• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

i5-4690k OC

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

pqwoerituytruei

Member
Joined
Nov 24, 2011
My primary OC exp includes a Phenom II 965 and a GTX 650 Ti Boost

I want to OC it for fun, in a ideal world i would say the goal is 4.8 GHz stable
so far i used the "Optimized CPU setting" in my UEFI Bios's lowest setting 4Ghz

i upped the multiplier to 42 and it seems stable, only have about 10min to Adida64 stability test on it though.

The main 2 things i want to know is the temperature limit and the "safe" voltage limit. still have other questions but those are the most important

About how long should i run adida64 stability test to see if it is stable, i know with prime i ran it for at least 2 hours, usually did 4-8 hours but never had a crash after 2.5 hours, but that was on a 2009 AMD chip

I noticed my 4th core seems to run about 5C cooler than the others under high load, that useful info?

What is target CPU cache speed and should i mess with it?

I don't really trust the so called "Optimized CPU settings" so can anyone tell me if there is anything stupid wrong with them
I assume there is still no way to screenshot the BIOS with it being UEFI, so please forgive the low quality images
 

Attachments

  • pwBqsBP.jpg
    pwBqsBP.jpg
    165.9 KB · Views: 12,129
  • hjM5smC.jpg
    hjM5smC.jpg
    181.7 KB · Views: 9,244
  • unENi9V.jpg
    unENi9V.jpg
    185.2 KB · Views: 9,221
  • VxkgTsi.jpg
    VxkgTsi.jpg
    181.5 KB · Views: 9,150
  • o29UorR.jpg
    o29UorR.jpg
    148.6 KB · Views: 9,120
  • 6U0n5mK.jpg
    6U0n5mK.jpg
    170.5 KB · Views: 9,161
  • uiHawoA.jpg
    uiHawoA.jpg
    380.7 KB · Views: 9,126
Last edited:
Better images:
adida64 already pushes 70C and linpack pushes 80C
Fan settings are supposed to be standard silent, silent,silent; i think the bios has a bug and sets them to what ever it wants to (edit: changing the so called 'optimized OC settings' does that)

i tried upping the 1.15 to 1.20 and running @ 4.3Ghz, it was very unstable, thought i was able to almost load my xubuntu desktop
 
Last edited:
Please attach your images directly to your posts. There is a built in forum tool for doing this. Click on Go Advanced and then click on the little paperclip tool at the top.
 
done, added to 1st post, thought there was a limit of 3 attachments and my original photos were over the size limit
 
Would this OC be safe (voltage/temps wise)

using offset voltages
+0.1 CPU voltage input
+.15 adaptive vcore
 

Attachments

  • oc.png
    oc.png
    362.9 KB · Views: 9,073
  • oc0.png
    oc0.png
    352.4 KB · Views: 9,129
A max. core temp. of 83C on Core #1 at a Vcore of 1.282v is fine assuming that's what you're asking. TjMax is ~100C for Haswell (Tcase = 72.72C for the i5-4690), so at a core temp of 83C Distance to Tj is 17C.
 
so that means i have ~17C of head room, but i had a peak temp of 89C on the 1st core (core0)
100-89=11
the temps relative to the heat that my case expels is so much different then with my phenom II running prime at 4.2Ghz, the core was in the upper 50s at peak and the case exhaust was warm, haswell is running 80C+ and the exhaust feels cool, then again it is a much lower TDP, anyway guess i wont be using a OC to keep warm, lol

it really seems like the IHS is bottlenecking the cooler, or is my logic flawed in comparing temps (core to case exhaust) from my old AMD chip to this one

thanks

BTW some of the screenshots show a vcore of 1.312
how much voltage head room do i have given my NH-D14 can keep it cool
given the goal is just not to waste having a K skew chip and Z97 board
 
A Vcore of 1.28-1.312v seems too high for only a 4.4GHz clock. You should be able to hit 4.6GHz w/ even less voltage, and 4.4GHz shouldn't require more than approx. 1.20v. You probably need to adjust the cache frequency, cache voltage, and enable CPU LLC for a lower and more constant voltage under load. And if you have the option to, disable Turbo Boost.
 
the vrm for amd is on the motherboard, that is why the case exaust got so hot with that.
the vrm is on the cpu with this cpu so the cpu makes all the heat.
 
A Vcore of 1.28-1.312v seems too high for only a 4.4GHz clock. You should be able to hit 4.6GHz w/ even less voltage, and 4.4GHz shouldn't require more than approx. 1.20v. You probably need to adjust the cache frequency, cache voltage, and enable CPU LLC for a lower and more constant voltage under load. And if you have the option to, disable Turbo Boost.
well the cache freq/voltage is stock

should the cache clock match the core clock? it not what % of the clock should the cache be?

