Overclock.net › Forums › Intel › Intel CPUs › My 2500k Overclocking Log on ASUS P8P67
New Posts  All Forums:Forum Nav:

My 2500k Overclocking Log on ASUS P8P67

post #1 of 6
Thread Starter 
Hello. I picked up a 2500k two days ago along with an ASUS P8P67 WS Revolution motherboard and I have been working to overclock it. As always, I read all the information I could find on Google and OCN. I started posting in the Sandy Bridge Stable Club with questions and comments, but it occurred to me that I hadn't seen a full OC log from anyone yet. There are tons of guides, but I thought it would help to put together a real work experience with specific hardware that perhaps will help others. I will edit this OP or post additional messages in this thread with updates.

My goal is the highest, (relatively) safe OC for 24/7 Folding@Home and gaming - not one-time benchmarks.

My System Specs (copied here in case my sig rig changes in the future):
  • ASUS P8P67 WS Revolution motherboard, 1202 BIOS
  • Intel i5 2500k CPU
  • 2x2GB OCZ Reaper HPC DDR3 @ 1600MHz 6-8-6-24-1T stock
  • OCZ Vertex 3 SSD with latest firmware
  • Hyper 212+ Cooler with MX-2 TIM
  • 2 x ASUS GTX 480 @ 800/1600/2000 SLI (not relevant to this OC)
  • Corsair HX850W PSU (90% efficient)
  • case, fans, HDDs, and other non-relevant hardware

Based on my research, I set the following to start:
  • Enabled power saving features or set to auto: C1E, C5/C6, EIST (I would normally disable these, but I read that ASUS motherboards are more stable with them enabled)
  • RAM voltage set to 1.65V stock
  • VCCIO set to 1.15V because I read this should be at most 0.5V lower than RAM ovltage
  • PLL Overvoltage Support enabled since I'm shooting for an OC abote 4.5GHz (may disable this later)
  • Spead Spectrum enabled (I would normally disable this, but I read that ASUS boards are more stable with it on)
  • Switched CPU voltage and other relevant settings to manual in the BIOS
  • LLC set to 75% (Ultra-high)
  • VRM Frequency set to 350 (from 300) based on research
  • Duty and phase control set to Extreme.
  • CPU current set to 140% based on other guides
  • Set other BIOS settings as per usual (ie. AHCI mode, etc.)
  • Disabled motherboard components that I'm not using

What I have done so far:
  1. I tested for stability at stock settings.
  2. To get a ballpark of what my chip is capable of, I left vCore on auto and increased the multiplier to 4.5GHz. The automated voltage was set to 1.3625V and the system was stable after some quick prime95 tests
  3. I increased to x48 and x50 just to see if the system would boot with everything on auto (who isn't curious?). It booted into Windows but with a voltage around 1.55V - way too high! At least I know it's possible.
  4. Based on some average voltages required for 4.8GHz, I set my vCore to 1.4V (vPLL still on auto) and set to stress test.
  5. I received lots of BSOD 101 errors as I slowly raised the vCore from 1.4 to 1.43 in small increments. I read that BSOD 101 usually indicates that more vCore is needed.
  6. At 1.43V, the system passed prime95 1344 and 1792 FFT tests for 15 mins each. I set up a blend test and went to bed. In the morning, I found that some workers had stopped after about 50 minutes.
  7. I read that lowering the vPLL is supposed to help the system get stable at lower vCore so I tried 1.75V, 1.7V, 1.6V, and 1.55V to give me an idea of the effect. None of these really improved stability (yet), but I did find that PLL Overvoltage needed to be enabled at 1.6V or lower to boot.
  8. After reading that some chips liked higher vPLL instead of low, I dialed it all the way up to 1.89V just to see what would happen. Still not stable so I increased slowly to 1.45 vCore. At these settings, the system passed 30 minutes of prime95 1344 and 1792 FFTs. It passed an hour blend test. I set up F@H and folded without errors or freezing for about 20 hours.
  9. Now that I thought I was stable, I was going to try to lower the voltages to see what my minimum was. I tried lowering the vCore slowly down to 1.44, but I got more BSOD 101s. Then I tried lowering the vPLL slowly. I tested 1.7V to 1.89V in small increments. To make a long story short, my original settings were not fully stable after all - but vCore was probably close. I decided not to go above 1.45V.
  10. I have dropped my multiplier to x47, set vPLL to Auto, vCore to 1.44V and started over. As of now, the system has passed prime95 1344 + 1792 for 45 minutes. I suspect the system is stable, but I'm going to try to push the voltage lower and re-test.
  11. Testing at 1.41V, 1.625 vPLL wasn't quite stable and produced a failed worker in prime95. I have tested at 1.42V for about 30 minutes prime95 1344 + 1792 FFTs and no problems yet. I suspect that 1.415V will be my minimum stable with vPLL set to 1.625. However, I may try to push lower. I am seeing Vdroop of about 0.2V in the OS.
  12. I am now stable at x47, vCore at 1.415V, vPLL at 1.625V. I have tested with prime95 1344 + 1792 FFTs for about 2 hours and prime95 blend test for 4 hours. I have done 12 hours of F@H and that will run 24/7. Temperatures max at 72C when stress testing and 56C with F@H. Next I will try to overclock my RAM and possibly test lower PLL voltages.

