Welcome Guest ( Log In | Register )

Outline · [ Standard ] · Linear+

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

views
     
hafiez
post Dec 17 2017, 09:51 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



one noob Q guys....

I'm on 8700k, Z370 gaming k6, crucial ballistix sport ddr4 ram. for cooling, I'm using D15s.

I just play around with the OC tune settings which came with the mobo. I've selected the highest to 5GHz.

with the preset, the mobo set the cpu V to 1.4, and I play around by bringing it down to 1.2v.

at first I received BSOD, return to BIOS and change the value slowly upward, until I get 1.248v with no BSOD.

for ram, I didn't change any values. but it runs on 2400MHz.

for temperature, when stress test it hovers around 80c - 93c.

if I use CPU-z "stress cpu" button, it plays around 80c on load.

on normal usage, idle on 33c-37c and 65c on load.

am I doing anything wrong or I'm exposing the chip to danger?

This post has been edited by hafiez: Dec 17 2017, 09:53 PM
hafiez
post Dec 18 2017, 09:06 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ Dec 17 2017, 11:09 PM)
CPU-Z is not a suitable software to test stability of your OC, neither a good software for benchmarking purpose.

For CPU overclocking, the most standard people would go is Prime95, but take note that it is way too much stressful that your cooling can actually handle the heat.

For me, a couple of hours (3 hours minimum) of Realbench or AIDA64 are more representative of real-world applications.
Between these two, Realbench is better since it stresses both CPU and GPU at the same time.

Anything lower than 100c would be safe. Moreover you're using an air cooler, which is less worrying than custom watercooling with PETG tubing etc. Your normal usage at 65c max looks fine though, just need to  retest with a proper stress tool as I mentioned earlier to confirm stability.
*

do u delid ur proc?

Btw, for the cpu V, why need until 1.3/1.35v ?

hafiez
post Dec 19 2017, 09:52 AM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ Dec 18 2017, 09:50 PM)
No delid, my chip is doing 5GHz @ 1.28v under custom wc with max temps never go above 90c.
Every chip was born unique, meaning each CPU comes with its own stock voltage (VID).
If you're lucky enough you will find a good chip, while some others might get bad chips with high VID.
It's all about draw of luck, silicon lottery.
*

when I open the intel overclocking utility, the max frequency only show 4.99GHz, never went to 5GHz. in youtube video, I see people getting 5.01GHz, what I'm doing wrong?

the youtuber using the same bios and settings as me. he didn't change any values.

and, RAM only run on its specification? the memory I use stated 2400MHz, when I try change it to the higher number like 3000MHz, it wont boot.

so, when back to bios and change it back to the default values.

QUOTE(x-1o8-x @ Dec 18 2017, 10:30 PM)
thanks bro for the input, guess i'll leave it at 0.08v voltage offset for now
*

hi, how you install/use your thermal paste with NHD15s ?

pea style or spread style?

This post has been edited by hafiez: Dec 19 2017, 09:55 AM
hafiez
post Dec 19 2017, 01:05 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(x-1o8-x @ Dec 19 2017, 10:49 AM)
i used pea style. however based on my previous experience it doesnt really matter which style you use as long got enough tongue.gif

the 4.99ghz fluctuating to 5.01ghz is quite common I guess btw, that's just how core clock speed fluctuates. RAMs usually run at their stated specs, so maybe in your case it's 2400MHz @ 1.2v, making it go 3000MHz is basically overclocking.
*

I see, thanks bro. yes, the ram spec is 2.4ghz @ 1.2v.


QUOTE(owikh84 @ Dec 19 2017, 11:02 AM)
How about the CPU frequency showed in CPU-Z?
I would disable BCLK spread spectrum in UEFI to prevent/minimize the BCLK fluctuations
*
ok already turn off the BCLK Spread Spectrum.

the difference is, I can get 5g or 4995mhz most of the time

in cpu-z, it shows 5000.0 MHz 50x100mhz

sometimes 5001,5002.

in AIDA it shows 5.00Ghz

only in intel extreme utility it barely go above 4995mhz. maybe it just a rounding problem, nothing serious.

edit: what is your proc cache value? 43 or 48, is it ok?

