Discussion in 'Article Discussion' started by arcticstoat, 9 Mar 2011.
I love mine - working great. Massive improvement over the previous i7 i860 machien I had before
Heard from Scan over the weekend. Components were shipped Friday afternoon and will have them on Monday
I'm not really satisfied with the Gigabyte P67A-UD4-B3, overclocking on this board is nearly impossible... On stock it's ok, but at ~4.5-4.7GHz it throws out BSOD 124 all the time at random intervals.
Thank you, but I was wondering if it would fit length wise not height wise
Length of what ? 560Ti ? Do you realize Define R3 has no problems with HD5870 length and with some forcing even 5970 fits.
no offense, but you shouldnt blame the board due to your lack of OC skills. you simply need more voltage somewhere.
i have the same board (minus the b3 p67) and i barely had to push to get to 4.5ghz prime stable.
Lack of OC skills. Yes, of course. 4.3GHz works on VCore 1.3761V. 4.4GHz BSOD's after a hour or so. Lack of OC skills, of course.
RAM is at 1.53V, with 1.5V set in BIOS and 1.5V needed for RAM. Memory controller voltage is at default (1.05V), pushing it to 1.15V made no change. Still BSOD code 124 for anything over 4.3GHz.
Right now i run at 4.3GHz voltage set to normal + DVID+0.000V, this pushes the CPU to the previously mentioned 1.3761V and this one is stable (5 hour uptime so far). This is with everything turned on except turbo (means HT, EIST, C1E, C3/C6 on).
I tried with C1E/Turbo/Hyperthreading/Overheat protection turned off, voltage set to 1.4V (1.38V real), memory controller at 1.15V, memory voltage at 1.6V, everything else left at defaults. 4.4GHz random BSOD, 4.5GHz random BSOD, 4.6GHz random BSOD, 4.7GHz random BSOD, 4.8GHz Windows fail at boot.
Maybe it needs more than 1.38V real for more than 4.3GHz, but i'm not willing to send my 2600k to RMA in 2 weeks because of more than 1.4V VCore.
If you have so much knowledge then please, tell this stupid man with no OC skills where is he making a mistake.
lol. obviously the term "no offense" doesnt take where your from.
taking on a patronizing tone will get you nowhere
Well then please explain where i lack some voltage. Maybe there is some super hidden setting called "stable OC", but unfortunately i'm unable to find it. Each and every OC guide i did read said you turn off C1E, Turbo etc (done); set the multiplier to the required frequency (x44-x48, done), set CPU Core voltage to the needed level (1.4V, done; this is afaik the limit you can take without risk of damage to your CPU), set the memory controller voltage higher if needed, but not above 1.2V (1.15V, done).
What voltage is the one you are refering to ? Setting Vcore to 1.45V ? No thanks. I want to use my cpu for more than 2 weeks.
Edit: And don't forget i talk about 2600k, which means hyperthreading turned on. That is something you don't have on 2500k, which can also affect the stability of OC (and it can show up bugs in board which are not present with 2500k).
Edit2: And you know, the worst part is that you boot at 4.6GHz no problem. Start OCCT and run Linpack for a hour, no problem. You start idling, no problem. And then later, just for fun, it BSOD's out of nowhere, with no reason. If it would be predictable like "ok, it fails at load" or "ok, it fails at idle", then i say it's fine, i know where the problem is. But when you have voltages pretty much maxed out, it is ok at idle, it is ok at load, yet it is doing BSOD and someone says "you simply need more voltage somewhere", it makes you at least irritated.
i understand your frustration, but the chances that you got the only sandy bridge cpu that wont hit 4.5ghz are pretty slim. the problem must lie in your settings.
if i were you, i would start over. loosen your RAM settings considerably to rule that out. turn off all power saving features and HT. since my fiancee is on her computer ATM, i cant go through and check what settings im running, but i know its only running a 1.35 vcore. whats your vdroop look like? my board has a pretty high droop even with LLC enabled.
