odd I7 overclocking behaviour

robinlawrie

robinlawrie

robinlawrie

Associate
Joined
26 Oct 2009
Posts
93
Location
London
Hi all.. i recently got one of the i7 4 gig bundles from ocuk, the one with the ud5 motherboard, some patriot ram and a noctua cooler.

i completed the system with a gtx285 2 gig, 2 corsair p128s in raid 0 and a 1tb hdd for backup. its a very nice system and amazingly fast, but since i got it i had frequent cold boot problems, until i loaded the memory profile in the bios. this fixed the cold boot issues, i think 1t command rate was on, 2t is stable.

however it wont run intelburntest for more than about 20 seconds before bsod'ing *strangely it does this on auto (4 threads) but if i manually set 8 threads it completes a run of 5.

i did a little reading up and tried boosting the vcore initially, then the various other voltages.. tiny little bits at a time, but increasing the voltages had one effect... boot crash. the temperatures on the cpu never went above 72 degrees whilst running linpack or rendering (i do 3d..)

i then began to suspect my psu.. it one of the few not-new components in my pc. its an OCZ powerstream 600.. cost a packet 5 years ago, and since i was running a 150watt cpu and 2 6800 ultras on watercooling then, i assumed it would handle this rig.. but it seems to me that maybe its just not got enough grunt on each of the 2 12v rails..20a on one and 18a on the other.

im gonna order myself a corsair hx650 which has 55a on a single rail.. is this a good choice?

also, any other explanations for crashing on boot? - while its still cold, when i increase the vcore (it was only on 1.27v from ocuk and i gradually bumped up to 1.36v)

id really like to get this to the advertised "guaranteed stable" state without having to rma anything.. most likely its my psu i guess...?
 
I'd get a bigger watt power supply If I were you to be guaranteed not getting anymore problems, your graphics card alone uses 180watts.
 
Ibt had a bug in older versions and you need to run 16 threads I believe, though would recommend running Linx 10 runs on all memory or prime 95 on blend for about 8 hours.

What is your current psu putting out on it's rails? Everest, hwmonitor and occt can show this
 
Ok very odd... i tried HW monitor, but didnt believe the result, so tried occt and everest.. can it possibly be true?!

vcore 1.25v
5v rail 5.08v
12v rail 2.11v !?!?

how could my pc even run for a second with 2v on the 12v rail?

bios reports 12.13v on the 12v rail...

hmmm im glad i decided to order the new psu..

i did go for the corsair hx650, it should surely be plenty for my rig? even if the cpu and gpu take 200watts each ive still got 1/3rd spare? the ssds take no power at all and the backup drive is only used occasionally..

ETNiES- you advise going for a larger psu, but your sig shows a machine thats pretty similar to mine, (lower consuming gpu maybe, but a lot more other components) and youre using a 600w ocz there?

anyway hmmm.. 2v on the 12v rail.. odd because the psu has under and overvoltage warning lights on it.. and they are all green at the mo.. a bug in the reporting software?
 
hmm.. ok well i got the corsair hx650 psu, and rigged up my watercooling setup (black ice triple rad with noctua fans, eksupreme cpu block , Laing DDC 1+ Ultra pump/res.)

with the ocuk OC settings (4 ghz (incl turbo) 1.288v vcore) the machine boots fine, and does most of the things i use it for. idle temps are higher than expected at around 38 degrees, but when running occt or IBT the temps never really go above 60 degrees.

however, it still bluescreens out of IBT or occt after either a few seconds or a few minutes.. i can increase the vcore, and have tried up to 1.4v but it doesnt seem to make much difference. changing other voltages (or even dropping the memory multiplier (taking the ram down to 1140 or something) generally leads to a crash on boot or while loading windows.

at 1.4v (1.38 reported by occt) the temps are creeping towards 45 idle and 70 load.


another strange thing is it seems just as happy / unhappy at 4.2 itll boot fine and run 3dsmax, but IBT, occt and also SupCom Forged Alliance after some time) will bluescreen the machine.

finally i also noticed the ram timings seem odd:

chan 1 9,9,9,28
chan 2 9,3,auto,auto
chan 3 9,auto,auto,auto

but if i load the xmp profile they change to 9,9,9,28 throughout, but doesnt seem to help stability..might even be worse.

grr, thought it was the psu...think it could be the ram?
 
robinlawrie said:
hmm its also odd that the bundle i bought was supposed to have corsair ram i believe, but it has patriot, which ocuk dont even sell ?
Click to expand...

you looking at the right website? i got my patriot from ocuk and have just checked and can see 2 diff sets for sale. Them changing the ram is not unusal if there a lack of stock they would just use another brand with same timings i would assume.
 
yep its a gigabyte ud5..

regarding the patriot ram, i know they sell the brand, but not the 1600 mhz stuff (that i could see) .. maybe its just no longer stocked.

think im getting somewhere though.. im at 4.2, (nice and neat! 200 bclk, ram at correct 1600 speed and timings)

ive ditched occt and IBT for the recommend Linx (is that cheating? find a stress tester that doesnt crash?) and got 10 runs out of it before it bluescreened this time.

vcore is at 1.38, but drops as low as 1.34 under load in hwmonitor (thats with load line calibration enabled btw.)

am i supposed to look at the load voltage when i choose my vcore? i could push it to 1.4 under load, but then it would be a 1.45 or something in the bios and at idle, which scares me a bit.)

im hoping my cpu block will "settle in" over the next few days as my idle temps are quite high.. (42 degrees idle and 68-70 degrees under load.)


im also getting a bit of variation between the cores.. one is hitting 74 degrees after 30 mins linx, but another never passes 68.. idle is smaller differences (at this moment: 42, 39, 40, 38 for core temps)

need to get a backplate so i can clamp it down a bit tighter..

i still cant really touch the other voltages without making it crash on boot though.. particularly the qpi/vtt, which id like to raise, but it just wont have it if i increase it. have to leave all on auto..
 
I have the same bundle as you and got the exact same problem, blue screening after few mins (and less) using prime95 and OCCT at 4.0Ghz.

Sent it back for testing and they have told me they stressed it successfully over night, and yet it is still blue screening for me after few mins. I have more than enough power from PSU and temps are fine. Don't figure.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom