i5-2500k Overclocking Help

travisisgoddd

Distinguished
Jan 2, 2012
18
0
18,510
So I have had my i5-2500k for a few years now and never really got into the whole overclocking thing until recently.

Specs:
MSI P67A-G43 (B3) Motherboard
i5-2500k 3.3ghz
Hyper 212+ Heatsink + Fan

What I was at first attempting to overclock my CPU, I was aiming for a 4.5 ghz overclock but no matter what voltage I set, I always got BSOD code 124. I am also aware of the maximum voltage to not pass is 1.52 set in place by intel, but never ventured out beyond 1.4 in personal tests.

I decided to pull back the ratio's down to 43 and set a manual voltage to 1.325 which is the most stable thing I got. I ran a ~6 hour P95 stress test with no problems earlier, I am running another test now over night to make sure. A hour or two ago, I set the voltage down to 1.3v to see if I can get it any lower and got a BSOD 124 after a rough hour of P95 torture test. I guess my question is, is a 1.325v voltage out of the ordinary when it comes to running a 4.3ghz overclock? In my opinion it seems like a very high voltage and I do not know if I should be worried about this. It seems as if the majority of i5-2500k owners can get a 4.5ghz-4.6ghz overclock with the voltage I am setting for a 4.3ghz.
 
use dynamic vcore instead of fixed voltage. 1.325 is within range for 4.3ghz on the 2500k.

http://www.overclock.net/t/910467/the-ultimate-sandy-bridge-oc-guide-p67a-ud7-performance-review

i prefer that guide.

how far can you overclock stable with stock svid vcore... meaning not touching any voltages at all and leaving them on auto?

for initial overclock testing you should leave everything untouched on auto, leave the memory at stock which will be 1333mhz, and TURN OFF all power savings like speedstep, c1e, c6, etc. use a monitoring program like hwmonitor to log vcore and temps(you can use realtemp if you want for temps). use prime 95 version 26.6, dont use newer versions like 28.5.8

i keep my 2500k(lapped) at 1.32v@4.4ghz with my hyper 212 set at normal fan curve and i get about 72c max temp p95 with 68f ambient. though i can go to 4.6ghz with 1.38v, vcore/temps/noise are not worth the paltry 4% performance increase that i will never notice when im transcoding 1080p bluray. i mostly game anyways which it makes no difference.
 
I checked the bios, the MSI P67A-G43 has no options for setting wattages, amp limits and LLC. I set the Vcore to auto and its VCORE is running a 1.304V minimum, and 1.344V Maximum. I can see why setting it before to 1.3V manually can mess since its needs a minimum that is .004 than what I set, but I ran a stable build at 1.325, wouldn't that cause the system to BSOD since it needs a 1.344V maximum? Should I go back to a 1.325V or just leave it at auto?
 
if you are talking about the maximum vcore reported by hwmonitor or some other app... the 1.344v is showing you vdroop that happens when you drop from 100% load to an idle state. that isn't the load voltage. your load voltage is likely .02 lower. if you are running 43x@4.3ghz, what does the bios monitor say your vcore is compared to the holding vcore at 100% load in p95? you should have all the power savings off so that you dont get a 0.935v reported when the cpu drops to a 16x multiplier and a different power state. you can turn those power savings back on later when you get a stable overclock.
transient_no_vdroop.jpg
 
When I enter the BIOS with the VCORE on auto, the bios reports 1.304V being used, while under load in P95, HWMonitor is reporting a VCORE of 1.320V. I will post some pictures of my BIOS to just clarify what options I do and do not have when compared to other MSI MB's. If anything that should be enabled/disabled please let me know.

http://puu.sh/gK9pY/f85aeb6457.jpg
http://puu.sh/gK9on/a84f216252.jpg
 
all that looks good, though you should turn off the intel xmp profile for initial overclocking stability.

where are the pictures of your voltage options for vcore etc.? what other options do you have besides auto for the vcore? do you have normal? on my gigabyte mobo if i change to normal then dynamic vcore becomes unshaded and available for modification.

you should be able to adjust 'dvid' or dynamic vcore. if you do 44x with everything on auto are you crashing in p95 or are you just losing some workers?

if you are then just add 0.010v offset to the dvid setting and retest. keep moving it up 0.010v at a time until you are stable then try removing 0.005v.

vdroop control is basically LLC. but with a modest overclock you shouldn't need it.
 
The MSI P67A-G43(B3) has no options when it comes to dynamic vcore or offsets, what you see is what I have. It is pretty depressing that this specific board was advertised as being a pretty good overclocking motherboard but still doesn't give the users specific options like that. Anyway, I changed the CPU ratio to 45 and the VCORE to auto to see the minimum and maximum voltages of that specific CPU ratio, I ran P95 for about 10 minutes, maybe less and then BSOD code 124. I will test 44x at auto here in a moment and edit my post with the results.

Edit: Put the CPU ratio to 44 and VCORE on auto, HWMonitor is reporting a minimum of 1.320V and a maximum of 1.352V. Going to start a P95 test and I will do another edit to report if it is stable or not.

Edit 2: Ran a ratio of 44 at auto vcore in P95, 2 minutes in my computer locked up and played a very very loud static sound so I restarted and dropped the ratio to 43 and vcore at 1.325V. I do consider this stable but I am still bumming that I can't run a 45 ratio at auto Vcore without almost instantly BSOD during P95.