So, I was trying to overclock my i5-2500k sitted on Asrock p67 Pro (not pro3) but as soon as at 4GHz I start to get these strange system shutdowns where power is simply cut off out of the blue. Then after few seconds PC powers on automatically. Hard cycle I believe it's called? First I thought that it might be PSU but even more strange thing is that I can play games fine and without any crashes but as soon as I quit the game and return to desktop, shutdown follows soon with 80% propability. It may also shutdown when I'm browsing the internet or doing something other non-demanding. Sometimes full day may go by without shutdown but then it happens again the next day. After shutdown it may sometimes shutdown again after loading up the Windows desktop. Shouldn't bad PSU cause shutdowns under load and so that it hits the fail safe on and power switch must be cycled at the back before PC can be turned on again? So this kind of shutdown with automatic reboot might point more towards CPU or mobo? Btw. max temps are 60C for CPU and 45C for mobo, but I think it's not that either as shutdowns aren't happening under load. Now I have kept CPU at 3.9GHz where it seem to be stable without shutdowns and I have also monitored voltages with software that shows min/max values but so far they seem to be well within tolerances. Of course it would be hard to see voltages at the moment of shutdown because I haven't found software with usable logging function. Otherwise I'm pretty much at loss and I don't have any spare parts to switch and test things separately. Either I try my luck and buy new PSU (current one is about 5 years old) or maybe the mobo simply sucks for overclocking as it doesn't have that many power phases (I think 4+1 or something by looking at the mobo). I overclocked using following guide: http://www.overclock.net/t/1198504/...-guide-sandy-bridge-ivy-bridge-asrock-edition So I haven't touched any BCLK settings and only upped the CPU multiplier. Vcore seemed to go up automatically with multipliers so I had to lower it a bit using negative offset voltage, using Intelburntest to find the lowest stable voltage which is now about 1.150V under load and about 0.900V idle. But raising Vcore or adjusting LLC doesn't seem to affect these shutdowns. I tried to go for basics by reseating all the cables and replaced the old PSU power cord and connected it straight to wall socket instead of (also old) power strip. I almost thought that it solved the problem as I didn't get shutdowns for 2 days running at 4GHz but then on 3rd day it suddenly shut down again after I finished playing game and went to browse the internet. Has anyone else had similar hard cycle problem and did new PSU solve the issue?
Check if you have EIST C1E C3E or other power saving features enabled. Also the best way to OC this chip is to disable all except EIST and use offset voltage. Oh and always re-test new settings with Prime95 running in background with low thread priority while doing usual stuff in foreground. Also never use very high load balancing. VCore in Sandy Bridge is supposed to decrease when under load. If you use super aggresive load balancing, computer might restart or do shutdown cycle... Also you can try DEcreasing PLL voltage to somewhere around 1.7V PSU is mostly an issue under full load OR when starting computer (the first few seconds). All in all it's a lot of things, best way is to reset it do defaults, disable all power saving features except of EIST, keep memory on lowest settings like 1333MHz and just keep increasing offset voltage and multiplier.
I had only EIST and C1E enabled like that OC guide advised and LLC level 5 which is lowest. When I disabled C1E now it seems that CPU never drops to idle state and idle temps are few degrees higher because Vcore stays up. I try to run it like this for a while to see if it still shuts down, maybe mobo/CPU can't handle light load voltage transitions too well because it never shuts down during full load.
I've always disabled the power saving stuff until I was sure the overclock was stable.Then you enable them and see what effect it has so you can narrow down whether that's the cause or not.You must have a good chip if you can get 4 out of that voltage I always need 1.2 for 4 and it went up rapidly after that.1.3 for 4.4 and after too much for the gains.
Can you have power savings enabled at 4.4? I used Asrock Xtreme Tuning Utility in Windows to lower Vcore step by step and test stability in between and it seemed that I could have gone a little lower on VCore but PC didn't start up if I used more than -0.130V offset voltage in BIOS. But no much joy having low voltages if it still keeps shutting down for some weird reason
Yeah you can after you make sure you're stable.I ran stress tests for a quick check and then run it it for a while doing everything I usually do as I find stress tests don't always give an accurate idea over the long term.I do everything from the bios as I don't like doing it with those utilities.
I have EIST enabled at 4.8GHz. This way OS is in control and frequencies go all the way down to 1.6GHz and 1.08V when in idle. Also EIST is DISABLED for first 90 seconds after the boot into OS. It kicks in after 90 seconds so don't get puzzled by that behavior.
