Intel socket 1156, Xeon 3440, Overclockin goodness

[MercuryDoun]

So i recently purchased a really nice MSI socket LGA 1156 MoBo from ccokeman(Thanks man!) And rather than be "boring", lol, and put an i3/i5/i7 into this Mobo. I decided it would be fun to put a Xeon 3440(i7 equivalent server processor, closest relative is the i7-860) into it instead. Now i didnt have alot of time because i started the build late, and i need it stable to play video games this weekend with my brother. So i decided to give it a quick OC right off the bat after installing windows, because i felt as though the 2.53ghz the Xeon processor had to offer stock sucked(lol 2.53Ghz hyperthreaded i7 quad-core, not good enough, i might be a nerd eh?)... So i set the vcore to 1.3v, bclk 191 and multi x19... and adjusted buncha random voltages away from Auto, and just slightly above stock(just to make sure the MoBo didnt over volt itself)... and low and behold, it booted into windows and Primed for 2+ hours now...

 

My very first attempt at the OC i did 3.0ghz at 1.1vcore @ about 45c on the hottest core... and it ran fine for about 15 minutes(know its not 'stable' after 15 minutes, but really was testing temp), so i jumped it to 3.4ghz at the same voltage, and ran fine for another 15 minutes at about 48c... So realizing now that my cooling was going to hold up(Zalman CNPS10X Flex w/ 2x 65 CFM low dB fans), and realizing that it was 12am and i have to be at work at 9am, i decide to quit playin around and jumped the voltage to 1.3v and put it to 3.6Ghz. It's been priming for over 2 hours now with no issue, and has only hit 69c on the hottest core, likely due to the massive voltage increase i gave it, which is probably WAAAAAY more than i need for the 3.6Ghz clock speed..

 

Here is my question to you all. My brother at this time will never need more than 3.6Ghz at this point in time. So its perfectly fine where its at. So should i spend some time getting the vcore as low as possible for the current 3.6Ghz OC to maintain lowest temps and highest energy efficiency? Or should i see what this Xeon can REALLY do and take is as high as i can first?

 

I guess what im really asking is does anyone care to see how well this Xeon CPU can do? Because i wouldnt mind finding out how high it can OC, but also its not my computer, and i could spend all those hours doing much more productive things than trying to melt my brothers processor. But if some people besides myself are actually interested ill see what i can do about OC'in this bad boy.

 

 

 

P.S. Im really tired because its after 2am now. And i may or may not have had a few beers due to my friends Birthday being today, so i please forgive any terrible typos.

[cjloki]

overclocking anything always interests me (not nearly as much as securities analysis) so go for it ! just remember "Heat Kills", get better cooling... let us know how it goes...

[dihartnell]

overclocking anything always interests me (not nearly as much as securities analysis) so go for it ! just remember "Heat Kills", get better cooling... let us know how it goes...

 

 

Me too but perhaps some sleep would be useful. I would hate to see you make a mistake and fry your brothers CPU. I can wait until tomorrow to see what it can do.

[MercuryDoun]

Me too but perhaps some sleep would be useful. I would hate to see you make a mistake and fry your brothers CPU. I can wait until tomorrow to see what it can do.

 

Oh its definitely waiting until this weekend or beyond. Just wonder if anyone is actually interested in over clocking the server processor. Thats the reason i juiced it so much. I needed to be sure it was going to be stable because im not gonna be able to fix it this weekend if it keeps BSOD'in or something every half hour when hes using it.

 

Next week or maybe on the Weekend. Ill sit down and start back from a much lower voltage and push it as high as i can. Looking at it, im only going to get a 19x Multi, and 20 with Turbo. So thats kind of a pain, I dont think ill be able to push it much past 4.2ghz due to an insanely high QPI speed. But either way, should be fun.

 

I dont think there is anything wrong with the cooling. When it was running full 100% load, it was at 1.344vcore... thats Alot of voltage for 3.6Ghz considering 3.4ghz ran just fine(admittedly for only a short time, so who knows if it was really stable or not) at 1.1v. But again i really just need it stable this weekend so he can use it.

 

Which brings me to question the usability of the Load-Line Calibration on this MoBo... isnt LLC Supposed to PREVENT vdroop? Because the reason i eneded up at 1.3v was because with LLC enabled it was dropping my vcore from 1.2 to 1.1 when the cpu hit 100% load... So i restarted the machine, went back in a set the vcore to 1.3, and again it went from 1.3v to about 1.2v after it hit full load. So i disabled LLC, and the core voltage actually increased from 1.31x to 1.v344 after it hit full load. Isnt that like butt backwards of what LLC is supposed to do?

