Help needed on previous overclock

[Dlaw]

I did a thorough cleaning of my sig system today, and after putting it all back together, I ran Prime 95 to give the new TIM a quick burn, and it locked up almost instantly.

So, I started messing with settings, and this is what I have so far;
CPU Freq. = 240
Multi = 17
VCore = 1.450, up from 1.385
NB Freq = 2400
NB Voltage = 1.1250, up from 1.1125
HT Bus = 1920
LLC = Enabled (options are enabled, disabled, off)


Temperature under load after 10 minutes was 57C at the socket, and 32C at the cores. I can't seem to manage to get it to run any longer. Do I need to up the voltage more? It seems like a hefty bump, especially since I've almost added 1/10 of a volt to my previously stable overclock. If I can get a screen shot of my next P95 run, I'll post it.

 

[RGone]

You did not save your old settings as a profile? My memory gets sort of weak after awhile and I think I had one setting and it was not what I seem to have to use now.

Sounds like you need to up the Vcore for sure. Be a little easier if we had the 3 CPUz captures and the HWMonitor capture to see what it is logging during the P95 Blend run. You could have ram set-up differently than before...it is just iffy to say what is the difference without a little better data.
RGone...

 

[Dlaw]

I did save the old settings, and I also wrote them down. The settings in the profile matched the ones that I wrote down, but P95 failed almost immediately.

I'm going to do a run at stock to see if that's actually stable. It looks like I'm going to have to redo this overclock.

 

[Dlaw]

The stock run was successful. Here are the pics:

 

[Dlaw]

So, since I had to redo all of this anyway, I decided to take a few steps that I had skipped last time, including trying to find my max overclock with the two locked cores unlocked. I found my max FSB, which turns out to be somewhere between 250 and 255. I'm currently running at 250, with a multi of 15.5 for 3.875GHz while running as an x6. Here are the screens after 20 minutes:


CPU VCore = 1.4v with LLC enabled


CPU-NB voltage = 1.1250v
Memory voltage = 1.55v





The difference between core and socket temps according to the stock test I ran this morning is ~16C, so it appears as though the core temps are currently ~49C

And in case anyone wants to see them, here are the screens from my previous overclock:









Edit: core #6 dropped out after 36 minutes. Going to give her a vcore bump and hope it doesn't push my temps over the edge.

 

[RGone]

You keep very good records Dlaw. According to the before clean-up HWMonitor you were running more Vcore under load than you are now with 6 cores. It could be that LLC is working a little different this go around.
RGone...

 

[Dlaw]

Well, the vcore bump didn't do it. I started prime, went outside to smoke, came back and checked on it and it was good. Then I went and took a shower and when I came back it had restarted on me. LLC does seem to be acting very different, as before it would push much harder than it does now.

My question is, could that be due to the PSU degrading? Or is it more likely that it's the board or CPU?

I'm gonna try one more run with another vcore bump, previously it was ~1.415v, and let her go while I go shopping with the family. Hopefully this will do it.

On a side note, it seems to me that now the LLC on my board is acting more like the LLC on others, am I mistaken?

And thanks RGone, my PC health pics are in there own folder in my pictures library, and I have a specific notebook just for taking PC related notes. I did that for just such an occasion as this, lol.

 

[caddi daddi]

oh,man! don't listen to Rgone.
get out here on the edge of the cliff with me, don't keep notes that makes rescue and rebuild way to easy......

 

[RGone]

No man it is he you cannot listen to...

Anybody that uses an 800 + Horsepower 4WD Porsche to go get cigs and then has to drive 4.5miles back over his dirt road to get to the cabin to smoke in the yard...well he ain't all there as some might say. That is C_D. Hehehe.

Nevertheless you did keep dang good records Dlaw.
RGone...

oh,man! don't listen to Rgone.
get out here on the edge of the cliff with me, don't keep notes that makes rescue and rebuild way to easy......

 

[Dlaw]

Thanks guys, lol. So, my system is 3 hour and 36 minute P95 stable!!

Here are the shots:









Vcore is now at 1.425, before it was at 1.4125. I'd say that's as far as this baby is gonna go for me. What do you guys think?

 

[caddi daddi]

I THOUGHT THE THUBANS LIKED THE CPU/NB CLOSE TO 2000?????
and the ht link even with it.

 

[Mandrake4565]

Anybody that uses an 800 + Horsepower 4WD Porsche to go get cigs and then has to drive 4.5miles back over his dirt road to get to the cabin to smoke in the yard...well he ain't all there as some might say. That is C_D. Hehehe.

Off topic but this is probably why Caddi and I think alike More V Core!!!!

 

[RGone]

Off topic but this is probably why Caddi and I think alike More V Core!!!!

No not another hollering pour the coals to 'er matey.
RGone...

 

[trents]

I THOUGHT THE THUBANS LIKED THE CPU/NB CLOSE TO 2000?????
and the ht link even with it.

That ain't so but it is true the Thubans like to keep the CPUNB and HT Link frequencies close together.

 

[Dlaw]

I can't get mine stable with the HT link much above where I had it in my previous overclock. Maybe I have a weird chip?

 

[RGone]

From previous overclock HWMonitor, it shows Max 1.46VCore @ 4 cores 4080Mhz. Temps 65c and 36c.

Now showing Max 1.46Vcore @ 6 cores and 3875Mhz. Temps 66c and no package temp due to unlocking of the 2 cores.

I would imagine you could now lock the thing back and push the cpu mhz to just north of 4.0Ghz like you had it before and all would look just about identical to your previous speeds.

Unlocking those cores can actually unlock a weak core and throw the calculations off as to what is going on. But you could get 4.0Ghz out of the thing unlocked if the CPU Temps were lower. They are not so temperature seems the limiting factor unlocked. Bumping 3.9Ghz with 6 cores should do plenty of real work.
RGone...

 

[caddi daddi]

looking at the "cputin" temp witch is socket temp, it looks like he needs a good bit more airflow to the vrm section. it's not going to help with the clock but might help keep the vrm section alive.

 

[RGone]

looking at the "cputin" temp witch is socket temp, it looks like he needs a good bit more airflow to the vrm section. it's not going to help with the clock but might help keep the vrm section alive.

Yeah man I saw that as well, but did not say anything since I guess he is more or less trying to return to his previous settings. Although now he has the other two cores unlocked and it would not hurt to keep the VRM's cooler if he stays in unlocked mode and does a lot of work. Sure as heck would not hurt to see the cpu temp go down.
RGone...

 

[Dlaw]

Yea, my main limiting factor right now is where the computer is located. The only room I have for the box is in a cabinet that I took the door off from in the built in desk in my house. I'm waiting to save some money up and redo the whole desk, so that I can have it sitting a little more open. Here's a picture to show you what I mean:

Sorry it's so big.

 

[Dlaw]

So, should I be worried about the 66C socket temps? I wouldn't be against locking the two cores back up and pushing it back to 4+GHz, I just figured that since I had to do all of this work again, I might as well see what I could do with them.