Sign in with
Sign up | Sign in
Your question
Solved

Finally Stable ?! AMD Phenom II X4 965 Processor BE , Success ?!?

Tags:
  • AMD
  • Overclocking
  • Product
Last response: in Overclocking
Share
September 16, 2010 5:14:48 AM

Well after much trial and tribulation today it seems i have finally came to a "stable" OC of 4000MHz!
With a Core voltage of 1.48V, I was having problems with stress tests using LinX but realized maybe i should try the actual recommended stress test in the actual AMD overclock sticky.

I downloaded and updated 3Dmark06 and passed the test seemingly pretty well with a high cpu temp of 55C, a low of 36C, and an idle of about 40C and lower.

I followed the overclock guide pretty closely with upping the voltage every time i blue screened with the 20x multiplier i had set from the stress testing. With this Core voltage of 1.47v i was able to successfully pass the 3Dmark06 test. Does this mean i am finally actually "stable" at this OC, i was reading in the guide that if i passed this successfully, any long term stress testing could just ultimately end up messing up my cpu, which i don't wanna do!

The core voltage in BIOS i have set is 1.47 and it fluctuates on load with 1.47v-1.488.
Here is my CPU-z validation link, any tips or pointers that would help me fine tune this would be greatly appreciated !



;)  Big thanks to simon12 and saint19 for their helpful responses and help during these OC tests.

More about : finally stable amd phenom 965 processor success

September 16, 2010 10:38:53 AM

Spoke too soon ? Or should i be happy with passing 3Dmark06 test? After i posted this i decided to run a 1 hour OCCT stress test on the cpu for good measure, when i woke up i realized from the "system has recovered from a serious error" that unfortunately i had at one point or another received a BSOD during the test.

Just curious as to whether i should be happy with passing the initial 3Dmark06 on the 4000MHz OC and disregard the failed 1 hour OCCT cpu stress test?

I went into bios and upped the voltage to 1.48, gonna run 3Dmark06 and a OCCT 1 hour cpu stress test , hopefully it passes and i can be "stable" at this 4000 OC.


I am just wondering am i on the right path and mindset for this current OC phase i am in , feedback would be wonderful.
m
0
l
a b K Overclocking
September 16, 2010 10:45:53 AM

Bench stable, and really stable are different things.

you need to run prim95 for at least 4-6 hours to make sure you are truly stable (i recommend overnight).

you are still on the right path (and kudos for doing your research, most people come here looking for easy settings...) but you may find upping the voltage won't stabilize you. there is a point where voltage will no longer help. you can try of course, but its possible you will need to lower the clocks slightly in order to maintain a stable setting.

i usually find a max stable overclock that passes a 4 hour test, then drop back 50-100mhz to be sure i'm absolutely safe. But its up to you. either way, you need to pass a prolonged 100 percent load situation. not just a video benchmark.
m
0
l
September 16, 2010 11:04:45 AM

Well , i guess what i have to accept is that maybe this chip just cant handle a 4000MHz OC , i just got out of BIOS from upping the voltage to 1.48 , started a Linx hour and a half test and my computer didn't even BSOD it just turned off.

I guess this is where i take the advice of the overclock sticky and be content with my stable OC of 3900 MHz or things may become disastrous? Although i have not messed with things like memory, and north-bridge , is a 4000 MHz OC really out of the question right now and should i just accept the 3900MHz OC ? I don't wanna harm my system but i did have that 4000 OC in my head as a stable goal. :( 
m
0
l
a b K Overclocking
September 17, 2010 12:35:03 AM

400mhz may still be attainable, with more voltage. however the system turning off and not BSODing is usually indicative of a heat problem. the system shuts itself off to avoid damage from excess temps. were you monitoring temps at the time? both CPU and Mobo temps need to be looked at.

usually bringing it down 100mhz isn't needed. you could just lower the Base clock slightly to drop it 20mhz. that may be enough to stabilize it if 4Ghz is not attainable. however based on the current situation i would not recommend more voltage, until cooling is improved. what CPU cooler do you have? and how many case fans?
m
0
l
September 17, 2010 5:45:24 AM

I am using the Zalman CNPS9900ALED 120mm 2 Ball Low-noise Blue LED CPU Cooler

http://www.newegg.com/Product/Product.aspx?Item=N82E168...
So it is a pretty nice CPU heat-sink and very OC worthy.
and i have 1 case fan, A rear 120mm.

