Page 2 of 4 FirstFirst
1
2
3
4
LastLast
  1. #21
    Moderator Cilraaz's Avatar
    15+ Year Old Account
    Join Date
    Feb 2009
    Location
    PA, USA
    Posts
    10,139
    Quote Originally Posted by Drunkenvalley View Post
    Aaah. I just recall people saying that the Asus boards' misbehavior could be fixed by BCLK set to base value and RAM setup manually, etc.
    In the vast majority of overclocking situations (Sandy Bridge only), the BCLK should remain at 100MHz with Spread Spectrum disabled. Spread Spectrum will fluctuate the BCLK to avoid creating a sympathetic EMI field with nearby computers. Spread Spectrum isn't really a bad thing to have enabled, but most people prefer the "cleaner" look of a flat 100MHz BCLK, and I haven't heard of any stories of problems with it disabled.
    Last edited by Cilraaz; 2011-08-15 at 08:34 PM. Reason: Re-worded Spead Spectrum's definition

  2. #22
    Quote Originally Posted by Cilraaz View Post
    1. Don't use IBT to test stability on Sandy Bridge chips. Prime95, while it takes longer, is better at finding instability. You should be able to pass a 4-8 hour Prime95 run (Blend, max memory) when you can really call the system stable.
    2. By "level 5 PLL", I assume you actually mean level 5 LLC (load-line calibration).
    3. I assume you mean that you have voltage set to 1.355v in the BIOS. What does the voltage drop to under load?
    4. Is PLL Overvoltage enabled? (Hint: It should be)
    5. What other settings (other than vCore, LLC, and multiplier) have you changed and what are they?

    The log says 3.32GHz because that's what the CPU is. Its default identifier is what's read by most programs, so its default speed is shown by most programs.
    2. Yup
    3. I was able to lower it to 1.34 vcore in BIOS, under load it goes to 1.356v
    4. It was set to auto, I just now set it to enabled (internal PLL overvoltage is what I'm assuming you were referring to)
    5. I only changed the multiplier to 44x, the LLC level to 5 and the vcore to 1.34v (EDIT: and now the PLL overvoltage to enabled)

    Does anyone see anything wrong? Operating at mid-high 60s Celsius under load atm.


    Edit: hmm.. now it seems cpu-z is constantly reading 1.356v as my core voltage while and while not under load even though my cpu vcore is 1.34v

    Also, to be sure I get an accurate test what should I disable?(ie sleep mode) I'm also assuming it's okay to turn off my monitors over the night I'm testing while I sleep.

  3. #23
    Moderator Cilraaz's Avatar
    15+ Year Old Account
    Join Date
    Feb 2009
    Location
    PA, USA
    Posts
    10,139
    Quote Originally Posted by MrSniffles View Post
    3. I was able to lower it to 1.34 vcore in BIOS, under load it goes to 1.356v
    You really don't want your LLC high enough that it would increase your vCore under load. If your LLC is on a scale of 1-5, back it down to at least 4. At "lower" voltages like the mid 1.3's, you can get away with using LLC to keep your vCore stable regardless of load. Some are pro-LLC; some are anti-LLC. It really depends on whatever you prefer, but in any case, try not to let LLC increase that voltage under load.

    Quote Originally Posted by MrSniffles View Post
    4. It was set to auto, I just now set it to enabled (internal PLL overvoltage is what I'm assuming you were referring to)
    I sure was.

    Quote Originally Posted by MrSniffles View Post
    5. I only changed the multiplier to 44x, the LLC level to 5 and the vcore to 1.34v (EDIT: and now the PLL overvoltage to enabled)
    That should be fine. If you got a bum chip, you might need a little more voltage, but, as an example, I'm running x46 @ 1.355v.

    Quote Originally Posted by MrSniffles View Post
    Does anyone see anything wrong? Operating at mid-high 60s Celsius under load atm.
    Stress test load? If so, that's absolutely great. If you mean gaming load, then it might be a little high, but definitely still safe. The danger point of Sandy Bridge chips isn't until the mid-to-high 90s, but most people typically like to keep them under 75°C-ish. I let mine hit in the 80s when I was doing an Intel Burn Test run with the AVX instruction set... just to see how high I could get it.

