P9X79 & 3930K BSODs

[multra]

Hi there,



Just got a new machine from Santa (myself)
Specs:

i7-3930K
P9X79 Deluxe
Corsair Dominator GT 32GB CMT32GX3M4X1866C9
Corsair cooler H100i
Coolmaster HAF X case

which produces BSODs under load(random times)
My non OC temperatures:

idle: 39C-41C
max load: ~62C - 65C (with Prime95)

Having said that, I did use the XMP 1.3 profile to make sure the CAS timing for my Corsair RAM were correct which bumps the temperatures.
However, with or without the XMP, I BSOD after running Prime95 or a demanding game (borderlands2, Crysis 2) at random times.
I am at a loss to what i am dealing with here.

Would anyone please point me into the right direction Initially i wanted to overclock this system to 4.5Ghz (with my poor experience with OCing), but having problems at stock demoralizes me quite heavily.

Many many thanks!!
<3

PS please see the screenshots:

Idle:


CPUZ- Idle:


CPUZ - Memory tab:


CPUZ - SPD tab:



Prime95 Blend stress test:


Prime95 Blend stress test CPUZ:





BIOS screens:

 

[bluezero5]

what's our bsod debug code 00000124?

 

[multra]

Thanks for reply.
I am running another prime95 test now. will snap a picture the moment it BSODs.
Max core temp just hit 68C. Not sure if that's normal under load.

 

[multra]

BSODed with 0x0000001E
Max temperature was 70C.

 

[bluezero5]

0x0000001E BSOD..

this is not as common, this is corrupted file/driver which can be caused by many things.
here's a quick read. http://www.almuseo.com/what-causes-0x0000001e-bsod.php
there's a Auto-solution program there, if you wish, you can try it.

On to the reason:
second on that, how that 'started', might be ram or CPU Vcore, OR your video driver/Audio driver, or just an OS file mis-written at somepoint.

I notice you are using Off-set voltage, which is ok if you are not exceeding 4.0Ghz, but I found above that, absolute voltage gives more stablitlity. I will also allow max CPU Capacity to 140% in DIGI power to besure it can temporary deal with power peaks without crashing.

FYI, if you BSOD is 00000124 then it is related to CPU VCore voltages. (mostly)
and your problem 'might' start from something else RAM related.

If it is not too much trouble, you can try to reinstall your OS. start from scratch. as my experience with 0000001E is that it is usually later into more accumulated problems, and without more details I am not sure how to help.

if you re-install your OS, (make sure you got all the drivers you need from MOBO to Drives and everything, otherwise, don't attempt this unless you know what you are doing!) but in case you feel ready to do so. Next thing is try to OC your chip from the BIOS, set Vcore to 1.4V and slowly up the CPU Multiplier until it can you get a BSOD in prime95 blend then. Then down it by 20Mhz, and that should be your 24/7 CPU speed. (I will guess it be around 4.5Ghz at least.) all the while, keep RAM speed at 1600, have all timings on Auto.

then turn RAM profile to XMP, and use the suggested timings. (your case 1866) then do Prime95 again, it should be stable given you set the RAM voltages as recommended too. Then if nothing crashes, you are good to go. If it crashes this time, now you know your RAM might have some issues, you can try to install the RAM 1 and each time, and see if you have a faulty unit. Lets hope you don't need to come back telling us you do!

Good luck bro.

meanwhile, if you have more info on your crash pattern, let us know, we might be able to help more. For example:
1, does it crash if CPU speed slower?
2, does it crash besides at Prime95? how long into it does it crash? etc.

 

[multra]

thanks so much for your response.
I've taken your advice and wiped the os, loaded all the necessary drivers, updated windows. I've reset the bios as well. going to start non oc first. running prime95 now. max core temp reported at 59c with cpu running at 3.5ghz at turbo, max cpu core voltage at 1.224v. is that normal?
I will update the moment it craps out.
(sorry for typos, posting from my phone)
I will report back

 

[redduc900]

Lower the System Agent voltage (VCCSA) back down to the default of .925-.971V before you fry your chip at the 1.30V+ that it's set to now. If anything, w/ 32GB of DDR3-1866 installed you may need to bump VTTCPU and / or the DRAM voltage. Start by raising the VTT to 1.12V and the DRAM voltage to 1.55V, then boot to Memtest86+ installed to a thumb drive and check the RAM for stability.

 

[multra]

thanks redduc900!

update to my previous post,
my pc ran stable with prime95 at stock bios for 2 or so hours. I got excited, abit careless and went ahead and changed to xmp profile(1866). prime95ed for some more without crashes! by that time ive read redduc900's post and lowered VCCSA to .950. (I believe after I switched to xmp, it set VCCSA to 1.3). left VTT and DRAM on auto, to test one thing at a time. So far Prime95 is running without Bsods.

Currently running at 3.8ghz turbo (increased from 3.5ghz after I changed to xmp. I thought xmp only changed ram? sorry if this sentence makes me look idiotic, I'm very green to this). vcore voltage is 1.320 max. 65c -68c

I also ran 3dmark 11 followed by prime95 with no crashes. so far 3 plus hours of stress and bsod free.

I have to say the stock H100i fans are loud at 2k rpm. I tried changing the fan profiles to 'quiet' using corsair link software which drops rpms to 1k but then my core temperatures hit 72c.
providing my room ambient temperatures are 25c-27c, are my core temperatures seem reasonable at max load?

thanks so much guys for you support!

 

[bluezero5]

Awesome!

looks like your rig is stable for now! always a relief!
Hopefully that will never happen again! (if it does, we will help as usual.)


Do stay around, this forum is pretty nice.

