P8P67Pro + 2500K, I want 4.7Ghz!

Discussion in 'Intel Processors' started by Akerhage, May 29, 2011.

  1. Akerhage

    Akerhage n00b

    Messages:
    11
    Joined:
    May 29, 2011
    Have som problems when overclocking, this is my first time doing this so little noob.

    Tests:
    4.6Ghz @ 1.36v - BSOD after 90min with prime and I dont want the VCore higher. (but I think it should work here, because 4.5 at 1.32 seems to work)
    4.5Ghz @ 1.35v - seems to work fine, still running tests (2hours)
    4.5Ghz @ 1.32v - Have run tests 1hour, worked fine.

    The temps are ok, from 58-68 aprox in theese settings
    In cpu-z the voltage says 1.32? - running it on 1.36 i BIOS - why is that??

    So is there anything I can do now or is it just to accept that I cant be able to have the cpu at speed over 4.5 without going further with the Voltage?

    If there are possible ways for me to achieve 4.6 with a Vcore under or at 1.35 please give me some tips for things to try out before I give up at this speed. And can you guys confirm that 1.35 is the critical 24/7-voltage? Or is it okay maybe to have 1.38 if I can run 4.7 with that settings?

    Thanks!
     
  2. Ocellaris

    Ocellaris Ginger @le, an alcoholic's best friend.

    Messages:
    18,798
    Joined:
    Jan 1, 2008
    Last edited: May 29, 2011
  3. Akerhage

    Akerhage n00b

    Messages:
    11
    Joined:
    May 29, 2011
    No I understand that, but maybe someone can guide me to other settings that may lead to the 4.6 or 4.7 to work with 1.35 =)
     
  4. GoldenTiger

    GoldenTiger [H]ard as it Gets

    Messages:
    18,569
    Joined:
    Dec 2, 2004
    I know this is a basic but it's worth checking.. did you enable Internal PLL Overvoltage in the BIOS? That is required for multi's 45-47+ depending on the exact CPU's multi wall. No one knows the exact "critical 24/7 voltage" but many have run CPU's at 1.45v as 24/7 usage with no degradation at all for months now, some have said trying 1.52v+ for 24/7 lead to some degradation (i.e. overclock starting to fail and need more voltage, not working at lower speeds gradually as you keep downclocking it, etc.). 1.45-1.46v is generally regarded as safe, personally I'm going to stick to that or less with my 2600K (it's currently doing 4.8ghz @ 1.395v and has been since the chips came out). Do ensure you have Load Line Calibration (LLC) on Ultra High as well, and the VRM management settings on "Extreme".
     
  5. FemmeFatale

    FemmeFatale [H]Lite

    Messages:
    121
    Joined:
    Oct 21, 2010
    Why not just stay at 4.5ghz?
     
  6. Ocellaris

    Ocellaris Ginger @le, an alcoholic's best friend.

    Messages:
    18,798
    Joined:
    Jan 1, 2008
    Read the thread I linked... it is exactly what you are looking for.
     
  7. trick0502

    trick0502 [H]ardness Supreme

    Messages:
    5,133
    Joined:
    Apr 17, 2006
    i followed the linked thread and got my cpu to 4.5 @1.275 in bios (1.282 cpuid).
     
  8. Akerhage

    Akerhage n00b

    Messages:
    11
    Joined:
    May 29, 2011
    thanks, I have followed a guide to get to this point, but still stuck at 4.5, but will check the link, thanks!
    And I can stay on 4.5, but want to try out to fix with this and it is pretty funny to do this and learn something new!
     
  9. Akerhage

    Akerhage n00b

    Messages:
    11
    Joined:
    May 29, 2011
    As requested, my settings. I have done some changes with LLC, PLL, VCCIO, Cstates and VRM. And I have seen that thoose settings did make the VCore to be a little bit lower and no BSOD with that as I got before. So please check my settings and comment on those!

    A question about System Agent Adapter, should the Initiate Graphic Adapter be at PEG/PCI?
     