    Quote Originally Posted by MrSniffles View Post
    Edit: hmm.. now it seems cpu-z is constantly reading 1.356v as my core voltage while and while not under load even though my cpu vcore is 1.34v
    Are you sure it's not set to 1.356v in the BIOS? LLC won't cause your vCore to raise while idle and it shouldn't be that far off from your BIOS.

    Quote Originally Posted by MrSniffles View Post
    Also, to be sure I get an accurate test what should I disable?(ie sleep mode) I'm also assuming it's okay to turn off my monitors over the night I'm testing while I sleep.
    You shouldn't need to disable anything. Sleep mode shouldn't kick in since the CPU is active (you can disable it if you want to be safe). It is safe to turn your monitors off.

  4. #24
    Quote Originally Posted by Cilraaz View Post
    You really don't want your LLC high enough that it would increase your vCore under load. If your LLC is on a scale of 1-5, back it down to at least 4. At "lower" voltages like the mid 1.3's, you can get away with using LLC to keep your vCore stable regardless of load. Some are pro-LLC; some are anti-LLC. It really depends on whatever you prefer, but in any case, try not to let LLC increase that voltage under load.
    .
    the LLC is a level scale of 1-10, would level 5 be ok in this case?

    Edit: I went ahead and set my LLC to 4 and ran prime95 for a couple minutes, while running my temps were 63-64C and my core clock was 1.32v. While idling my core clock jumps up to 1.356v. Does this seem right and should I lower my vcore of 1.4v?

    Edit 2:While playing wow I noticed that the vcore only increases when my core speed goes to 1.6ghz(idle speed?) and lowers as the core speed increases. Do I have something enabled that shouldn't be??

  5. #25
    Moderator Cilraaz's Avatar
    15+ Year Old Account
    Join Date
    Feb 2009
    Location
    PA, USA
    Posts
    10,139
    Quote Originally Posted by MrSniffles View Post
    the LLC is a level scale of 1-10, would level 5 be ok in this case?

    Edit: I went ahead and set my LLC to 4 and ran prime95 for a couple minutes, while running my temps were 63-64C and my core clock was 1.32v. While idling my core clock jumps up to 1.356v. Does this seem right and should I lower my vcore of 1.4v?

    Edit 2:While playing wow I noticed that the vcore only increases when my core speed goes to 1.6ghz(idle speed?) and lowers as the core speed increases. Do I have something enabled that shouldn't be??
    No. The vCore voltage should remain the same or drop under load. What do you mean by "lower my vcore of 1.4v". You're idling at 1.356v and at load at 1.32v. Where is 1.4v coming from? Your idle voltage should be pretty equal to what you have set in the BIOS.

  6. #26
    I actually have an i5 2500k @ 4.5ghz on a z68 as well...


    Only thing is though : /

    I have to run at level 7 in order to not get a BSOD x101 error.

    Level 5 and level 6 give me that error. I also have the voltage set to 1.360v but while gaming it goes to 1.404 - 1.44volts but the highest temp I get is 62 degrees celcius...
    Last edited by Lirik; 2011-08-16 at 02:57 AM.

  7. #27
    Moderator Cilraaz's Avatar
    15+ Year Old Account
    Join Date
    Feb 2009
    Location
    PA, USA
    Posts
    10,139
    Quote Originally Posted by Lirik View Post
    I actually have an i5 2500k @ 4.5ghz on a z68 as well...


    Only thing is though : /

    I have to run at level 7 in order to not get a BSOD x101 error.

    Level 5 and level 6 give me that error. I also have the voltage set to 1.360v but while gaming it goes to 1.404 - 1.44volts but the highest temp I get is 62 degrees celcius...
    Increasing voltage under load is bad. Increasing voltage by 0.08v under load is horrendous. Regardless of temps, LLC raising your voltage to 1.44v is unsafe. If you must use LLC, find a level that keeps your voltage the same under load and idle, then adjust your vCore appropriately.