You mentioned plugging directly into the wall socket and a "power strip" so for all we know it's just normal variations in your power. Without a UPS you just don't know. I see more power variations in periods of heat and cold. I've never seen a power cord go bad. You said you used an online guide to set your OC. It's important to only follow the method that is suggested and not to copy other peoples' settings, since every system is different. So you're overclocking and undervolting and wondering why it's unstable??? The guide you mention recommends 10 minutes Prime95 runs to test for stability, and as a final test a 1 hour run. Clearly, this is not always indicative of total stability. Did you pay attention to this part of the guide you followed: "I do not recommend negative voltage: The offset affects idle voltages per multiplier, and Intel has already set the voltage for the idle multiplier. You don't want to drop below what Intel has already tested for you." Usually when I see this type of thing it's the mobo, and not the PSU. I'm definitely seeing q67/q77 mobo's coming back defective at this point(office machines with lots of hours, but never OC'd). Many(most?) 2500k's do 4.0 at stock settings - maybe you got a dud? Did you buy the system new, and are only now attempting to OC - or is it a used setup that your buddy overvolted for 5 years? I've never had a problem with using SpeedStep while OC'd. Recently I saw a Q77 board that was resetting randomly. After testing components separately, I noticed that any OS would crash on this board if any GPU was in the x16 slot. My point is that your problem could be caused by lots of things. If you didn't have problems before overclocking, then the obvious solution is to return to default(or at least more conservative) settings.
Or play around with voltages which is half the fun.Keep temps under control really no danger.A lot would do 4 with stock but definitely not most.I agree with the every chip is different and guides should be just that and not a bible.If you're wanting to overclock without playing with it a bit don't bother unless it's super conservative.
Well I'm not straight copying frequency settings etc. but just using settings that are supposed to enable best overclock. I went up from lower frequency step by step and always tested for stability. I said that raising voltage doesn't seem to affect to these shutdowns. How are you supposed to overclock sensibly anyway if using that +0.005V offset causes voltages to automatically go up through the roof along with multipliers? Using 40x with +0.005V offset load voltages seemed to go over 1.3V already and increasing per each step so I don't think people are running their 5GHz systems at 1.3-1.4V using only positive offset because voltages and temps would be crazy. Or maybe this voltage bumping is just some kind of Asrock thing, I don't know. I bought CPU and mobo used but I have been running it with mild 3.7GHz overclock for a while without problems. I know second hand stuff may and do have wear and tear and they aren't like new. And maybe there is some degradation causing overclock to fail and I'll just have to accept it that this specific CPU/mobo can't go more than 3.9GHz. Unless I get another mobo or CPU there is obviously no way of knowing which one it is that's bad.
For a 40x multipler, very little should be needed actually. Also, did you forget to disable Turbo Boost perhaps? Normally, when changing the CPU multiplier, TB should automatically get disabled (I think), but it doesn't hurt to make sure and disable it manually. In case it helps, my 2500K runs stable with a 42x multiplier using: Turbo Boost: Disabled All power saving features (EIST and whatnot): all "enabled", "auto" or "default" (whatever the default setting of the BIOS is) VCore: 1.19v XMP: Enabled VDroop: Auto LLC: Auto BCLK: Auto (should be ~100, the default for Sandy Bridge). So basically, all I did was change VCore from "auto" to "1.19" (after overnight testing sessions with prime; 1.18 was stable but I figured slightly higher than stable is best) and disable Turbo Boost. Everything else is on "auto" or "default". Are you overclocking you RAM? I've heard it can severely hinder CPU overclocking. For my 4.2GHz OC, I just use XMP for the RAM (which sets 1600MHz for RAM) and 1.5v RAM voltage (because XMP would set it to 1.6v which is too high for Sandy Bridge and 1.5v is stable here.) To see if RAM hinders you (although I doubt it), try disabling XMP, which will set your RAM to 1333MHz (the native RAM speed for Sandy Bridge.)
Yes they are. And believe me that with I7s and their hyper threading temps go significantly higher when running all 8 threads. 5GHz Prime stable SB (i7s + HT at least) running on <1.4V is very rare. I'd say golden sample if it exists. Also all SB chips are designed for maximum 24/7 voltage of 1.4V so only problem is temperature, you just need proper cooling. Using negative offsets is just very bad idea especially in combination with EIST or C1E. I'd strongly advise against it.