This post has been edited by hafiez: Dec 19 2017, 01:08 PM
hafiez
post Apr 19 2018, 10:42 AM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



last week I update to latest bios for my mobo

I didn't realize it wiped my 5GHz (8700k) settings (my bad I didn't save it). I only realize it when I see my pc didn't run at 5GHz on load.

so I just play around again and include all values that I remembered. I'm not a extreme overclocker where by I can remember everything laugh.gif (learning from utube only... blush.gif )

after I key in everything and let my pc run. do some stress test, I found out my cpu speed wont go down to stock when on idle. what did I went wrong yea?

my cpu now stays at 5GHz all the time. but the temp around 40-45c on idle. on load it will go up to ~57c. speed remain.
hafiez
post Apr 19 2018, 11:08 AM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ Apr 19 2018, 10:57 AM)
Your CPU is running at 5GHz constantly even on idle because power saving features have been disabled in the OS and BIOS settings.

In OS, go to Power Options, select Power Saver power plan.

In BIOS, go to CPU Configurations, look for power management settings, enable Enhanced Intel Speedstep Technology (EIST) and C1E.
*
all are enable. this is what makes me wonder though...

and, is it normal for the temp to go up to 100c when doing Small FFTs test? I stop the test.

on Blend test, the temp seems normal for me.

edit:

OK, now I understand what u mean by OS Power Plan.

I change from High Performance to Balanced now.

Previously I changed to HP from Bal.

the speed now are at stock or lower on idle.

This post has been edited by hafiez: Apr 19 2018, 11:21 AM
hafiez
post Apr 19 2018, 12:12 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ Apr 19 2018, 11:49 AM)
It's normal that Small FFTs being hotter than Blend test.
Small FFTs demand less memory than the Blend test, so the CPU spends more time processing data and less time waiting for data.

I think this can be explained here: https://superuser.com/questions/981466/in-p...0-for-al/981477

In most situations you will find Small FFT can easily pass the test but fails almost quickly in Blend.
I'd use custom Blend 1344K min/1344K max FFT with 90%+ RAM capacity stressed.
*

ok, prior to ur reply i did some googling and the results said about the AVX which are enabled in newer version of PRIME. Version 26.x seems no AVX, so that the temp tested using this version will produce lower temp reading. Im on v29 btw.

Regarding the test, on the current settings i manage to get stable on Blend with 1.25v. But when on Small FTTs, immediate BSOD..

I increase the v to 1.27, run small FTTs, no BSOD after several minutes BUT the temp went high 😅...

Whats ur advice regarding the test. Im just normal user with gaming, surfing net n doing some minor work on PC.

hafiez
post Apr 19 2018, 12:48 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ Apr 19 2018, 12:28 PM)
I believe you did not run Blend correctly like I suggested earlier, which is 1344K/1344K FFT (min/max) and with 90%+ RAM tortured.

Small FFT won't stress the IMC like the Blend does.
*
ok, I'm running the custom test now.

suggested time to test, 3 hours plus OR 10 minutes?
hafiez
post Apr 19 2018, 06:00 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ Apr 19 2018, 12:49 PM)
At least 3 hours, up to 8 hours.
10 minutes is seriously too short to find errors.
*

alright I've done my observation for several hours

I re-do all my settings and I start from 0

previously I'm using 5GHz preset by asrock but I edit few things;

max core ratio 50
cpu cache ratio 43
BCLK freq 100
AVX ratio offset Auto

Core voltage 1.25-1.30 (fixed)
Load Line Calibration Level 1

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

then I cleared everything but only edit these ^^ 6 settings, instead of using preset+self edit

--------

when running Prime95 - Custom 1344 Min/Max 90% RAM ;

on 1.25v, BSOD

on 1.27v, System Freeze

on 1.28v, I get worker stopped windows only 7/12 cores working

on 1.285v, only 11/12 workers working after 1 minute

on 1.29v, only 11/12 workers working after several minutes

on 1.30v, all workers working fine on 10 minutes test <-- this I need to test overnight because I do short test to see the workers failing or not upon increasing the V.