When i was stress testing at 4.0 using just 3Dmark06 i passed it and temps were with a high cpu temp of 55C, a low of 36C, and an idle of about 40C and lower. Although when i tried running linx for that one hour and the pc shut off , it happened so quick i dont know what the temp spikes were.

At this point should i maybe think of raising the north bridge ? If so or not, what is the purpose of doing that ?

Right now i am running very stable at 3.9 at 1.42v , its just funny that that extra 100 mhz is being such a stickler to get stable. I want 4.0 to be stable for gaming, right now there aren't many games that put a constant 100% grind on the cores for any extended duration, i was thinking if it passes 3dm06 and pays fine heat wise when im gaming, i should be fine possibly?
m
0
l

Best solution

a b K Overclocking
September 19, 2010 11:04:39 AM

sorry about the late reply, was away for the weekend.

basically, yes. 3dark will put way less stress than a full load stress test. chances are, under 100 percent load the temps are too high and its shutting off. re run the stress test using HWmonitor to log temps. you will probably find yourself hitting the 65 degree ceiling.

adding more case fans could help, as a single exhaust won't be the most efficient setup. and i assume it is the heat holding you back, not the stability of the chip.

you could keep running at 4.0 without fixing the heat issue. but given a little extra load than 3dmark 06 (which really wont put much) and you could face system crashes. and system crashes can be more than a minor inconveniences when badly timed, even ruining your OS install and forcing a format. so i wouldn't recommend it. in reality, the performance difference between 3.9 and 4.0 is negligible, so you aren't really losing much. though as an OCer i do understand that need to push as far as you can. but for that, i recommend re-running a 100 percent load and monitoring temps. which will most likely confirm heat as a culprit. you can then go about improving airflow and hopefully keeping 4.0 stable.
Share
September 19, 2010 6:01:23 PM

Hey there 4ghz is attainable. i just crossed this barrier and finally got it stable. I was running 3.8-3.9ghz very stable (overnight on prime95) @52c. The 4ghz mark took me a while of different thing to keep it stable. As you know there are many variables heat, nb frequency,ram timings, ram frequency, etc... Don't give up it can be done as I did it. If I can anyone can LOL



This is a 2 hour screnie of a 2 hour prime95 @ 4ghz temp was 56c

http://img291.imageshack.us/img291/488/cpuid918.png
m
0
l
a b K Overclocking
September 19, 2010 10:41:07 PM

Quote:
If I can anyone can LOL


not true. all chips are different. just because yours was able to hit 4ghz, doesn't mean they all can. Don't come here giving incorrect info when you don't know what you are talking about please. (don't mean to sound so hash, but its happening so often lately, and its only a matter of time before someone breaks something)

but, i do think the barrier here is the heat, not the chip. so 4ghz may still be attainable. assuming OP can improve airflow a little.
m
0
l
September 19, 2010 10:55:34 PM



but, i do think the barrier here is the heat, not the chip. so 4ghz may still be attainable. assuming OP can improve airflow a little.[/quotemsg said:


welshmousep

I realize that is true (Not all chips are created equal) Maybe I should have been a little more specific about his issue. I do apologize about that. I too was stuck at 3.9 as all of us do know overclocking there is always a risk. I will keep my thoughts to myself...
Sorry
m
0
l
a b K Overclocking
September 20, 2010 5:20:04 AM

I don't mean to discourage you from posting, just to be mindful of the content. many people come here thinking that there is some single setting they can input to hit a certain clockspeed. and its that sort of misinformation that can cause serious problems. with overclocking, more than anything else on these forums, you need to be sure of what you are saying. and to be sure that it will not be misinterpreted.

anyways, I'm sure OP gets the point now. so lets just see if he can confirm heat as the barrier here.
m
0
l
September 21, 2010 12:27:31 AM