    Last edited: May 30, 2011
  10. Akerhage

    Akerhage n00b

    Messages:
    11
    Joined:
    May 29, 2011
    The new BIOS for P8P67Pro is now avalible on the Asus website! 1606, I have just updated the BIOS with it!
     
  11. Akerhage

    Akerhage n00b

    Messages:
    11
    Joined:
    May 29, 2011
    My new OC-settings. (2011-06-01)
    4800Mhz @ 1.4V (To High VCore??)

    Max temps with Prime95: 57,64,65,62
    VCore CPU-Z: 1.376-1.384
     
  12. Solitude

    Solitude [H]ard|Gawd

    Messages:
    1,567
    Joined:
    Dec 16, 2010
    please update here and there because I hope to run @4.8 as well when my new mobo arrives. I will put my results here too.
     
  13. Akerhage

    Akerhage n00b

    Messages:
    11
    Joined:
    May 29, 2011
    Will inform you if I succeed! =)
     
  14. zUg_zUg

    zUg_zUg n00b

    Messages:
    56
    Joined:
    Nov 10, 2005
    did you have any issues with the PC not booting up and just getting stuck at a black screen with a underscore blinking? its happened to me a few times already and ive done everything from reseating the cpu, reseating ram, and even using individual ram sticks. its getting quite frustrating
     
  15. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    53,876
    Joined:
    Feb 9, 2002
    Quite often you need more voltage than that. When I take 2500K's and 2600K's to 4.4GHz+ I often need 1.40v to do it. Additionally you can try enabling CPU-PLL overvoltage which sometimes helps get an extra multiplier or more. As for the voltage being too high, I wouldn't worry about it. I believe they have a maximum ceiling of 1.52v. Anything under 1.5v should be plenty safe. You usually don't need to make very many adjustments to the motherboard settings. You'd be surprised how far you can get tuning almost nothing on Sandy Bridge / LGA1155 based systems. Usually I just need to increase the CPU voltage, CPU-PLL overvoltage, load-line calibration / vdroop control, and power phase control since ASUS, Gigabyte and MSI all use some form of dynamic phase switching which is less than ideal for higher end overclocking results.

    The automatic setting is fine. PEG or PCI settings are only needed if you have more than one type of graphics card installed and want to change which one is the primary adapter. It won't have any effect on overclocking.
     
    Last edited: Jun 4, 2011
  16. Akerhage

    Akerhage n00b

    Messages:
    11
    Joined:
    May 29, 2011
    Thanks for the answers.
    I havent had crashes before windows loads. So I have managed to get into windows all the time.

    And for the voltage setting to 1.52, that is the VID they are talking about. And I dont think its ok to go over 1.4v for a 24/7-setting.

    I am "pretty" stable now at 4.7@ 1.4v, really dont want to primetest all the time, a little bit afraid of cpu degradation due to all the primetests I am doing.

    Have CPU PLL @ 1.706, LLC-ultra high, cstates enabled,phase and duty to extreme.
     
  17. tangoseal

    tangoseal [H]ardness Supreme

    Messages:
    7,323
    Joined:
    Dec 18, 2010
    Dude I run 1.42 to hit 4.8 stop being a sissy.

    4.5 is a joke to obtain stability at but you hit a wall of low voltage works no more once you go higher.
     
  18. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    53,876
    Joined:
    Feb 9, 2002
    Its perfectly OK to go beyond 1.4v for 24/7 operation. Again Intel's VID limit is 1.52. Anything from 1.52v or lower should be perfectly fine for every day use so long as your thermals are in line. CPU degradation will not occur from prime testing.
     
  19. zUg_zUg

    zUg_zUg n00b

    Messages:
    56
    Joined:
    Nov 10, 2005
    @ dan

    if you running at 1.4 do you need active cooling on the chipset?
     
  20. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    53,876
    Joined:
    Feb 9, 2002
    No. I've done this many times using the stock chipset cooling.
     