++ in CPU-z it shows 1.296v-1.312v during stress test and during idle.

hmm.gif

This post has been edited by hafiez: Apr 19 2018, 06:02 PM
hafiez
post Apr 21 2018, 06:04 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



owikh84, let say only one worker failed or stopped during the prime test, what should I do?

increase RAM voltage?

I read, some said increase the ram voltage

problem is, my ram is 2400MHz and 1.2v

I'm using crucial ballistix sport lt 8gb and single

--edit

right now I'm running at 1.3v 5GHz on normal usage.

This post has been edited by hafiez: Apr 21 2018, 06:05 PM
hafiez
post Apr 22 2018, 08:36 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ Apr 21 2018, 10:47 PM)
As keyo said earlier, most probably is your CPU needing more juice.
Increase your vCore.

BTW, did you check your VID in UEFI BIOS (stock CPU voltage with everything default)?
With VID info it'd be much easier for me to guess your voltage required for 5GHz.
*

i set the vcore to auto

I check at cpu-z it plays around 1.312v ~ 1.328v but only on 4.9ghz. Maybe because avx offset i put on auto and it -1 upon testing.

Or do i need to reset back all those settings?

Edit

Ok when i open HWinfo, it shows 1.439-ish

This post has been edited by hafiez: Apr 22 2018, 08:38 PM
hafiez
post Apr 23 2018, 12:07 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ Apr 22 2018, 11:14 PM)
Clear CMOS, go inside UEFI and read the CPU voltage with everything untouched.
That's your CPU VID.
Almost forgot about the inaccurate voltage reading on Gaming7 that you told me earlier  doh.gif
Maybe I should also get the latest batch of this particular board to double confirm it.
*

this? All auto already...

» Click to show Spoiler - click again to hide... «


This post has been edited by hafiez: Apr 23 2018, 12:16 PM
hafiez
post Apr 23 2018, 06:18 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ Apr 23 2018, 03:51 PM)
Based on your screenshot, 1.088v is your VID, which is exactly the same as my 8700K.
It is a good chip, better than average chips.

With NH-D15S, it is able to do 5.0GHz @ 1.328v avg core temps: 39C Idle, 86C load Prime95.
On powerful custom loop with dual 360 rads, it just needs 1.280v, avg core temps: 37C Idle, 72C load.
The 0.048v difference between D15S and custom loops is because of "leakage" at 80C.

So if you want to run 5GHz at low vCore you better keep your temp below 80C.
This can be achieved by upgrading your cooling or delidding your chip.
*
ok I've set the settings to OC and run prime95 and this time vcore fixed to 1.328v

avx offset set to auto (previously either 1 or 7)
llc set to line 5 (previously line 1)

with this settings I see some improvements;

1) stable temperature, no jumping or spike
2) stable speed @ 5GHz
3) temp max 82c, average 60c

BUT one core / worker stop running at 45mins mark

last time it will failed at 15th mins

increase some vcore..? like you and keyo said before

does it fails because of temp also?
hafiez
post May 1 2018, 11:28 AM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



am I doing the right thing?

in bios I set 1.33v

and this is the result after few weeks of testing, no workers stopped. biggrin.gif

» Click to show Spoiler - click again to hide... «


This post has been edited by hafiez: May 1 2018, 11:36 AM
hafiez
post May 2 2018, 08:06 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ May 1 2018, 01:31 PM)