Well i am running a pretty weird stock ram config right 5 gigs of ddr2. I was talking to some people and basically that might be one of the main reasons, my ram not supporting the OC properly. I am not even taking advantage of the dual channel atm, and i didn't even know till recently, i went ahead and and purchased 4 gigs of ddr2 , two 2 gig sticks, and 2 more fans to lower temps. If i still have problems i am gonna cry ha ha but .... alas, another problem presents itself, my ram was supposed to be on my doorstep this morning , fedex website said it was left on doorstep and delivered... yeah its not... so lol WTF so stressed...
m
0
l
a b K Overclocking
September 21, 2010 4:27:37 AM

Shouldn't be the RAM. you have a black edition chip, so overclocking is done by raising the multi. this is independent of the rest of the system unlike FSB OCing. so the RAM should make no difference.

the fans however should help considerably, and you can also try re-applying some TIM. as i said, im fairly certain that heat is the issue here. so hopefully the new fans will make enough of a difference to get you stable.
m
0
l
September 22, 2010 12:34:01 PM

Fans come in tomorrow, as well as some arctic silver 5 thermal compound, gonna remove the stock compound that came with the chip and apply the new stuff. As of right now i still have had 0 success with anything over 3.8 . Today i started with 3.8 which is my most stable OC to date. Passed a 4 hour OCCT test @ 100% load , also passed a 2 hour 100% load Linpack stress test at 3.8 but anything above that so far is not even stable for more than 5 at worst to 16 mins at best in 100% cpu load tests.

I started at 3.8 with the stock voltage of 1.375 which is a multiplier of 19, moved the multiplier up to 19.5 @ 3900 MHz but have not had a stable multiplier thus far raising the voltage up by .1 each time , as of now i am up to 1.383v @ 3900 still going up in voltage each test i fail to try and find the sweet spot, it seems i got quite a while to go since i have seen people @ 3.9 and 4.0 hovering anywhere from 1.40-1.5+volts in their 3.9 and 4.0 stable overclocks.

I have been monitoring temps and they vary depending on the stress test, Linpack does a rhythm of 40% load to 100% load during its tests and my temps rang from 40c to 51c at the highest. OCCT being a 100% stress test constantly, the temps stabilize at around 50c these dont seem to be high temps for this OC of 3.9 .

Oh well i know this is trial and error, maybe things will change here soon.
I think the fans will help more when they come in but like i said my temps arent getting insane for this OC, anymore suggestions? I will definitely keep you posted especially when the fans come in, you have been a great help, all of you.
m
0
l
a b K Overclocking
September 23, 2010 4:12:11 AM

Well it could indeed be that you have hit the limit of the chip. as stated before, they are all different and some just don't OC well. you are correct that the temps are not high, but the OC could be causing MB or NB temps to rise also. If heat doesn't solve the issue, and the voltage bumps are doing nothing, then i'd probably assume you have hit the limit. sorry to say it.

EDIT: just re-read the OP, and it doesnt sound like a chip stability issue. you still aren't bluescreening correct?

im thining it could perhaps be the PSU. what brand/model PSU do you have? it may be worth also monitoring the CPU vCore and checking for vDroop. a faulty PSU buckling under load could also be the cause of this issue. given the system is just cutting out and not crashing, i would put it down to heat or power. a bad chip still seems very unlikely.
m
0
l
September 23, 2010 8:55:34 AM

welshmousepk said:
I don't mean to discourage you from posting, just to be mindful of the content. many people come here thinking that there is some single setting they can input to hit a certain clockspeed. and its that sort of misinformation that can cause serious problems. with overclocking, more than anything else on these forums, you need to be sure of what you are saying. and to be sure that it will not be misinterpreted.

anyways, I'm sure OP gets the point now. so lets just see if he can confirm heat as the barrier here.



