Overclock.net › Forums › Intel › Intel CPUs › [closed] i7-5820K: Can't get my OC stable (Need Advice)
New Posts  All Forums:Forum Nav:

[closed] i7-5820K: Can't get my OC stable (Need Advice)

post #1 of 15
Thread Starter 
Hey guys,

I'm having massive issues getting my OC stable. I'v read several OC guides and tried several different approaches to get the OC I'm aiming at (4.6GHz), but to no / nearly no results. I'm not completely new to the topic of OC but I'm far from an expert. This OC isn't my first one, but by far the most troublesome I ever did.

AIM:
  • Stable 24/7 workstation
  • Max clock possible without exceeding 1.3v Vcore
  • RAM running at XMP (3000MHz)

OC Basic Setup: (This is what I started with)
  • XMP: Enabled
  • CPU Strap: 125MHz
  • BCLK Frequency: 125.0
  • Vcore: 1.2v
  • VCINN: 1.9v
  • LLC: Auto
  • Intel Virualization Technology: Enabled
  • Fast Boot: Disabled
  • Boot Logo Display: Disabled
  • POST Delay Time: 5s
  • Setup Mode: Advanced Mode


The best OC approach I had so far was by following Silent Scone's OC guide at the overclockers uk forums. This is what I found out following his guide:

OC-P1 (4,3GHz, 1.20v Vcore): POST => OK; WIN boot => OK; CPU-Z: verified; RealBench 7h+ (crashed just before finishtin the 8h test (Error: LuxMark-x64 has stopped working); Temps: 69°C max.
OC-P2 (4,5GHZ, 1.20v Vcore): POST => OK; WIN boot => OK; CPU-Z: verified
OC-P3 (4,6GHZ, 1.23v Vcore): POST => OK; WIN boot => OK; CPU-Z: verified (sweet spot?)
OC-P4 (4,7GHZ, 1.30v Vcore): POST => OK; WIN boot => OK; CPU-Z: verified
OC-P5 (4,8GHZ, >1.30v Vcore): Not tested. Too much Vcore needed.

I figured, that the sweet spot must be OC-P3 (4,6GHZ, 1.23v Vcore) so i concentrated on that profile. By trying to validate my OC with AIDA64 and RealBench - as recommended in most guides - I hat some issues regarding stability using Windows 10. So I stepped over to good old Prime95 (v27.9). I've some experience in using Prime95, so I thought it might be a good idea. My plan was to find out the best settings for my OC by target-oriented stress testing at specific FFT sizes. Unfortunately, I can't get the system stable enough to get prime 95 to run for more than a few seconds. I've tried to increase the Vcore step by step up to 1.30v, but to no result. The system didn't get any more stable.

I hope you guys can help me.


My Tools:
AIDA64 Extreme => isn't working properly under Win10 (Error: System Out of Resources)
RealBench => seems to have issues under Win 10 aswell, not sure about that (Error: LuxMark-x64 has stopped working)
Prime95 v279/v285 =>
HWMonitor
CPU-Z

My System:
OS: Windows 10 Pro
Mainboard: ASUS X99-A | BIOS v1801 (latest)
CPU: i7 5820K
RAM: G.Skill Ripjaws DDR4-3000 16GB
Graphics: Palit GTX 980 Super Jetstream
Cooler: Noctua DH-15
HDD: Samsung 850 Pro

Guides I'v read / watched so far:
http://www.hardwareluxx.com/index.php/reviews/hardware/cpu/33127-oc-guide-bringing-haswell-e-to-its-limits.html (EN)
http://rog.asus.com/365052014/overclocking/rog-overclocking-guide-core-for-5960x-5930k-5820k/ (EN)
http://forums.overclockers.co.uk/showthread.php?t=18645325 (EN)
https://www.youtube.com/watch?v=Nj4ixZE7EWo (EN)
https://www.youtube.com/watch?v=Ikq6CJBv_uA (EN)
https://www.youtube.com/watch?v=Mj1Tcd2BW0o (EN)


Regards,
Chopchop2
Edited by ChopChop2 - 8/22/15 at 11:46pm
post #2 of 15
Thread Starter 
UPDATE:

I'm slowly getting of the ground. The 4.6GHz profile seems to be too hot and too Vcore greedy. I switched to the 4.5GHz profile. I'm confident to get this profile stable under / around 1.30v Vcore and hopefully with acceptable temps.