  8. #28
    Quote Originally Posted by Cilraaz View Post
    No. The vCore voltage should remain the same or drop under load. What do you mean by "lower my vcore of 1.4v". You're idling at 1.356v and at load at 1.32v. Where is 1.4v coming from? Your idle voltage should be pretty equal to what you have set in the BIOS.
    I have my vcore set to 1.4v on the BIOS. Why is it idling at 1.356v then? I've also been getting some BSODs (mainly 0x124 and some 0x101(think I may have fixed this one)) while stress testing. I got this from a guide:

    0x124 = add/remove vcore or QPI/VTT voltage (usually Vcore, once it was QPI/VTT)
    0x101 = add more vcore
    but I'm not really sure what to do about the 0x124 BSOD.


    On a side note, I don't know if this effects anything or not but it seems my cpu has 3 different clock and core voltage settings for 3 different states, 1.6GHz @ 1.356v(IDLE), 3.2GHz @ 1.344v and 4.4GHz @ 1.32v(Prime95 full load). Granted, these are not always the exact numbers but it seems fluctuate between them for different amounts processing requirements.

  9. #29
    Quote Originally Posted by Cilraaz View Post
    Increasing voltage under load is bad. Increasing voltage by 0.08v under load is horrendous. Regardless of temps, LLC raising your voltage to 1.44v is unsafe. If you must use LLC, find a level that keeps your voltage the same under load and idle, then adjust your vCore appropriately.
    I think I'm missing something here... is the whole point of LLC to increase your voltage under load?

    Also what would be considered a "starting" voltage value?

    1.35 at level 5 or something?

  10. #30
    Moderator Cilraaz's Avatar
    15+ Year Old Account
    Join Date
    Feb 2009
    Location
    PA, USA
    Posts
    10,139
    Quote Originally Posted by MrSniffles View Post
    but I'm not really sure what to do about the 0x124 BSOD.
    I'm not sure why you're sitting so much under your BIOS value, but 0x124 usually requires more vCore. What kind of RAM do you have, though? Sometimes if you have either a high quantity of RAM or high speed RAM, you can get 0x124s if you don't bump the QPI/VTT voltage.

    Quote Originally Posted by Lirik View Post
    I think I'm missing something here... is the whole point of LLC to increase your voltage under load?

    Also what would be considered a "starting" voltage value?

    1.35 at level 5 or something?
    Depending on who you ask, LLC is a savior or evil. LLC was created to allow you to maintain a steady voltage between idle and load. The problem is that it goes against actual Intel specs for the CPU. Check out the following images:



    In this image, you see no LLC being used. When load hits the CPU, the voltage drops. When the CPU goes back to idle, the voltage spikes up to the CPU's VID (or, in an overclocker's case, their vCore setting). Now look what happens when we add LLC:



    LLC is fighting to keep the voltage level. When the CPU goes from load to idle, it receives the same voltage spike as before, but now it's coming from a higher starting point. This is the "micro-spike" that I've spoken of before. If this goes too high, it can cause potential problems. In your case, LLC is putting the voltage to 1.44v under load. When your CPU goes idle, you're spiking OVER that 1.44v, which can potentially be dangerous.

    Intel specifications list maximum VID for the i5 2500k as 1.52v, but this isn't taking LLC into account. If you set your vCore to 1.52v in BIOS with no LLC, you'll see a voltage drop to about 1.4v under load. Therefore, it's been assumed that if you're using LLC, you're safe as long as you remain under 1.4v (as it then shouldn't spike higher than the 1.52v max VID).

  11. #31
    Great post that really clears up a lot of confusion, I appreciate it.

    Also you said to mr. sniffles: "What kind of RAM do you have, though? Sometimes if you have either a high quantity of RAM or high speed RAM, you can get 0x124s if you don't bump the QPI/VTT voltage."

    I have a feeling my RAM might be conflicting with my OCing... not sure, but my ram is at 1600MHZ 2 x 4gb.

    I mainly say this because I know a lot of people use ram at lower MHZ if I recall correctly.

    My ram:

    http://www.amazon.com/gp/product/B004I763AW
    Last edited by Lirik; 2011-08-16 at 03:44 AM.

  12. #32
    Quote Originally Posted by Cilraaz View Post
    I'm not sure why you're sitting so much under your BIOS value, but 0x124 usually requires more vCore. What kind of RAM do you have, though? Sometimes if you have either a high quantity of RAM or high speed RAM, you can get 0x124s if you don't bump the QPI/VTT voltage.
    My ram is G.SKILL Sniper Low Voltage Series 8GB (2 x 4GB), what should I bump the QPI/VTT voltage to if I do need to bump it up?

