Welcome Guest ( Log In | Register )

Outline · [ Standard ] · Linear+

 Skylake/Kaby/Coffee/Comet/Rocket Lake OC V1, Show off your OC here

views
     
reconvision
post Dec 27 2017, 12:50 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
Hi guys. I'm new to overclocking. I have overclock my i7 8700k with noctua nh d15 dual fan setup. I'm able to get stable at 4.8ghz at 1.26v in prime95 for 2 hrs. But the temperature is just spiking crazily from 70 to 85. Even in cpu intense game like battlefield 1 & assassin creed origin, temp are also fluctuating from 68 to 84. Is this even normal? Do i really need Aio to run at that kind of voltage?
reconvision
post Dec 27 2017, 04:01 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(owikh84 @ Dec 27 2017, 03:49 PM)
Not sure why ASUS MY doesn't bring in the Code this time. It's a good board for those who want to have the appearance of the Formula but don't plan to go for custom loop and don't want to pay extra for the EK hybrid VRM heatsink/wb.
Hi bro, welcome to the OC club.
Your temps look normal at such CPU frequency, vCore and cooling.
D15 is the best air cooler that money can buy and it outperforms most AIO coolers. During my testing, with push pull fan configuration it performs on par with Cooler Master"s Nepton 240m  AIO. Some higher-end AIO coolers especially those 360mm will win by 1-5 degrees Celsius.
Wait for Corsair H150i 360mm AIO cooler to be released on Dec 31. I read Vortez's leaked review (before it went down) they claimed that it beats the NZXT X62 by 14 degrees C. But be prepared to pay around RM 1k for the price lol.
*
Thx bro! Ok at least i know the temp i get is within the normal temp at such vcore. Coz some ppl claim to be running 5ghz on air 24/7 make me have high hope of reaching there but it seems with our hot weather , i do not wan to run my cpu at low 90 Celsius everyday xD. Now i should decide either to finally venture into aio or delid.

reconvision
post Dec 29 2017, 10:22 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(x-1o8-x @ Dec 28 2017, 05:58 PM)
that is pretty low TBH, im getting 90c with a D15S and also dual fan setup, are you using the latest prime95?
*
I use prime95 v26.6. What I heard is the new version of prime95 have avx instruction. So I avoid using the new version since most games dont use avx instruction.
what overclock setting are you using now??


This post has been edited by reconvision: Dec 29 2017, 10:23 PM
reconvision
post Dec 30 2017, 09:33 AM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(x800 @ Dec 30 2017, 08:12 AM)
avx instructions can generate quite a bit more heat. I believe that's why your temp is a lower than the other guy.

try one run with prime95 ver that supports avx and see, that'll show a sort of worst case scenario for the temps
*
Yeap. Thats when u need avx offset for the rescue. Btw anyone experiencing avx bug whereby even idle at desktop trigger the avx offset ?

This post has been edited by reconvision: Dec 30 2017, 09:34 AM
reconvision
post Dec 30 2017, 01:58 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(x-1o8-x @ Dec 30 2017, 10:39 AM)
I only have offset of -0.08v to the core voltage @ 4.7ghz and everything else is auto. but yea I'm using the latest prime95 which has avx instruction so my temps are like 90c on core3/4, 1.200v under load.
*
Ya. Today i found quite a lot of ppl with d15 are reporting almost the same temp underload at 75-80 in game with no avx instructions. So i can said that our temp is normal and the other way around is to delid it which will bring our 8700k to 60c underload. Tempting to do it but i still dont want to void the warranty yet sad.gif. The tim from intel is just garbage.
reconvision
post Dec 31 2017, 12:34 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(owikh84 @ Dec 31 2017, 12:21 PM)
AVX offset has been a debatable issue since Kaby Lake. Some said it's a bug while others saying otherwise.
It could be that you have certain OS software running in the background that's using AVX instructions, triggering the AVX offset thingy. Here's the list of AVX-based software: https://forums.anandtech.com/threads/what-c...d-avx2.2498660/
Some games might be using AVX instructions, or NVIDIA driver itself is implementing it: http://www.overclock.net/t/1641867/avx-offset-bug
This doesn't happen with Skylake-X though or he needs more games to test: http://www.overclock.net/t/1643198/coffee-...ks-while-gaming
One guy said this bug is triggered when C-states are enabled in the UEFI: http://forum.gigabyte.us/thread/2704/psa-avx-offset-bug-real
I personally don't use AVX offset because I want my OC to be all out for worst case scenarios.  biggrin.gif
Check out the OC guide in the top post of this thread for GIGABYTE's Z370 GAMING 7. It will guide you OCing your 8700K to 5GHz.