I got you. I will keep that in mind.. I do agree with you on the heat or power...
m
0
l
September 23, 2010 2:32:08 PM

Its not the power i just got a brand new 900w psu. I may have met the chips potential at 4.8. I will conclude that after i apply the new thermal compound, and 2 fans.
m
0
l
September 23, 2010 5:21:58 PM

AS5 has a break in time of 200 hours of normal (power on use and cooldowns) use before its at it's optimal keep that in mind....
m
0
l
September 23, 2010 11:30:39 PM

I don't understand your above post.
m
0
l
September 24, 2010 2:54:42 AM

the compound has to SET before it works at optimum efficiency.
Like when you take some elmers glue and glue some paper together. If you move it right away it hasnt set yet, You have to wait a bit till its at optimum. Same thing with most thermal compounds.
m
0
l
September 24, 2010 3:49:24 AM

I see , so it takes 200 hours of use before its optimal?
m
0
l
September 24, 2010 5:20:25 PM

Best answer selected by sockpirate.
m
0
l
a b K Overclocking
September 27, 2010 6:37:42 AM

For AS5, yes there is a cure time. however elvyn is incorrect in stating msot are the same. in fact TIM with a cure time has become very uncommon. as5 when very quickly out of favor because of it. the best compounds on the market right now have little to no cure time whatsoever.

do you pinpoint the problem after? still curious to know for sure what caused the issue.
m
0
l
September 27, 2010 8:16:15 AM

Well for some reason the recommended timings for my ram seem to be what was giving me the problems, like i was prime95 stable for like 5+ hour tests then i would play ffxiv and maybe 1-2 hours in i would get screen lock and have to manually restart.
I set everything to default other than the ram voltage, which is recommended 2.0v and slowly started raising the multiplier on stock voltage of 1.375 i can get up to a stable 3.8 OC and things seem to be working fine. I know this contradicts some of what the amd BE sticky says especially about setting the timings to the recommended settings but for some reason 4-4-4-12 was giving my system trouble and locking me up, so im running 5-5-5-18 , i know not as good as the aforementioned but its allowing my OCs to run a lot smoother whether its just a 3.8 OC or if its a 4.0 OC.

Any insight as to why this is ?
m
0
l
September 27, 2010 8:19:22 AM

ps. Temps are a lot more stable now with my 2 new 125 mm fans , bought a little spot fan for my video card , and installed a 60mm fan on the side panel also blowing onto the vid card. Prior to that i was only running the after market HS and one exhaust fan out the back.
m
0
l
November 1, 2010 7:52:44 AM

everyone wants to go at the almighty 20x... why waste your time? I'm at 14x at 4004MHz, nothing special. Stock fan. running about 55°C high. Core Voltage is 1.404. I had it this way for a month or so. The only problem i ever ran into was me tripping over my power-chord and unplugging it. lol.
m
0
l
June 20, 2011 8:14:03 PM

sockpirate said:
I am using the Zalman CNPS9900ALED 120mm 2 Ball Low-noise Blue LED CPU Cooler

http://www.newegg.com/Product/Product.aspx?Item=N82E168...
So it is a pretty nice CPU heat-sink and very OC worthy.
and i have 1 case fan, A rear 120mm.

When i was stress testing at 4.0 using just 3Dmark06 i passed it and temps were with a high cpu temp of 55C, a low of 36C, and an idle of about 40C and lower. Although when i tried running linx for that one hour and the pc shut off , it happened so quick i dont know what the temp spikes were.

At this point should i maybe think of raising the north bridge ? If so or not, what is the purpose of doing that ?

Right now i am running very stable at 3.9 at 1.42v , its just funny that that extra 100 mhz is being such a stickler to get stable. I want 4.0 to be stable for gaming, right now there aren't many games that put a constant 100% grind on the cores for any extended duration, i was thinking if it passes 3dm06 and pays fine heat wise when im gaming, i should be fine possibly?

I didn't look to see how old this thread is but I have the same cpu and I have my cpu clocked at 4.0ghz I leave my voltage at 1.48 I've never ran prime 95 at this voltage but I have ran it at 1.50 voltage for 24 hours and it was stable,however, you have a decent cooler but I do not advise you to use it when overclocking this processor to 4.0ghz. I actually recommend a corsair H50 or H70 if you are brave enough for liquid cooling and if you are not the Noctua NHd-14 is the best air cooler on the market and out performs the H50 and H70 in cooling however this air cooler is over $100.00 and it is massive, but I have one myself and in my opinion it is worth every penny. My current temp at 4.0 ghz with a northbridge of 2.2ghz is 34 celcius and when I ran prime 95 for 24 hours max temp was 46 celcius at a voltage of 1.50.
m
0
l
!