LLC=load line calibration, i assume

I have no idea how to calculate the cpu input, i am just guessing i need to use 1.9 after the offset is applied
i assume the cpu input is the voltage that the vcore (and everything else) has stuff has access to
i need to sleep now, eyes wont stay open

@caddi daddi thanks for clearing that up
 
i am not sure how much voltage i need to give the cache, i may be able to lower the voltage a bit more (have not tried less than this), still not sure of what i am doing

cpu input is at +0.08
additional cache/cpu voltage is at +0.075
cache/cpu clock is at 43x
llc is enabled
everything else is auto

my objective is to get the best OC i can without giving of the scaling feature that drops the vcore/click speeds when idle, i dont need to be running at 4+Ghz with a high vcore just to play video that is using the GPU for decoding, would disabling turbo prevent that?
 

Attachments

  • oc.png
    oc.png
    397.3 KB · Views: 8,974
well it seems setting the additional offset voltage was pointless, it does 4.3Ghz stable with everything on auto... except the clock multiplyer
@redduc900 disabling turbo disabled my OC capping my cpu at 3.5Ghz
 
Last edited:
the vrm for amd is on the motherboard, that is why the case exaust got so hot with that.
the vrm is on the cpu with this cpu so the cpu makes all the heat.

I was under the impression that the on-die VRM had been removed from Haswell refresh and Devil's Canyon in addition to the "better TIM" (why not solder it? Buggers...) in an effort to improve temps.
 
I was under the impression that the on-die VRM had been removed from Haswell refresh and Devil's Canyon in addition to the "better TIM" (why not solder it? Buggers...) in an effort to improve temps.
Nope. Just the TIM and some added caps on the improved PCB. They did not change the entire way the CPU works on a refresh. :thup:
 
Nope. Just the TIM and some added caps on the improved PCB. They did not change the entire way the CPU works on a refresh. :thup:
and when Intel said TIM they were referring to the amount of glue they were suing to hold the IHS down, there was too much putting extra space between the die and the IHS, there paste is actually better then NH-T1, but paste need to be ultra thin to work, not a thick blob
replacing the paste would help on the older chips cause you remove the thick layer of glue in the process, when i say thick i mean the thickness of a sheet of paper

this is based on the research i did a few weeks ago

-------------------------------

If someone could tell me how the CPU input, cache voltage, and a vcore are related that would be great
i assume a lower input voltage could result in a lack of power for the vcore
if i knew how to calculate the required cpu input
i have a good idea of safe vcore levels, but not safe cache voltages
 
Looks like they did a bad job with the tim on yours, big difference between the core temps. Mine is a little better but still has a 5c difference between the highest and lowest. I think that you need to enable C1E and EIST to allow it to idle at lower MHZ and voltage. I am having the same issues, can't tell what does what so I haven't been able to add offset voltage. Mine will do 4.4 and pass AIDA64 on default settings with the Vcore showing 1.146. Adding voltage to the CPU actually crashes mine but I am unsure as to why that would be. I got a Gigabyte mobo with a new bios that I can't find any info on yet so It may be a while before I can see what this chip will do.
 
and when Intel said TIM they were referring to the amount of glue they were suing to hold the IHS down, there was too much putting extra space between the die and the IHS, there paste is actually better then NH-T1, but paste need to be ultra thin to work, not a thick blob
replacing the paste would help on the older chips cause you remove the thick layer of glue in the process, when i say thick i mean the thickness of a sheet of paper

this is based on the research i did a few weeks ago

-------------------------------

If someone could tell me how the CPU input, cache voltage, and a vcore are related that would be great
i assume a lower input voltage could result in a lack of power for the vcore
if i knew how to calculate the required cpu input
i have a good idea of safe vcore levels, but not safe cache voltages
well...
The gap wouldn't change with different pastes. They just used something better, but not the best (solder essentially). Shorten the distance between the core and ihs would help..it shrinks that gap.

Check out the guide for info on those options (I'm mobile).
 
My thermal paste reads on the back......

Features:
Diamond particles measuring 0.0000015 cm in a heat-transfer suspensuin solution decrease space between conductive compounds and provide superior thermal performance versus standard metal based compounds. Rated at 5.3 W/mk
Extended -30c to 250c stable temperature range ideal for performance cooling
Lighter, easier to spread stable composition won't crack or dry out.

Thermal paste gaping plays a big role in your cooling. I'd de-lid that bad boy and put the cooling device right on the cores. Always seems to work best.. (water block )
 
Back