Regards,
Chopchop2
post #3 of 15
Where are you getting your CPU voltage reading from CPU-Z? HWMonitor? etc?

Anyway, take a look at this guide: (It's a pdf file)

https://drive.google.com/file/d/0Bz2VRRbLPrZnYjlVR2kzYkVvMFE/edit?usp=sharing
Gaming Computer
(19 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core 2 Quad Q6600 Asustek P5Q-PRO Nvidia 9800 GTX+ Kingston DDR2 8gb 
Hard DriveOptical DriveCoolingOS
Western Digital 1 TB Pioneer Noctua NH-U9B Windows 7 Ultimate 64-bit 
MonitorKeyboardPowerCase
Dell Monitor Logitech G110 Antec 750W Thermaltake Aluminium Old Case 
MouseMouse PadAudio
Logitech G500 Gaming Mouse Generic Mousepad Soundblaster 5.1 VX 
  hide details  
Reply
Gaming Computer
(19 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core 2 Quad Q6600 Asustek P5Q-PRO Nvidia 9800 GTX+ Kingston DDR2 8gb 
Hard DriveOptical DriveCoolingOS
Western Digital 1 TB Pioneer Noctua NH-U9B Windows 7 Ultimate 64-bit 
MonitorKeyboardPowerCase
Dell Monitor Logitech G110 Antec 750W Thermaltake Aluminium Old Case 
MouseMouse PadAudio
Logitech G500 Gaming Mouse Generic Mousepad Soundblaster 5.1 VX 
  hide details  
Reply
post #4 of 15
Thread Starter 
Quote:
Originally Posted by benjamen50 View Post

Where are you getting your CPU voltage reading from CPU-Z? HWMonitor? etc?

Anyway, take a look at this guide: (It's a pdf file)

https://drive.google.com/file/d/0Bz2VRRbLPrZnYjlVR2kzYkVvMFE/edit?usp=sharing

Hello benjamen50,

thanks for replying to my post. The CPU voltages I mentioned were the voltages I set in the BIOS. I've checked them (for each profile) with CUP-Z under Windows. The voltages shown in CPU-Z were matching my BIOS configuration (e.g. 1.29v in BIOS => 1.289v in CPU-Z).

Thanks for the guide. That's actually exactly what I did and how I did it (usual approach in most guides). The two main differences were that I directly started with 4.3GHz and manually set the VCINN to 1.9v according to the guide of Silent Scone.

I did my testing with Prime95 v27.9 and checked temps / voltages in parallel with HWMonitor. I started my testing with a 5min run of FFT size 1344k to test the Vcore stability, followed by a 15min run of FFT size 864 to test the general stability of the OC. If the system was BSODing I've checked the BSOD-messages with BlueScreenView. According to the BSOD-errors, I've then adjusted the OC-values for the next run (e.g. error 0x124 => increase Vcore, etc.).

If the a profile passes the 15min of FFT 864K, I usually stop playing around with it and start a long term stress test (8h) to see if it stays "stable". For those stress tests I've used RealBench up to now. The reason is, that it produces a lot less heat than Prime95.

Is my approach to test thinks OK or should I do it in a different way? How long should I run Prime 95 / other stress tests to validate stability?