But I don't think you can comfortably run 5GHz OC with your current air cooling, unless you hit silicon lottery with a really low VID (stock voltage with everything defaults) and delid. So it really depends on good is your CPU which varies from one chip to another. Get a bigger and better cooler like the NH-D15/S or 240mm+ AIO.

It is advisable to check your VID through the UEFI first before proceeding for OC. Check your CPU core temps with your CPU running at stock clocks with Realbench v2.56, then only proceed for 4.7GHz on all cores and so on.

For record, my 8700K has 1.088v VID and it's capable to do 5.0GHz running under a high-end custom loop at 1.280v on Prime95 v29.3 custom blend 1344K.
*
wow! thanks for showing me the link as well. Appreciate your work notworthy.gif Yes I solve it by setting avx offset to zero now. so i can have 4.8ghz all day .
reconvision
post Dec 31 2017, 02:17 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(owikh84 @ Dec 31 2017, 01:13 PM)
[attachmentid=9470946]

Alright guys, currently I am testing my 5.0GHz OC on the Hero X with the an air cooler in the form of Noctua NH-D15S (2x NF-A15 fans).
To my surprise, for some reason I need to pump more voltage as compared to my custom loop.

[attachmentid=9470959]
Noctua NH-D15S: 1.328v, avg core temps: 39C Idle, 86C load
Custom loop: 1.280v, avg core temps: 37C Idle, 72C load

Yes that's right, 0.48v difference!  blink.gif

At first I couldn't believe it so I re-tested using back my custom loop again I can confirm that there's no chip degradation in which I am still able to obtain stability at 1.280v under custom loop same like before, I don't know why LOL. rclxub.gif
*
interesting. So i might be able to get some voltage down when i pair my cpu with that new corsair 360mm AIO cooler xD.
reconvision
post Dec 31 2017, 03:01 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(owikh84 @ Dec 31 2017, 02:26 PM)
I understand that the lower temps, the lower the voltage required but seriously did not expect as much as 48mV.. .Funny LOL.
Guess something is not right when the temps go above 80C, don't know...

btw, Corsair H150i Pro RGB 360mm AIO cooler appeared in a recent LTT video. Seems like the launch is not today, but most likely to be released during CES 2018 that will be held in Las Vegas between 7-12/1/2018.

[attachmentid=9471095]

*
thumbup.gif looking forward to it. hopefully it will arrive at our shore soon.
reconvision
post Jan 1 2018, 04:20 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
Just went to chevk my bios setting again. Found out my vccio (auto) at 1.336v and voltage agent (auto) at 1.152v . Are these value too high??? My dram voltage it set to 1.35 default.
reconvision
post Jan 1 2018, 07:11 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(owikh84 @ Jan 1 2018, 06:29 PM)
Those values are almost similar to ASUS's boards running at auto setting.
At DDR4-3200, I have mine defaults at 1.3v and 1.125v respectively but I decided to reduce the VCCIO to down 1.125v and I can still run it stably. As a result, my CPU temp can be reduced a bit.  biggrin.gif
*
Which software are u using to test the ram stability? I have just tune both vccio and vccsa to 1.2v. I am now running prime95 blend test for 2 hrs. Kinda weird, when i click start blend, prime95 stop working msg pop out. but when i try to test again, it is running an hr without any crash. Not sure if the ram is stable or no.

Edit: After 2 hrs of prime95 blend test, no error or warning or crash. So i guess is stable.

This post has been edited by reconvision: Jan 1 2018, 08:30 PM
reconvision
post Jan 2 2018, 09:12 AM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
Not sure what really happen. I choose to run the prime95 custom and set everything according to your pic u share with 14000mb with 15min tome to run fft. Once i click start, a few worker stop immediately while others in self test mode. My cpu is not stress too coz the temp is at idle temp. Leave it for 2 hrs still not done. I try to stop the test but it wont let me do it. So i try to end task in task manager, an error pop out . I try again amd finally it close. Then i check my ram usage is now staying at 5.4gb. Really need help... Im very confuse now. One thing i notice is those worker who stop immediately are showing swap error something. ( Btw i shutdown my pc as i hv to go work)

This post has been edited by reconvision: Jan 2 2018, 09:15 AM
reconvision
post Jan 2 2018, 09:34 AM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(owikh84 @ Jan 2 2018, 09:27 AM)
Errors pop up means your OC is not stable. Most likely is due to imsufficient voltage etc.
Try launching Prime95 under Administrator mode to fix the RAM usage issue.
*
Im still noob in prime 95. What can i do in prime95 administrator mode to fix the ram usage?