  13. #33
    Moderator Cilraaz's Avatar
    15+ Year Old Account
    Join Date
    Feb 2009
    Location
    PA, USA
    Posts
    10,139
    For both of you: At 8GB, it might be helpful to bump your VTT up slightly. I raised mine to 1.09v when I upped my system to 8GB. Also, double check your DRAM voltage. Finally, you could try lowering your CPU PLL voltage. Some people have stated that they received more stability between 1.72v and 1.78v, rather than the default 1.8v. I have mine set right in the middle at 1.75v.

    For MrSniffles: Make sure it's within the RAM's specification. If the system is trying to set the default 1.5v for RAM that's specified at 1.25v, that could cause problems.

  14. #34
    Quote Originally Posted by Cilraaz View Post
    For both of you: At 8GB, it might be helpful to bump your VTT up slightly. I raised mine to 1.09v when I upped my system to 8GB. Also, double check your DRAM voltage. Finally, you could try lowering your CPU PLL voltage. Some people have stated that they received more stability between 1.72v and 1.78v, rather than the default 1.8v. I have mine set right in the middle at 1.75v.

    For MrSniffles: Make sure it's within the RAM's specification. If the system is trying to set the default 1.5v for RAM that's specified at 1.25v, that could cause problems.
    I changed my CPU PLL from 1.8v to 1.75v, my DRAM voltage from 1.5v to 1.25v and my QPI/VTT from 1.05v to 1.25v. I couldn't find the VTT you were talking about changing to 1.09v. Also, I saw a few other voltages under the DRAM voltage that read 0.750v (think they had "channel" in the name), do those or anything else need to be adjusted as well?

    Edit: After changing all of that I still got a 0x124 BSOD while stress testing.

  15. #35
    Moderator Cilraaz's Avatar
    15+ Year Old Account
    Join Date
    Feb 2009
    Location
    PA, USA
    Posts
    10,139
    Oh dear god. 0.2v on VTT is not a slight change. I'm talking more like upping it to 1.09v. 1.25v is definitely in an unsafe range.

    QPI and VTT are the same voltage. They're just termed differently depending on motherboard brand.

  16. #36
    Herald of the Titans Sephiracle's Avatar
    10+ Year Old Account
    Join Date
    Aug 2010
    Location
    Colorado
    Posts
    2,729
    @Cilraaz

    Thanks for the post. I wish any of the stuff you guys are talking about is listed the same way in the MSI bios.

    http://www.tomshardware.com/forum/26...clocking-guide

    Has the best guide for mine anyways. Just by changing the Vdroop to low and changing my vCore to 1.38 I can get mine to do 5.0ghz. It gets up to 1.424 under load w/ Prime95 and gets up around 80-85. I need to take some time to play around with the settings since I actually just found that guide, but MSI's don't have LLC levels, they just have the one setting for it.
    LoL: Kr1sys
    WoW:'06 - '11 '14-?' : Krisys - Blood/Frost DK | Sephiracle - Arms/Prot Warrior | Sephyx - Shadow/Disc Priest | Petergriffin - Huntard


  17. #37
    Quote Originally Posted by Cilraaz View Post
    Oh dear god. 0.2v on VTT is not a slight change. I'm talking more like upping it to 1.09v. 1.25v is definitely in an unsafe range.

    QPI and VTT are the same voltage. They're just termed differently depending on motherboard brand.
    oh ok lol, I misunderstood you turned it back down to 1.09v *crosses fingers*

  18. #38
    alrite so I followed your advice and everything Cilrazz. I have honestly come to believe that my CPU is probably just not cut out for 4.5ghz unless I go up to 1.390volts+ which I don't think is worth it for 24/7 usage.

    I "loaded optmized defaults" in the bios and I will be starting at 41x multiplier and seeing where it takes me. Currently I am at 41x, which I'm pretty sure the auto voltage is at 1.270v. I am also currently running the intel burn test while typing this and it hasn't crashed (3rd loop in). At 4.5ghz no matter what vcore/LLC combo I tried (I stopped going up on vcore at 1.375) I blue screened after 20-30 seconds during the burn test.