  21. TwistedAegis

    TwistedAegis [H]ardForum Junkie

    Messages:
    8,958
    Joined:
    Oct 7, 2009
    Hmm, thanks for this. I seem to have hit a wall at 4.7 (on my 2600k) but was thinking 1.4+ was too high. I'll have to go back and see if I can push it up to 5.
     
  22. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    53,876
    Joined:
    Feb 9, 2002
    Again 1.52v is their maximum safe limit according to Intel's own white papers. Remember to account for vDroop / LLC when setting your voltages. Also keep in mind that many 2600K's can't go beyond a 44x-47x multiplier. Yours might be one of them. Even fewer can go beyond 48x. Though Kyle's got one 2500K that does 50x multipliers on most boards just fine.
     
  23. TwistedAegis

    TwistedAegis [H]ardForum Junkie

    Messages:
    8,958
    Joined:
    Oct 7, 2009
    I'm pretty sure my maximum is 47x anyway; even with upping my voltages, putting bclck up to 101 created stability issues, which was disappointing as I figured with 4.7 stable at only 1.35, I had some breathing room. But I never really pushed all the way to 1.4 as I didn't want to run it 24/7 at that level, however now I'll give it a go and even push beyond it a bit just to see if I can get it stable, although I may not leave it at that if it goes much higher than 1.4.
     
  24. tangoseal

    tangoseal [H]ardness Supreme

    Messages:
    7,323
    Joined:
    Dec 18, 2010
    LOL some people dont know how to overclock. There is more to OC than adjusting multi and voltage. There is oh .... about 5 more settings you need to look at... PLL, LLC, etc....

    I have hit 5.0 ghz on my 2500K and 5.4 on my 2600k ... oh there was some voltage but I needed to see the max. Then of course I clocked them to 4.8 since it was a lower voltage and honestly anything over is just a waste of power bill.
     
    Last edited: Jun 7, 2011
  25. tangoseal

    tangoseal [H]ardness Supreme

    Messages:
    7,323
    Joined:
    Dec 18, 2010
    Where are you getting this information from? I have seen 4 different 2600ks all friends of mine in the IT world that easily surpassed 5ghz and they are not even close to the same batch numbers.

    I think 4.7 is the maximum overclock of people who are afraid to use a little electricity!

    and....

    Run that chip with a little more aggressive load line and bump it to 1.43v and you will laugh at your self for not hitting 4.8-9 earlier. Also drop your CPU PLL down a few 100ths of a volt. I bet your PLL is a tad too high or set to Auto... that is meh. Also a junk ass power supply will not deliver clean enough of a wavelength of electrical flow to your components which will cause instability. I was NEVER able to hit 5.0+ with my last PSU. I got a AX1200 (top of the line PSU) and boy what a difference in clean power.
     
    Last edited: Jun 7, 2011
  26. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    53,876
    Joined:
    Feb 9, 2002
    ASUS' documentation on the subject:

    1. Approximately 50% of CPUs can go up to 4.4~4.5 GHz
    2. Approximately 40% of CPUs can go up to 4.6~4.7 GHz
    3. Approximately 10% of CPUs can go up to 4.8~5 GHz (50+ multipliers are about 2% of this group)

    Many CPUs (according to the information I've seen) can't go past 48x. Many also can. All the 2600K's I've personally worked with can with no problem. And 103MHz BCLK x 48 = 4944MHz which is pretty damn close to 5GHz. Most systems should be capable of getting near these speeds if they are using quality components.

    On this we agree. Adding to that; I've personally seen chips that wouldn't do more than a 49x or 50x multiplier on a particular board even if the CPU could. So in reality some boards / systems could hold back some CPUs. Keep in mind I often use the same two or three CPUs when I overclock on the test bench and I've used CPUs I have certainly taken to 50x plus which won't go beyond 47x or 48x on a given board. On some boards I simply have to make more adjustments to get over that 48x "hump" while on some boards I can't do it at all.

    He definitely needs to get over the "running at 1.4v+ is not safe" bit. That is surely holding him back.
     