Nice try!
Seems like one worker is not running as shown in the red boxes above.
You probably need to increase vCore a lil bit more to get that fixed.
Run RealTemp GT (browse inside RealTemp folder) to monitor temps of all the 6 cores.
The PCIe @ x8 lane instead of x16, is your GC installed on the PCIe x8 slot? or are you running SLI/CF?
*
up the vcore to 1.335

here is the 1 hour test result

the utilization is 100% showed in HWiNFO and Task Manager, only sometimes drop to 80% but not long till 95%++ again

I dun understand why in RealTemp only show 25~ % load

in the Prime95, the self test passed when I open the minimized window. just don't know why it didn't show like others.

btw I already relocate my gpu to slot x16. I didn't realized it sweat.gif

Attached Image

This post has been edited by hafiez: May 2 2018, 08:07 PM
hafiez
post May 2 2018, 09:59 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ May 2 2018, 08:55 PM)
When you see your CPU temps hit 100C, it means your OC is unsuccessful.
Thermal throttling kicks as soon as you reach the TJmax (100C), means your CPU speed will auto-reduce to a safe temp.
Looks like you need a more powerful cooler, or delid biggrin.gif
BTW, launch Realtemp GT.exe instead of non-GT to monitor all the 6 cores.
*

aiya...okay didnt know that

But i only changed the vcore

I will try to change the LLC later to 1, currently Auto

RealTemp GT didnt have in the folder, i7TurboGT got.
hafiez
post May 3 2018, 09:32 AM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ May 2 2018, 10:18 PM)
GT should be there, download Realtemp again from here:
https://www.techpowerup.com/download/techpowerup-real-temp/
*

ok this is the result after I changed the LLC from Auto to 1, vCore remain. 4 hours. biggrin.gif

the temp below 100c, the highest was 92c and it only happened few minutes after Prime95 starts.

after 30minutes, the temp only plays around 55-75. but mostly I can see around 60-65.

downloaded new GT.

all workers on prime passed just that one worker didn't show. only can see it when maximize the worker's windows.

the load in RealTemp looks ~100% after change the LLC to 1

Attached Image

This post has been edited by hafiez: May 3 2018, 09:33 AM
hafiez
post May 3 2018, 07:59 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ May 3 2018, 07:43 PM)
Good job.
Seems like you managed to solve the over-temp issue by adjusting the LLC to level 1 (least fluctuation).
My core temps fluctuated like 10c, but not as much as yours.
Maybe because power saving feature has been enabled, I could see your lowest CPU frequency going down to 800MHz and back up to 5.0GHz max. Or maybe you did not stress your RAM capacity that much, only Single-Channel @ 8GB or less, didn't see your Task Manager.
*
I guess that was the first reading when I open HWiNFO. I only hit Test in Prime95 after all apps are opened.

yes im on single channel only because I only have 1 sticks of ram blush.gif

along the test I do open task manager and see the utilization. it was 100% all the time.

and the Memory showing almost 100% of usage.

This post has been edited by hafiez: May 3 2018, 08:00 PM
hafiez
post May 3 2018, 08:37 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ May 3 2018, 08:29 PM)
That's Blend test.
Use Custom Blend better, insert 1344K FFTs in both min and max boxes, stress at least 90% of your RAM capacity.
Open up Task Manager as a proof.
*

yes i did this all this while since u told me this few weeks ago

I only do that custom blend.
hafiez
post May 3 2018, 08:57 PM

Look at all my stars!!
*******
Senior Member
2,980 posts

Joined: Jan 2007
From: Mount Chiliad



QUOTE(owikh84 @ May 3 2018, 08:41 PM)
Alright I could see the RAM usage in your HWiNFO screen, 95% average which is good enough.
Hmm.. as I said earlier, temp fluctuation might be due to your power saving options.
*

any advice on that?

Should i try to play with LLC maybe Level 2/ 3. Does this options got any relations?

Last time i tried level 5, bsod after few mins of prime laugh.gif

2 Pages  1 2 >Top
 

Change to:
| Lo-Fi Version
0.0810sec    0.37    7 queries    GZIP Disabled
Time is now: 30th November 2025 - 09:49 AM