Results
4.5GHz, 36x125, Vcore 1.28v, Ambient temp.: 26,0°C
FFT 1334k 5min: Passed; Max. Core Temp: 88°C (peak) / Avg.: 82°C
FFT 864k 15min: Passed; Max. Core Temp: 90°C (peak) / Avg.: 86°C
RealBench 8h: Passed; Max Core Temp: 81°C (peak) / Avg.: 75°C

Today I reconfigured my HDDs setup. I've switched my OS SSD to the second Intel SATA Controller to be able to use the first one for a RAID config. Afterwards I thought it might be a good idea to re-validate my OC, just in case using the second controller might have changed anything. While re-validating with FFT 864k I found out, that my current OC wasn't stable above 15min and under 30min, anymore.

Can such an instability be caused by the new SATA Controller or is it my OC that just isn't stable?! ... but why did the OC pass an 8h RealBench stress test if it's not stable?

It's getting really frustrating ...


Regards,
Chopchop2
Edited by ChopChop2 - 8/22/15 at 8:10am
post #5 of 15
I could be wrong, but since RealBench produces a lot less heat than Prime95, that would indicate to me that it is putting a fair bit less load than Prime95. That could by why the OC passed an 8h realbench stress test but not prime 95. Try Intel XTU (Intel Extreme Tuning Utility) benchmark runs / stability tests and see if it passes stable on those.
Gaming Computer
(19 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core 2 Quad Q6600 Asustek P5Q-PRO Nvidia 9800 GTX+ Kingston DDR2 8gb 
Hard DriveOptical DriveCoolingOS
Western Digital 1 TB Pioneer Noctua NH-U9B Windows 7 Ultimate 64-bit 
MonitorKeyboardPowerCase
Dell Monitor Logitech G110 Antec 750W Thermaltake Aluminium Old Case 
MouseMouse PadAudio
Logitech G500 Gaming Mouse Generic Mousepad Soundblaster 5.1 VX 
  hide details  
Reply
Gaming Computer
(19 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core 2 Quad Q6600 Asustek P5Q-PRO Nvidia 9800 GTX+ Kingston DDR2 8gb 
Hard DriveOptical DriveCoolingOS
Western Digital 1 TB Pioneer Noctua NH-U9B Windows 7 Ultimate 64-bit 
MonitorKeyboardPowerCase
Dell Monitor Logitech G110 Antec 750W Thermaltake Aluminium Old Case 
MouseMouse PadAudio
Logitech G500 Gaming Mouse Generic Mousepad Soundblaster 5.1 VX 
  hide details  
Reply
post #6 of 15
Thread Starter 
Quote:
Originally Posted by benjamen50 View Post

I could be wrong, but since RealBench produces a lot less heat than Prime95, that would indicate to me that it is putting a fair bit less load than Prime95. That could by why the OC passed an 8h realbench stress test but not prime 95. Try Intel XTU (Intel Extreme Tuning Utility) benchmark runs / stability tests and see if it passes stable on those.

Hi benjamen50,

I'll try that and report back. How long should I run a stress test to verify that my/a OC is "stable"? Is it still advisable to do stress testing with Prime95? I'v heard rumors that using Prime intensively would damage modern CPUs.

Regards,
Sebastian Himmel
post #7 of 15
Quote:
Originally Posted by ChopChop2 View Post

Hi benjamen50,

I'll try that and report back. How long should I run a stress test to verify that my/a OC is "stable"? Is it still advisable to do stress testing with Prime95? I'v heard rumors that using Prime intensively would damage modern CPUs.

Regards,
Sebastian Himmel

I've only heard this rumor affects CPU's that are running under an adaptive or offset voltage. A 0.1v+ increase due to using certain versions of Prime95, usually the newer versions. I stick to Prime95 v26.6 or v27.9.