  27. E4g1e

    E4g1e [H]ardness Supreme

    Messages:
    7,056
    Joined:
    May 21, 2002
    True. But in my case, the CPU temps are holding me back: Once I raised the CPU voltage above 1.4V regardless of the CPU clock speed, the load temps began to exceed the CPU's thermal danger zone of 90°C. And in my setup none of the air coolers - not even expensive ones - would have been of much help. I would have needed super-expensive water cooling or liquid nitrogen just to even use CPU voltages above 1.4V.
     
  28. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    53,876
    Joined:
    Feb 9, 2002
    I'm not sure what you can do on air actually. I've always used water cooling on all the Core i5 / i7 processors.
     
  29. QuiteSufficient

    QuiteSufficient Sufficiently [H]ard

    Messages:
    5,152
    Joined:
    Mar 2, 2001
    Mine does 5ghz with everything on auto except manual voltage on P8P67. No LLC, PLL, or any other adjustments.

    1.44v under load, 24/7 for 3 months now. Brother has same setup.

    OP, I think you will need more volts. I don't know what you paid for it, but how I look at it is that the chip was $179 before tax... it would suck to break it, but not the end of the world.


    That is your chip's limit then - which is unfortunate. My bro is doing it with the Corsair A70, which isn't exactly spectacular.

    Have you tried without hyperthreading? I've only played with one 2600k...would be interesting to see how that changes things.
     
    Last edited: Jun 8, 2011
  30. BababooeyHTJ

    BababooeyHTJ [H]ardness Supreme

    Messages:
    6,946
    Joined:
    Jan 21, 2009
    He is also running Loadline calibration on ultra high.

    With loadline set to ultra high it probably isn't for 24/7. I wouldn't go any higher personally. Intel's 32nm chip aren't exactly as hearty as their almost indestructible 45nm processors.
     
  31. tangoseal

    tangoseal [H]ardness Supreme

    Messages:
    7,323
    Joined:
    Dec 18, 2010
    Man you dont need LLC @ Ultra high. That is seriously for people hitting 5.1+ ghz etc... with the cooling to support it of course.

    Just run your LLC at 50% or High depending on your board. Also drop your PLL voltage a little.

    And lastly if your temps on Air are not lower than 90C regardless of the cooler it sounds to me like the control factor in your experimentation is your case's airflow. You need to improve airflow, buy an AC for your computer room/office, or just take the board out of the case, lay it on an ESD bag or foam and then test your system in fully open air environment.

    Temps are not an indicator of your chips ability to overclock and they damn sure are NOT an indicator of where your particular chip hits a wall. The wall you are hitting is not getting cool enough air to the heatsink where it matters.
     
  32. BababooeyHTJ

    BababooeyHTJ [H]ardness Supreme

    Messages:
    6,946
    Joined:
    Jan 21, 2009
  33. blade52x

    blade52x 2[H]4U

    Messages:
    2,965
    Joined:
    Sep 9, 2006
    It's hard to get these chips 100% stable at 4.5+ imo. While I got mine to do 4.6 with 1.35v under load, I still get the very rare reboot even though i am prime/occt and etc stable. Though it has something to do with me using offset mode which drops my voltage on idle. But I would rather keep that than waste energy when it's not needed. I would try 1.4v or higher but... It's just too hot now. Maybe in a few months.
     
    Last edited: Jun 16, 2011
  34. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    53,876
    Joined:
    Feb 9, 2002
    The energy wasted is minimal. It isn't as if a slight voltage tweak to your CPU is going to cause your electric bill to sky rocket.
     
  35. QuiteSufficient

    QuiteSufficient Sufficiently [H]ard

    Messages:
    5,152
    Joined:
    Mar 2, 2001
    How many of these chips have you owned? Just the 1 in your signature?

    My buddy and I bought 6 2500ks when they first came out and tested them all at once. I posted about it in an old thread. Some of them definitely hit a multiplier wall and started skyrocketing on heat with the same settings. That isn't just some magical bullshit people on the internet made up. And all of my testing at the time was on a big water loop.
     