 

[multra]

Bluezero5! thanks so much for your insight, its invaluable!
I tried searching around to see what "normal" temperatures are for the i3930k chip around but there is no definitive answer.
Would you consider my temperatures to be high or normal? I am trying to find a reference.

 

[bluezero5]

the TjMAX for 3930k is 91'C

I use Core-Temp, or Real-Temp temp programs.
cause some other programs sometimes record temps that I don't necessarily agree with.
you using HWmonitor.. should be ok, just observe the higher of the 2 CPU temp they have.

to me, High temperature means within 5'C of TjMax, like 86+'C
any under, is totally ok.

I run with 1.52Vcore on water, on my max load, my temp goes up to 71'C, and I am quite happy about that.

 

[multra]

great. i suppose my temperatures are not bad then.
My machines has been BSOD free for almost 20 hours under constant Prim95 stress.
I will start start raising multiplier shortly. Does 1.32v sound reasonable for a 4.2Ghz OC?

Once again, thank you guys so much for stabilizing my machine.
Merry Christmas!!!!!

 

[redduc900]

It shouldn't need that high a Vcore for 4.2GHz, since the average for that clock is ~1.25-1.30V.

 

[multra]

Just changed the H100i stock fans to 2x Noctua HF-F12. The noise reduction is glorious. Of course, that increased my core temperatures a bit though (70C at Prim95)

Another question if you do not mind.

As per reduc900's comment, I am trying to lower my core voltages to a minimum(safe minimum) in order to lower thermals. Again, this is for non-OC speeds. (Speedstep goes up to 3.8Ghz from 3.2Ghz)
I've read that "offset voltage" approach is the way to go if I want to maintain the speedstep, which I do. The only problem with it is it adjusts voltage for both idle and load voltages.
I have tried using negative offset of 0.06v and that yeilded:
0.8v - idle (old is 0.848v)
1.312v - load. (old is 1.320v)

My fear is that the idle voltage will be not enough if I keep increasing the negative offset.
What am i missing here? I'd like to lower my load voltage to say 1.26v without lowering the idle voltage to a point where my machine will halt.

 

[multra]

To answer my own question a tad.
I adjusted my CPU LLC to Normal: (was Auto)
idle min voltage: 0.800v (~1Ghz) 39C
load max voltage: 1.256v (3.8Ghz) 65C

Prime95 is running and the system appears to be stable so far.

Great readings I go off:
http://www.overclockers.com/forums/showthread.php?t=723719
http://www.overclock.net/t/1255029/need-some-help-understanding-llc-and-offset-ocing

 

[eskimochaos]

70C on Prime is not bad, like at all. When you use your computer for real things, it wont even come close to that. My ThinkPad has been running at 85*C for the last five years.

 

[Anonaru]

70C on Prime is not bad, like at all. When you use your computer for real things, it wont even come close to that. My ThinkPad has been running at 85*C for the last five years.

This here-- The H100 will let you up to 70*c on a 3930k quite quickly. I have found it resists temp increases very well once you get past this point. With some extreme RPM (3999RPM) fans added to the mix, my own H100 let me take my 3930k to 4.8GHz with an admirable 78*c top. Of course, I've been blessed with a stable at low vCore chip. Make sure you have ample CFM, but get fans that don't sound like a jumbo jet taking off

Glad to see you got past your BSOD issue, "multra", I've found that installing these high-capacity 4-module kits make acheiving a stable OC quite a bit more difficutl

 

[multra]

Thanks for the clarifications on the temperatures. It is reassuring to know they are within norm. Below are my idle and full load results if anyone is interested with the settings outlined in previous posts. So far the system is stable at idle and at full load.

My H100i stock fans were THE jumbo jets alright at full load (2500rpm) and not comfortable at idle either (1200rpm). That is probably because the case is a foot away from my head. I switched to to 2x Noctua NF-F12 fans and now I am loving it. They are dead silent at idle (600rpm) and still VERY quiet at full load (1400 rpm). AND they keep temperatures down very well at the same time, apparently. I currently only have the push configuration + 2x 200mm fans on top. I am sure I could probably get a few degrees off by adding 2 more Noctua fans to make it into push/pull. This will require some minor modding to the case, as my P9X79 mainboard has very tall heatsink that is in the way of the H100.

next step is to attempt the OC.
Once again, thanks so much for all the input, you guys are awesome!

 

[Anonaru]

Thanks for the clarifications on the temperatures. It is reassuring to know they are within norm. Below are my idle and full load results if anyone is interested with the settings outlined in previous posts. So far the system is stable at idle and at full load.

My H100i stock fans were THE jumbo jets alright at full load (2500rpm) and not comfortable at idle either (1200rpm). That is probably because the case is a foot away from my head. I switched to to 2x Noctua NF-F12 fans and now I am loving it. They are dead silent at idle (600rpm) and still VERY quiet at full load (1400 rpm). AND they keep temperatures down very well at the same time, apparently. I currently only have the push configuration + 2x 200mm fans on top. I am sure I could probably get a few degrees off by adding 2 more Noctua fans to make it into push/pull. This will require some minor modding to the case, as my P9X79 mainboard has very tall heatsink that is in the way of the H100.

next step is to attempt the OC.
Once again, thanks so much for all the input, you guys are awesome!

Yessir! We'll be around to guide you along the way The 3930 is a really fun chip to OC, I've had mine forever now and still play with it daily-- Man does its raw power really bloom at high OC, feels like alien technology compared to my ex-Deneb rig

 

[multra]

Hi guys,
It just happens i will be getting a 3960X processor in place of the 3930K (friend's ties).
Before i fry the processor, what is the safe range for the vcore voltage on that guy? still 1.25v-1.3+ ? before overclocking it.

many thanks!