Edited December 3, 2010 by MercuryDoun

[nickk47]

Interesting..I'm about to get a Xeon proc myself and I'm gonna oc the piss out of it cuz I'm curious how well it can handle more voltages and higher clocks, being a server processor, which kinda suggests to me that its suppose to be rock solid stable in stock speeds? What cooler are you using? I am gonna be using the Coolit evo internal water pump cooler

Edited December 6, 2010 by nickk47

[cjloki]

Which brings me to question the usability of the Load-Line Calibration on this MoBo... isnt LLC Supposed to PREVENT vdroop? Because the reason i eneded up at 1.3v was because with LLC enabled it was dropping my vcore from 1.2 to 1.1 when the cpu hit 100% load... So i restarted the machine, went back in a set the vcore to 1.3, and again it went from 1.3v to about 1.2v after it hit full load. So i disabled LLC, and the core voltage actually increased from 1.31x to 1.v344 after it hit full load. Isnt that like butt backwards of what LLC is supposed to do?

 

llc doesnt "prevent" vdroop, it reduces it...

you want to have some vdroop, and in fact it's an integral part of mobo/cpu design.

google search vdroop results

but there is a trade off, and that's what you're realizing in your testing...

when benching i set to standard or disabled depending on mobo...

[MercuryDoun]

Interesting..I'm about to get a Xeon proc myself and I'm gonna oc the piss out of it cuz I'm curious how well it can handle more voltages and higher clocks, being a server processor, which kinda suggests to me that its suppose to be rock solid stable in stock speeds? What cooler are you using? I am gonna be using the Coolit evo internal water pump cooler

 

My cooler is a Zalman CNPS10X Flex with 2x 65cfm fans. Coulda got better fans, but it needs to be sorta kinda quiet too. Right now its sitting at 3.8Ghz at 1.3v. Im going to spend just alittle bit of time to get it sorta stable at 4.2 or 4.4 depending on where my cooling lets me go. But 3.8Ghz is the goal of the processor. And so far with the quick OC i was required to do, its rock stable at 1.3v, im going to do my best to REALLY lower that voltage down. Its not that the temps are bad, i just dont like the idea of using a quick and dirty OC for an extended period of time.

 

With the Xeon's, Beware of the locked Multi's They keep them kind of on the low side, the 3440 has x19, and 20 with Turbo. So thats fine, but ive seen some with only 17 or 18 too. Which could give you alot of lip hitting something up there. I did try real quick at 4.2Ghz but i dont think 1.32v was enough for 4.2ghz, if thats the case, then my i7-930 is already superior, but they are also different sockets so maybe its not fair to compare them? But ill have to see if some other settings pan out to make it stable.

 

But really i haven't had very much time to sit down and have any real time to play with the Xeon and its OC. Hopefully this week i can get alot done. Ill come back with some real results once i have them.

 

 

llc doesnt "prevent" vdroop, it reduces it...

you want to have some vdroop, and in fact it's an integral part of mobo/cpu design.

google search vdroop results

but there is a trade off, and that's what you're realizing in your testing...

when benching i set to standard or disabled depending on mobo...

 

This i understand, but in my Asus x58 chipset MoBo, LLC enabled works exactly like id expect it too. It droop's just alittle at all times. But it almost never changes from 0%-100% load. Like if i set it to 1.1125, it will droop to something like 1.10xx which is what i currently have my 930 at now. BUt the voltage never changes aside from the fact that its slightly off right from the start. Turning LLC on this MSI socket 1156 Mobo is basically just saying, "Hey take whatever voltage you set in your BIO's and minus .1v because thats what im gonna do" Doesnt seem to make sense to me IMO. I dont think i had any other options besides enables and disabled for LLC. Ill have to look again and see if there is a standard or Overclock setting or something, maybe thats what im looking for.

[cjloki]

.... if there is a standard or Overclock setting or something, maybe thats what im looking for.

 

gigabyte llc bios options: auto, standard, level 1, level 2,....

[MercuryDoun]

gigabyte llc bios options: auto, standard, level 1, level 2,....

 

OK thanks, i think in my Asus Mobo, there was Disabled, Enabled, and Overclock... I dont remember what this MSI MoBo said. Ill have to look when i get home.