  36. Dan_D

    Dan_D [H]ard as it Gets

    Messages:
    53,876
    Joined:
    Feb 9, 2002
    Ultra high is the most you probably ever need on an ASUS board. High is usually more than sufficient. In fact I rarely find that setting to make any real difference once set to high.

    50% load-line calibration may not be enough. The high setting is usually enough in my experience. Ultra High is needed on occasion, but not generally. Not on ASUS boards at least. I've typically found CPU PLL overvoltage to work best around 1.9-1.95v. I've never reached a higher overclock by going over that. You are right in that ambient air temperatures are indeed important as is the air flow of the case. No argument there. However temperatures do not necessarily indicate a processor's overclockability. Especially if the ambient temperatures and case air flow aren't what they need to be. The temps will hold you back in those situations but that's not the CPUs fault. 4.8GHz or so is doable but honestly you need water cooling a lot of the time to pull it off. Few people have the air cooling, air flow and ambient temperatures needed for sustainable 5.0GHz overclocks on air. Not saying it's impossible but on all the chips we use, even on an open air test bench no air cooler would get it done. At least not well enough to stuff the motherboard and CPU into a case and use it daily. The load temperatures are usually too high for air cooling to succeed in those circumstances.

    No it isn't. I do it all the time. In fact I don't usually run into my first real hurdle until 4.7GHz. Actually that's not quite true. I usually run into the need to increase the CPU voltage around 4.4GHz or so on most boards. After increasing the voltage I usually see speeds of about 4.8GHz or so. It's multipliers of 49x or more where I find I've really got to start tweaking. Even then I still don't have to adjust that many settings to get the job done. Again CPU-PLL overvoltage, and some higher levels of load-line calibration and maybe one or two other things usually do the trick.

    I've done this myself with at least 5 or 6 different chips. Maybe more. I'm not sure how many Kyle actually has or how many of those I've used. I'm certain I've used at least three different CPUs. I used his first sample, one retail CPU and another engineering sample. I've built two rigs for friends using a 2600K and a 2500K. I've been able to hit 4.4GHz with no effort. I've tested their rigs at 4.8GHz with minimal adjustments. The same adjustments that got the job done on other ASUS boards. It's really easy to do this. But you are correct. Some CPUs have different thermal properties and some simply hit the multiplier wall faster than others. The motherboard also factors in as well despite the CPU being the biggest variable. I've seen chips that will do 5.0GHz fairly easily hit the wall at 4.7GHz on some boards. Of course this is before most vendors started providing the CPU PLL overvotlage setting. After that I've been able to get past 4.8GHz fairly consistently.
     
  37. tangoseal

    tangoseal [H]ardness Supreme

    Messages:
    7,323
    Joined:
    Dec 18, 2010
    OP, how are things working out for you so far? Were you able to achieve any semblance of success after all this information was being offered here?
     
  38. needmorecarnitine

    needmorecarnitine [H]ardness Supreme

    Messages:
    4,605
    Joined:
    Feb 10, 2004
    he said he is pretty stable at 4700MHz. In that post right before you called him a sissy
     
  39. QuiteSufficient

    QuiteSufficient Sufficiently [H]ard

    Messages:
    5,152
    Joined:
    Mar 2, 2001
    Dan my question was directed at Tangoseal, but my initial testing was on an Asrock Extreme4, and with the BIOS update provided at the time CPU PLL overvoltage was available.

    It is also possible that the manufacturing on the chips has gotten better over time, even though an official stepping change has not been announced?

    On the Asrock I had then only 1 of 6 chips would do 5ghz. 3 of them capped around 4.4, then heat shot up.

    The three machines I have built recently have been Asus P8P67/2500k, and all 3 have done 5ghz.

    I have not owned a 2600k, which is why I asked above if anyone had tried with and without HT.
     
  40. HardBytes

    HardBytes Limp Gawd

    Messages:
    278
    Joined:
    Nov 23, 2004
    @Dan

    You mention CPU-PLL overvoltage...do you really mean CPU PLL voltage? I can't find what you refer to in my BIOS.