If I remember right I wasn't able to disable turbo boost for some reason and it was stuck in "enabled". But I think it's off because CPU frequency doesn't raise under load and is 4GHz max at 40x100. So you are using fixed voltage and it stays always more or less the same? Maybe I try fixed voltage too as it doesn't differ much from current situation where I have all power savings disabled and voltage stays the same using negative offset and never drops to idle. Currently RAM is 1600MHz and 1.5V but I guess it doesn't affect much performance wise even if I drop it to 1333MHz. Altough I think RAM frequency shouldn't matter because it always stays the same and doesn't change, but who knows OK so I try fixed voltage then and keep EIST enabled (other power saving stuff disabled). At least that way idle voltage shouldn't drop too low in case that's what causes shutdowns. Thanks for the tips chaps!
You're looking at the status screen which just says if it's enabled or not. You need to go to another page in the BIOS to go to the option where you can disable it. I think it's a per-core setting, so you need to disable it for all cores. It makes a huge difference in some games. Fallout 4 and Watch Dogs 2 for example have a big dependency on RAM frequency. In Watch Dogs 2 for example, going from 1600MHz to 2400MHz makes a difference of about 25-30FPS (going from 50-55FPS to 75-80FPS.) Fortunately, most games do not depend on high RAM speeds. F4 and WD2 are the only ones I know of, actually.
I've seen 1 in the 20+ years I've been building computers.... Coincidentally, that was almost 20 years ago.... 3.7ghz isn't an overclock at all. That's the default maximum boost frequency set by Intel for the chip. You should have no problem hitting 4.1ghz on stock voltages. Even the regular i5 2500 could do that. The P67 Pro is just garbage. It is a 4+1 board. You should have found a P67 Pro3 or P67 Extreme3 at minimum. A Z77 Extreme4 would have been better though. I ran my i7 2600K with HT enabled at 4.5ghz at 1.35v..... Peak load temp of 65C on water (improperly setup) and 72C on air. I even had the power saving features enabled. For the record....an i7 2600K runs quite a bit hotter with HT enabled
Run furmark GPU torture test, it will heat your GPU and put some pressure on PSU. If your PSU survives (no reboot), there issue in CPU, otherwise its PSU.
I just thought that it might be the problem because my power cord stuff are maybe 10 years old or something. Oh yeah indeed so I should say that now I have mild overclock at 3.9GHz. No wonder there seem to be nothing about this P67pro mobo on internet so I suppose this was some kind of first failed prototype from Asrock and they quickly pulled it out from the market. And no wonder that I got it so cheap as those used Pro3/Extreme4 mobos cost almost more than new lol. I keep testing 4GHz using fixed voltage when I got the time and if it still shuts down then I think it's simply crappy mobo. Does that put more load on GPU/PSU than playing BF1 using max settings, for example? Because I can play completely fine without problems for hours and hours but shutdown occurs soon after I quit game or even if I don't play at all and just browse the internet. So doesn't seem like PSU problem to me, but what do I know lol.
System can crash due voltage / power changes. I used to believe PSU are fine, but I had one of PCs that kept crashing. Replaced motherboard, video card... turned out it was PSU. If you system will keep crashing under furmark, it would be more obvious that its not CPU causing the problem. Leaving it to PSU, GPU or motherboard. PSU issues are difficult to narrow down. Some times crashes obvious by adding load, sometimes it just craps itself for nothing. I just wouldn't be sure PSU is OK in your place and would check it with replacement (borrow from shop/friend/etc...).
Power cord failures are extremely rare and usually only occur due to physical damage to the cord. The P67 Pro wasn't a prototype. There's little information because it was more an entry-level or budget P67 based board. Even though it has the features available, it wasn't intended for overclocking.
So I have used fixed voltage of 1.22V for couple of days and 4.2GHz seems stable without crashes or shutdowns so far (but let's not get ahead of ourselves lol). I tried to enable speedstep and C1E but it caused system to freeze so I keep all power saving stuff disabled. If fixed voltage solved this shutdown issue then I guess this is the highest sensible OC for this mobo as using 1.25V couldn't do 4.3GHz and lowering PLL or DRAM frequency didn't help either. Altough I haven't messed with LLC much other than briefly trying lvl5 -> lvl3 but it didn't seem to make any difference. Edit: Still OK, except that one time while starting up it seemed to hang at BIOS check (can't remember what code that was at lower right corner), I pushed reset and it started up normally. If nothing serious comes up I'm satisfied with 4.2GHz and with GTX 970+16GB it should suffice for few years until another game title comes up that chokes up my PC at lowest detail settings and forces me to upgrade