Right now, as i said, i run on VCore Normal with DVID + 0.000V, which means the standard voltage for 4.3GHz. Real voltage is :
If you mean what is the voltage when i had fixed voltage set in BIOS, then it was 1.380V in idle and 1.366V at load when i had 1.4V set in BIOS.
thats what i meant. thats a much better vdroop than im getting with my board. is that with LLC enabled?
Ok, it seems that was with 1.38V set at BIOS, because right now i'm at 4.6GHz with everything turned off, running RAM at 1333CL9. 1.4V set in BIOS, while "idling" it is at 1.392V and while OCCT it drops to "1.380V".
BIOS Settings are :
So now it's down to one of the following issues :
1)Hyperthreading - which would be very unfortunate as this is the selling point of i7 2600k against he i5 2500k.
2)memory controller has serious issues running OC together with memory at DDR3-1333 at CL 7. Considering there is serious difference between CL7 and CL9 at DDR3-1333 (see the bit-tech review), this would be again unfortunate, but not as devastating as issue no 1.
3) virtualization support - to be honest, i forgot to turn this off (didn't noticed it), this is absolutely irrelevant to me, so if it would be this option, then i honestly wouldn't care.
The biggest issue with the BSOD code 124 is that it is pretty much impossible to detect it thanks to the total random nature of it. For example i'm now at 21 minute uptime with settings from these post, but is it really stable ? Won't i get a BSOD after 2-3 hours ?
FYI the CPU can do x48 as it POSTS, just it isn't stable enough.
hyper threading will always lower your OC headroom. im not familiar with the 2600k enough to say what the maximum speed with HT should be. if it fails prime with the current settings, id try it without HT. at least that will tell you for sure if thats whats causing your instability.
IIRC, ive also upped my QPI, system agent, and PCH voltages a little from their stock settings.
I think you missed the most important part about that BSOD - it is random. It is not connected to load testing or idling, it occurs randomly. You can do a load testing for a hour and it will not BSOD, you can leave then the computer idling for a hour and it will not BSOD, but once in a while you will get BSOD at load, or at idle, or when you are doing something....
That is the most frustrating part of the whole thing.
but it doesnt do it at stock speeds?
1 hour of prime really isnt long enough to be conclusive. i let it run over night before i begin to get my hopes up for success.
Exactly, it is rock stable at stock, up until 4.3GHz (including) with all features turned on and auto voltage.
Anything more and it becomes problematic. 4.7GHz is mission impossible (1.40V BIOS/1.382V real at load) is ending in BSOD in 3-10 minutes now, with all features turned off, regardless if HT is on or off. 4.6GHz seems stable for now, testing it with HT on for now (with HT off it went over 30 minutes of Linpack no problem).
Maybe with >1.4V voltage 4.7GHz would be doable too, but as i said, i don't want to kill my CPU in 2 weeks and considering i don't think i would have that bad silicon (especially if it POSTs at 4.8GHz setting, it is just not stable enough to finish Windows booting). So if it is not the CPU, then the only other option is it's the board.
whether the system will post at a given speed or not doesnt determine if the cpu is capable of running those speeds without errors. and without the voltage to support it, the speed is irrelevant.
as i said before, testing with HT off will simplify troubleshooting. once you get the system stable without HT, then you can worry about fine tuning it with HT enabled.
i would suggest pushing more voltage to other areas of the system. QPI and system agent voltages for sure.
Oh well, back to square one. 4.6GHz, 1.4V VCore, HT on, everything else off, DDR3-1333 CL9, auto voltages everywhere else - and it did BSOD after 2 and half hours uptime, out of blue, at simple browsing, without any significant load. Before that OCCT Linpack was running for a hour and half without any sign of error.
4.3GHz, everything auto, everything feature and power saving option turned on - works no problem.
So unless i got a super bad i7 2600k, the only other possible source for my issues is the board and the F1 bios. RAM is out of question, i was using it for over a month without any problems on my P55 setup.
4.3Ghz is usable speed for now, will see if the new BIOS will help.
I wish I could find an e-tailer selling a socket 1155 mini-ITX board. Have looked high and low. Some websites say they have stock but when you contact them they admit they don't!
Separate names with a comma.