Edit : This is the error (Cannot initialize FFT code, errcode=1005 Unable to allocate memory. One possible cause is the operating system's swap area is too small)

This post has been edited by reconvision: Jan 2 2018, 09:41 AM
reconvision
post Jan 2 2018, 11:23 AM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(owikh84 @ Jan 2 2018, 11:06 AM)
Right click on the Prime95 icon on your desktop , run as Administrator.

The error pops up because you're attempting to stress a lot of RAM more than the available capacity.

Don't simply enter 14000 MB in the Prime95. Check your avaiable memory through the Windows Task Manager then only assign the capacity to stress.
*
I mean by fixing the ram usage , do u mean i just run as administrator and stop the test that is runing in background?

I have 16gb (2 x8gb) so i thought it should be fine for 14000mb since i have nothing open at that time only core temp and hwinfo.
reconvision
post Jan 2 2018, 12:36 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(owikh84 @ Jan 2 2018, 12:19 PM)
Not sure if it's just me but there's a bug in the Prime95 that won't stress full RAM unless I run it in admin mode.
*
Oh lol. I think you misunderstood. Usually my ram usage in idle will be 1.3gb but yesterday after i use task manager to end task the prime95 (since it wont allow me to stop it) n close the prime95. I found out my ram usage now idle at 5.4gb. so im worry did i do something wrong to the prime95 test that it mess up my ram usage?

Edit: in the task manager , the memory show memory composition with 2.5gb compress. Is this the result of setting too much ram in prime95 causes window to activate memory compression?

This post has been edited by reconvision: Jan 2 2018, 12:51 PM
reconvision
post Jan 2 2018, 08:10 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
Reboot seem to get back to default of 1.9 ram in used. Now I realise is my mistake for putting 14000mb as the size coz now i check the available ram is 13.8GB. Those stupid skype apps and window defender is eating up the ram to 1.9gb ram. Also forgot about the window cached that taking out 2.1gb sweat.gif no wonder my prime95 crash instantly...

In the task manager, I saw available ram is 13.6gb whereas inside the open resource monitor i saw free ram is 12gb. which one should I follow for prime95?

This post has been edited by reconvision: Jan 2 2018, 08:29 PM
reconvision
post Jan 2 2018, 10:31 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
Ok. Starting my prime test again now with 10000mb. Why after i press start already have a few workers stop working and show 0-minutes 0 error 0 warning? Is that normal? I can see my cpu is doing work becoz the temp raise to 60-70c.
reconvision
post Jan 2 2018, 10:49 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(owikh84 @ Jan 2 2018, 10:35 PM)
Post the SS here with P95 screen, task manager etc.
Stressing 10000MB looks kinda small for a 16GB RAM.
Should take 95%+.
*
Found out the problem. All along im using prime95 32bit =.=. Now running with 64bit prime. All workers is running now. I check the free ram before i start the test and it only have 11000mb so i set 10000mb.
reconvision
post Jan 2 2018, 11:00 PM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(owikh84 @ Jan 2 2018, 10:52 PM)
bangwall.gif
*
Sorry for the trouble >.<. thanks for the help. Now i will pray to the overclock god for smooth test.
reconvision
post Jan 9 2018, 09:01 AM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
Bro Owikh84, the corsair h150i pro review is out.
(http://www.guru3d.com/articles-pages/corsair-h150i-pro-review,1.html). It seems like it is only slightly better than d15 . I guess delid would contribute more temp reduction.
reconvision
post Jan 9 2018, 11:50 AM

New Member
*
Junior Member
40 posts

Joined: Dec 2013
QUOTE(owikh84 @ Jan 9 2018, 10:26 AM)
Seems like the thermal results very from one review to another so I think we will need to read more reviews to compare. Different review sites used different CPUs and software to stress the coolers.

Mainstream Intel CPUs from Ivy Bridge till the newest Coffee Lake themselves have inefficient stock thermal paste and excessive gap underneath the IHS so these factors also need to be taken into the account. The thermal results won't be as accurate as the ones tested on the enthusiast-class CPUs (ie. 3930K etc) with soldered thermal interface. Furthermore, bigger chips like the 3930K have larger contact surface than the smaller CPUs.

G3D: performs similarly compared to Kraken X62 (based on 4790K)

Vortez: beats Kraken X62 by 13C (based on 3930K)

OC3D: outperforms Kraken X62 by 14.5C (7700K @ 5.0GHz)
*
Totally agree with you. I think the most resemble one to i7 8700k will be the i7 7700k oc to 5ghz from OCD. The result is astounding.

2 Pages  1 2 >Top
 

Change to:
| Lo-Fi Version
0.0565sec    0.33    7 queries    GZIP Disabled
Time is now: 29th November 2025 - 08:00 PM