For Intel XTU, I do usually 10 benches to do a quick and easy stable check for CPU overclocks. For the stress testing on Intel XTU, I recommend around 1-3+ hours. Honestly myself I don't even test long for my overclocks. I've only done up to 30 minutes to an hour.
Gaming Computer
(19 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core 2 Quad Q6600 Asustek P5Q-PRO Nvidia 9800 GTX+ Kingston DDR2 8gb 
Hard DriveOptical DriveCoolingOS
Western Digital 1 TB Pioneer Noctua NH-U9B Windows 7 Ultimate 64-bit 
MonitorKeyboardPowerCase
Dell Monitor Logitech G110 Antec 750W Thermaltake Aluminium Old Case 
MouseMouse PadAudio
Logitech G500 Gaming Mouse Generic Mousepad Soundblaster 5.1 VX 
  hide details  
Reply
Gaming Computer
(19 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core 2 Quad Q6600 Asustek P5Q-PRO Nvidia 9800 GTX+ Kingston DDR2 8gb 
Hard DriveOptical DriveCoolingOS
Western Digital 1 TB Pioneer Noctua NH-U9B Windows 7 Ultimate 64-bit 
MonitorKeyboardPowerCase
Dell Monitor Logitech G110 Antec 750W Thermaltake Aluminium Old Case 
MouseMouse PadAudio
Logitech G500 Gaming Mouse Generic Mousepad Soundblaster 5.1 VX 
  hide details  
Reply
post #8 of 15
Thread Starter 
Quote:
Originally Posted by benjamen50 View Post

I've only heard this rumor affects CPU's that are running under an adaptive or offset voltage. A 0.1v+ increase due to using certain versions of Prime95, usually the newer versions. I stick to Prime95 v26.6 or v27.9.

For Intel XTU, I do usually 10 benches to do a quick and easy stable check for CPU overclocks. For the stress testing on Intel XTU, I recommend around 1-3+ hours. Honestly myself I don't even test long for my overclocks. I've only done up to 30 minutes to an hour.

Hi benjamen50,

I've done 10 benchmarks with XTU as suggested without a problem (Speed: 4.5GHz, Max. Temp.: 80°C). For some quick results I'v done a 30min stress test. The stress test seems to be even more soft to the system than RealBench, but look for your self.




Regards,
Chopchop2
post #9 of 15
Most Haswell-E hexacores hits a ceiling at 4.5ghz. It isn't stable simply because your chip isn't capable.

My 5930k 4.5ghz takes 1.365v for it to be prime stable. Prime is really quick at finding any instability so I recommend tweaking your settings until it can pass at least an hour of it then move on to x264 encoding benchmark found on the devil canyon thread.

First it'll crash, as you make more refinements only individual threads will crash. I find working in reverse works best. Use the max "safe" voltages and decrease it until it crashes. vcore and input voltage are the two key voltages to getting a stable overclock.
post #10 of 15
My difference between 30 minutes of stability and 24 hours of stability for me was in the realm of .05V. This might be what you're running into. I usually do 1 hour before trying to move on to the next multiplier. It took me 5 bumps once I settled at 4.5 on my i75820k.
Main Rig
(8 items)
 
   
CPUMotherboardGraphicsRAM
i7 5820k ASRock X99 Extreme 4 ASUS ROG STRIX OC GTX 1080 ti G. Skill Ripjaws 4 
Hard DriveCoolingPowerCase
Crucial MX100 512 Custom Loop XFX Pro 1000W Limited Black Edition Modular Thermaltake Core X9 stacked 
CPUGraphicsRAM
i7 4710HQ NVIDIA GTX 860m 2x 4GB 
  hide details  
Reply
Main Rig
(8 items)
 
   
CPUMotherboardGraphicsRAM
i7 5820k ASRock X99 Extreme 4 ASUS ROG STRIX OC GTX 1080 ti G. Skill Ripjaws 4 
Hard DriveCoolingPowerCase
Crucial MX100 512 Custom Loop XFX Pro 1000W Limited Black Edition Modular Thermaltake Core X9 stacked 
CPUGraphicsRAM
i7 4710HQ NVIDIA GTX 860m 2x 4GB 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › [closed] i7-5820K: Can't get my OC stable (Need Advice)