SOLVED help getting a 955be to 4.5

[BeepBeep2]

the "pip" on the top of the cpu and two fires have been just a little tough on the poor thing.
i guess the next time i have it out i'll just have to take a pic of the top of it.
i really don't really think anyone will want to buy it after seeing that.

I would, after I get my 955 fully benched on LN2 and sell it so I can buy yours. I'm trying to find a better one, and normally you cant find good ones used since there was a long string of time from 09 to 11 where improvements were made in the process without a stepping change. Better off buying new than buying random used. Yours is strong though...really really strong and I would love to have it. But if ya keep beating on it, no chance. If you say something like "never had over 1.7v on air" people will be like "what's wrong with you"...and I'm already thinking that myself. You have a gem of a gem of a gem of a chip there and it is taking much unnecessary abuse.

In fact, I'd trade you an FX-8150 for it if I could get a few $$ on the deal. Would allow you to upgrade and really satisfy me! I will PM you with more details, if interested. Don't want to deal anything out here due to forum rules and I already said too much.

I would probably lap it again though, personally.

 

[caddi daddi]

the intent here is to get a 955be to run 4.5 p95 for two hours with no errors, if i ever get that done i'll figure out something else to do with that thing.
I never got to the computer store sunday because i had to fix that issue and while fixing that i also changed from an hdd to an ssd i bought a while back.
so, i am starting again.

 

[BeepBeep2]

caddi, I don't think you get it.
There is a point where increasing voltage makes clocks go backwards. If you can not get 4.5 under 1.625v, give up, because I'm telling you right now, it is not going to go faster if you give it more voltage. It will only go slower.

 

[caddi daddi]

ok so if i give up on this, what would i do with the thing, stare at it? it's one of the ugliest cases i have ever seen.
I had to try the extreme voltage just to get that out of my head.
you might want to read the thread, 'my bone pile before trying to reuse something i have ever had.
i have hed a few fx cpu, been everywere with those. but pm me further down the road after i throw in the towel on this project.

 

[BeepBeep2]

Well alright.
You still only need 1.55v to run 4114 MHz through Prime95 stable for several hours?

A common phenomenon when pushing voltage is that the CPU acts "sick" as we extreme overclockers describe it.

I don't know what week your 955 was made but out of 8-9 Phenom II 955/965 I've gotten to look at that were 2012 weeks, all of them really didn't scale much with voltage over 1.8v on LN2 or 1.5-1.55v on air/water, all the while older 2010 models scaled up to 1.9v.

I know a former buddy of mine "chew*" tested quite a few through 2011 too and they didn't scale with voltage.

AMD changed something in the manufacturing process mid 2010 that made the chips less voltage tolerant, but also seemingly cooler running and more efficient, we believe they implemented the low-K dielectric that Thuban had.

 

[caddi daddi]

where do i find a breakdown of the codeing on the cpu?
my 955 was bought in early/mid 2011 as best as i can remember and sat dormant (in my first build that would not boot for quite a while) so i am really only clocking a phenom 2 for the first time.

 

[BeepBeep2]

On the CPU IHS...

Contains model name, serial number, batch number/batch type/facility and week of manufacture.

If you bought it early mid 2011, then it is probably late 2010 or early 2011 weeks, about the time when Phenom II started hitting 7 GHz+ on LN2...which gives me more reason to be interested in your CPU.

 

[caddi daddi]

ok, got my first 10 ibt turns at standard setting.
guess i gotta fix something in the attaching.......

 

[caddi daddi]

and i'll back that up with this one.

 

[bassnut]

Nice caddie daddie .... we knew you could do it.

 

[RGone]

Nice caddie daddie .... we knew you could do it.

Me knowed he wuz gunna dues it or bust he ole gut trying.

Hehehe.

Congrats Caddi Daddi !!!


RGone...ster.

 

[caddi daddi]

thanks guys. that's just the first wrung on the ladder to 2 hours p95.
Rgone, thank you so much for spending the time on the phone coaching me along, without that i would still be at 4.3..........

 

[RGone]

Shett man I appreciate the kudos, but any man that run that 4-wheeled monster you got, can tweak a freeken lil ole cornputer foreshore.

RGone...ster.

 

[BeepBeep2]

IBT or LinX 10 pass takes...4-5 minutes at most.

It was kind of a given that you would be able to pass that, but 4-5 minutes of IBT is nothing in regard to stability.

caddi, whats the highest frequency you can bench Cinebench R11.5?

 

[caddi daddi]

beepbeep2,
that is the first rung of the ladder, that was the easy step.
now i step up to high in ibt and then to max.
after i can run it at max setting 10 times I move to the next step i then move to prime 95.
building blocks is the way i have to learn things.

 

[BeepBeep2]

beepbeep2,
that is the first rung of the ladder, that was the easy step.
now i step up to high in ibt and then to max.
after i can run it at max setting 10 times I move to the next step i then move to prime 95.
building blocks is the way i have to learn things.

Ok, but if you already found that it was unstable in P95 at any voltage what do you expect to change now that you passed 10 pass of LinX?

Good luck.

 

[caddi daddi]

heck if i know what i'm going to have to change, I had no idea what would have to change to get that step, so I'll start with a fsb at about 184 and start testing from there.
I can see one thing that is surly going to make this a slow go, I need the machine to be dead cold but the fans warm to get max airflow through the rad. that might take an open bench or a second power supply for the cooling system. right now i get about two attempts an hour due to wait time for the whole package to cool back off.

 

[caddi daddi]

first 10 minute run under prime 95.
I have opened the door!!!!!

 

[BeepBeep2]

Interesting, see you lowered temps.

Found out early on 10c difference means a good 100 MHz in clocks, I see you lowered temps and low and behold it looks good.

You are abusing that CPU hard though, 1.62v is what is normally used for...no more than 5 second CPU-Z validations on normal cooling

 

[liquidfire]

This is an interesting project how's prime going? What's the max temp your seeing?