Temperatures:
25C ambient
  • 76C max stress test at 1.44-1.45 vCore x48, 60C F@H 100%
  • 73C max stress test at 1.42 vCore x47
  • 71C max stress test at 1.40 vCore x47
  • 71.5C max stress test at 1.405 vCore 1.625 vPLL x47

This project is currently active so any advice is welcome. Please point out any errors or inconsistencies from my settings above. I hope this helps someone
Edited by klaxian - 11/7/11 at 9:12am
post #2 of 6
I got a similar experience when trying to find the optimal PLL and VCCIO here. Though I get allot higher temps, @ 1.35 vcore I am already hitting 80 during IBT.
The cpu current capability is a notch higher then i would like it to be, also I couldn't determine whether it was affecting the stability.
I find the most difficult part to be, getting it stable 12+ hours of prime blend. Cuz if it's not the system will crash..
Edited by grivy - 11/5/11 at 1:52pm
yes
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5 2500k @ 4.5GHz Asus P8Z68-V IGP HD3000 @ 1.9GHz  Kingston 8 GB 1600 
Hard DriveHard DriveHard DriveOptical Drive
OCZ Vertex 3  Samsung EG 2TB WD Red 3 TB Samsung SH-S223C 
CoolingOSMonitorKeyboard
Scythe mugen 2 Rev. B Windows 8 Pro LG 32LG3000 Logitech K360 Purple Boulder 
PowerCaseMouse
Antec Basiq BP550Plus Xigmatek Midgard Logitech MX Air 
  hide details  
Reply
yes
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5 2500k @ 4.5GHz Asus P8Z68-V IGP HD3000 @ 1.9GHz  Kingston 8 GB 1600 
Hard DriveHard DriveHard DriveOptical Drive
OCZ Vertex 3  Samsung EG 2TB WD Red 3 TB Samsung SH-S223C 
CoolingOSMonitorKeyboard
Scythe mugen 2 Rev. B Windows 8 Pro LG 32LG3000 Logitech K360 Purple Boulder 
PowerCaseMouse
Antec Basiq BP550Plus Xigmatek Midgard Logitech MX Air 
  hide details  
Reply
post #3 of 6
I have read VCCIO to match ram voltage. 1.5v ram = 1.05v vccio

Disable CPU speed spectrum, even Asus says so.

LLC set to 0%. Your voltage with it enabled is actually a lot higher than it shows in OS. Ypu are probably feeding it 1.5v instead of 1.45v. Disable it and you will find you need a higher voltage. Then there is the degradation, mini spikes, and low fluctuating idle voltages. Higher idle volts don't matter at all, volts only matter under load.

Disable C3, C6. Core parking is bad for stability when going from idle to load and vice versa.

Try CPU offset mode. You will have to set phase control to manual->ultra fast.

Disable iGPU, set it to offset power, set to -, set max value, set igpu llc to standard, overcurrent 100%, max frequency to lowest, and ignore igpu voltage monitoring.
post #4 of 6
Quote:
Originally Posted by amvnz View Post
LLC set to 0%. Your voltage with it enabled is actually a lot higher than it shows in OS. Ypu are probably feeding it 1.5v instead of 1.45v.
COST EFFECTIVE
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5 2500K @ 4.4GHz 1.24v Asus P8P67 Vanilla ASUS GTX660Ti DirectCU OC @ 1,250MHz/7,000MHz G.Skill Eco 1600Mhz 7-8-7-24 
Hard DriveHard DriveOptical DriveCooling
WD Caviar Blue 1TB (WD10EZEX) Intel 520 120GB LG SuperMulti CoolerMaster Hyper 212+ 
OSMonitorKeyboardPower
Windows 7 Home Premium x64 BenQ E2420HD Filco Majestouch 2 Tenkeyless w/ MX blues Seasonic SS-650HT 
CaseMouseMouse PadAudio
Corsair 500R Logitech G400 Steelseries QCK Asus Xonar DG 
AudioOther
Fiio E17 Alpen Beyerdynamic DT770 + Zalman clip-on mic 
  hide details  
Reply
COST EFFECTIVE
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5 2500K @ 4.4GHz 1.24v Asus P8P67 Vanilla ASUS GTX660Ti DirectCU OC @ 1,250MHz/7,000MHz G.Skill Eco 1600Mhz 7-8-7-24 
Hard DriveHard DriveOptical DriveCooling
WD Caviar Blue 1TB (WD10EZEX) Intel 520 120GB LG SuperMulti CoolerMaster Hyper 212+ 
OSMonitorKeyboardPower
Windows 7 Home Premium x64 BenQ E2420HD Filco Majestouch 2 Tenkeyless w/ MX blues Seasonic SS-650HT 
CaseMouseMouse PadAudio
Corsair 500R Logitech G400 Steelseries QCK Asus Xonar DG 
AudioOther
Fiio E17 Alpen Beyerdynamic DT770 + Zalman clip-on mic 
  hide details  
Reply
post #5 of 6
Thread Starter 
So I'm currently running at 4.7GHz with 1.42 vCore and 1.625 vPLL. I ran prime95 1344 + 1792 tests for an hour and a blend test for 3.5 hours so far. Seems stable to me, but I'll set up a longer run overnight. If this proves reliable, my next step will be to try to reduce the voltage a bit more.
post #6 of 6
Thread Starter 
@amvnz
Wow you pretty much contradicted everything I've read about overclocking SB on ASUS boards. I normally disable all the power saving features too but others say ASUS motherboards specifically are more stable with them on. Perhaps I will do some testing.

I also like you're idea off disabling the iGPU.

Do you have any proof of you're other suggestions? Thanks.

By the way, I seem to be stable now at 4.7GHz 1.415V with 1.625 vPLL. Still testing...

Sent from my DROID BIONIC
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › My 2500k Overclocking Log on ASUS P8P67