum of 30 min and that has caught every error i have ever had.

Never killed anything running prime, cpu-burn, toast, etc...

I guess if you aren't beyond your stable point of running a system and you run prime you might not have problems, but if your are like most of you and just have it overclocked without knowing its stable then yea I guess prime could kill a few things...

I use my pc's to run Folding@home 24\7 so having a system that isn't stable is out of the question but I guess if all you do is web-surf it would be ok to have a unstable overclock.


2 hrs prime wouldn't even qualify for being stable IMHO

prime can kill your hardware even if it is stable. the mosfets can blow and even if the system is stable it won't take 100% load at all times so the mosfets won't take a beating. prime is a nice program but after awhile it just goes too far. lol i wish particle would work on his benching program some more. i bet that thing stresses your system more than prime can. hell once you start it the system locks up until it is done its that powerful.

[Brother Esau]

Not trying to be disrespectful BroE, but I've seen an awful lot of Phenom boards die from running Prime for hours on end and it is usually the power sections of the boards that pop (IE: burnt Fets).

Prime95 is what killed my MSI DKA790GX, well that and a high CPUVid. The Asus 79-T has a very robust power section (8+2) and may be able to handle it but I for one won't be doing any 8hr prime runs.

Maybe I've been on the boat long enough to know better...

Dave .......could be that first generation hardware was inadequate Don't know. But I do know that most of the board related issues are from people pushing the HT.................(Thats a Fact).....did you do that? I know myself I have been very mindful of not doing that in any given case because of what I was witnessing happening to others boards..so I avoided it and have primed my machine for 9-11 hrs on numerous occasions same board since day one without issue

C'mon Dave I aint calling anyone out as a noob but this new standard of stability testing is not what I would call stable in which to judge

[Titan7171]

Dave .......could be that first generation hardware was inadequate Don't know. But I do know that most of the board related issues are from people pushing the HT.................(Thats a Fact).....did you do that? I know myself I have been very mindful of not doing that in any given case because of what I was witnessing happening to others boards..so I avoided it and have primed my machine for 9-11 hrs on numerous occasions same board since day one without issue

C'mon Dave I aint calling anyone out as a noob but this new standard of stability testing is not what I would call stable in which to judge

9-11 hours?...Easy for someone who runs their cpu at stock for "bios testing" lol.......put a little juice into it..........anyone could run prime at stock settings so whats the deal with that?

[cbjaust]

I agree with AMDGeeks.net and their definition of a stable overclock:

"Level 3
Certified 'Rock stable overclock'

* CPU-Z validation
* Super PI Mod 1M pic (1.4/1.5)
* 10 Hrs Stressprime pic (Prime95/SP2004)"

As a I side note does anybody know if Orthos picks up all cores of CPUs with more that 2 cores and automatically run 4 threads on them?

Cheers

[Daveburt714]

Dave .......could be that first generation hardware was inadequate Don't know. But I do know that most of the board related issues are from people pushing the HT.................(Thats a Fact).....did you do that? I know myself I have been very mindful of not doing that in any given case because of what I was witnessing happening to others boards..so I avoided it and have primed my machine for 9-11 hrs on numerous occasions same board since day one without issue

C'mon Dave I aint calling anyone out as a noob but this new standard of stability testing is not what I would call stable in which to judge

Bro, I'm not really sure the mobos have gotten that much better, especially concerning the PWM sections of the boards (Asus 79-T being the possible exception)... Most of my chips have been BE's, and I've only used the HTref to tweak my memory and it has never really required much over 220 on that clock (1180 Mem speed)...

When my DKA790 died my HTref was well under 220, I can't say for a fact that your wrong, but I'm really not sure high HT clocks kill boards...

From my experience, high V's kill stuff faster than anything else. When my MSI died I modded the AOD file to allow 1.55 CPUvid, and it popped after only 20 minutes of P95....

I'm not really sure that your assumtion that high HTRef clocks kill boards. I agree with ya the that with BE chips it's really not nescassary and can cause more problems than it's worth, but when OC'ing a locked chip as long as you keep V's under control I'm not sure it is a board killer either.... If I had a PhII 920 I wouldn't hesitate to crank the HTref to 250.

My biggest problem is running Prime95 for hours on end... Most Phenom boards still have weak PWM sections with inadequate cooling. Even the boards with Heatpipe coolers still don't do the job...

Most of the people I've seen kill stuff do it by running really high V's on their IMC/NB volts... My personal limits on these are:

1.55v's on the CPU core (Asus 79-T)
And 1.4v's on the IMC (bench only, everday I keep it under 1.375v)
The IMC is VERY sensitive to high voltage!

All that being said, for me IntelBurnTest will catch errors in 20 minutes that will seem stable even after hours of P95, and I don't have to worry about waking up up with a dead box...

[DavyBoy]

im with them guys, i have only run prime >1 hr maybe a handful of times and never >2 hrs,, too many horror stories lol

What horror stories??

1 hour of Prime95 isn't long enough to test for stability IMO...

There are people with i7 rigs running it for 12+ hours with 4GHz overclocks, and they don't have any problems...

[demonkevy666]

What horror stories??

1 hour of Prime95 isn't long enough to test for stability IMO...

There are people with i7 rigs running it for 12+ hours with 4GHz overclocks, and they don't have any problems...

take a look at the power regulation size of those boards compared to AMD's

they have more phases to them they're design for 300 watts max ours max out at 200 watts.

[jonspd]

I knew I shoulda snapped a SS when I killed prime after 10hrs

gonna try for higher on a later date anyways...