    So, either I go to 1.390v + for 4.5ghz, which I'm not sure is viable or I stay low near 41x - 43x with "auto" voltage on.

    Also, 44x crashed on auto vcore as well.

  19. #39
    So I ran Prime95 overnight with your settings, Cilraaz, and I woke up to yet another BSOD 0x124.

    Things I've changed so far: clock multiplier: 44x, Vcore: 1.35v, PLL: level 4 of 10, DRAM voltage: 1.25v, CPU PLL: 1.75v,
    QPI/VTT: 1.09v, PLL overvoltage: enabled and realtime OS ratio changes: disabled.

    ---------- Post added 2011-08-16 at 08:17 AM ----------

    I just got a 0x1A BSOD and now when I try to turn on my computer it turns on for a few seconds and then back off and then on again and then off again after a few seconds and so on. D:


    I was able to finally boot the system and reset my BIOS to the default settings, when I opened the BIOS I got a message saying something along the lines of "This system has encountered boot failure due to overclocking or voltage(i think?) problems".
    After this I'm not so sure I want to overclock anymore since it gave me quite a scare, does anything have any insight on this issue?

  20. #40
    Moderator Cilraaz's Avatar
    15+ Year Old Account
    Join Date
    Feb 2009
    Location
    PA, USA
    Posts
    10,139
    Quote Originally Posted by Lirik View Post
    alrite so I followed your advice and everything Cilrazz. I have honestly come to believe that my CPU is probably just not cut out for 4.5ghz unless I go up to 1.390volts+ which I don't think is worth it for 24/7 usage.

    I "loaded optmized defaults" in the bios and I will be starting at 41x multiplier and seeing where it takes me. Currently I am at 41x, which I'm pretty sure the auto voltage is at 1.270v. I am also currently running the intel burn test while typing this and it hasn't crashed (3rd loop in). At 4.5ghz no matter what vcore/LLC combo I tried (I stopped going up on vcore at 1.375) I blue screened after 20-30 seconds during the burn test.

    So, either I go to 1.390v + for 4.5ghz, which I'm not sure is viable or I stay low near 41x - 43x with "auto" voltage on.

    Also, 44x crashed on auto vcore as well.
    There's no reason that 4.5GHz should require 1.390v. I don't think I've seen anyone hit the overclocking wall before at least 4.6GHz on a Sandy Bridge setup. What motherboard are you using? It might be possible that you chose a poor motherboard for overclocking and it's holding you back. I'm honestly grasping at straws right now because a low 4's overclock shouldn't be giving you this much trouble.

    Quote Originally Posted by MrSniffles View Post
    So I ran Prime95 overnight with your settings, Cilraaz, and I woke up to yet another BSOD 0x124.

    Things I've changed so far: clock multiplier: 44x, Vcore: 1.35v, PLL: level 4 of 10, DRAM voltage: 1.25v, CPU PLL: 1.75v,
    QPI/VTT: 1.09v, PLL overvoltage: enabled and realtime OS ratio changes: disabled.

    ---------- Post added 2011-08-16 at 08:17 AM ----------

    I just got a 0x1A BSOD and now when I try to turn on my computer it turns on for a few seconds and then back off and then on again and then off again after a few seconds and so on. D:


    I was able to finally boot the system and reset my BIOS to the default settings, when I opened the BIOS I got a message saying something along the lines of "This system has encountered boot failure due to overclocking or voltage(i think?) problems".
    After this I'm not so sure I want to overclock anymore since it gave me quite a scare, does anything have any insight on this issue?
    All of your settings look good. I'd have a hard time believing that the CPU would need more than 1.35v for x44. You might be experiencing some oddities from your ultra low voltage RAM, or you may just need to bump your QPI/VTT a touch more (say 1.10v or so).

    The double boot you experienced is pretty standard with overclocked Sandy Bridge setups. Mine does it any time I completely power the system off.

    The 0x1A BSoD might not be overclock related (or at least not directly). Are you using the Intel SATA ports for your drives or are you using the Marvell SATA ports (assuming your motherboard has them)? Is there any chance that you could post the minidump file from the crash (usually in C:\Windows\Minidump)?

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •