cancel
Showing results for 
Search instead for 
Did you mean: 

Maximus V Extreme + 3770K - overclocking past 4.6GHz stability problem

feniks
Level 11
hello!

I am totally new to Maximus V BIOS and still learning it. Board is great and actually I'm impressed how well ASUS made it! it was long time no see with ASUS and I'm glad I gave it a shot again finally 😄

so far I updated BIOS to 704 and all drivers to latest versions (win7 x64 sp1 here), all installed correctly and works nicely! 🙂

One issue though (my fault for sure). I was wondering if anybody could give me a hand with MVE BIOS settings for stabilizing my 3770K OC at 4.7-4.9GHz ... I know the CPU is not a problem, because it used to run those clocks stable on a different board with minimal changes in BIOS.

I tried replicating the same voltage (vcore + PLL + vccio) settings on MVE BIOS and it all works up to 4.6GHz (even with PLL override disabled), rock solid stable.

However when I try running 4.7 or 4.8GHz and enabled Internal PLL override, I just can't get it fully stable under load. Tried nearly all PLL voltage settings between 1.55V and 1.85V and no dice. I am using Intel Burn Test 2.54 for stress testing, usually "Standard" 10 rounds for quick stability testing of CPU and it crashes almost always after a few rounds.

Here are my settings (3770K @ 4.7GHz for example):
multi 47
CPU spread spectrum disabled (default enabled)
BCLK 100.0MHz (default)
vcore 1.225v (used to be stable on a different board with 100% LLC), went as high as 1.250V and no luck still
CPU PLL enabled (default was auto)
PLL 1.85V (that was the setting making it nicely stable with lower vcore on a different board)
CPU LLC Extrmee (100%)
VCCIO 1.25V (auto was around 1.10-1.15V)
CPU VRM 500KHz (default 300KHz)

For time being I am running Extreme LLC settings (100% vdroop control) as I had no time in playing around with offsets yet (I will). I also adjusted Digi+ settings to Extreme levels and VRM to highest frequencies, so there is no things coming into play ... again, no luck.

any ideas what I am missing here? why does it seem like it needs MUCH more vcore than actually it needs?

Thanks for input!
.: R3C0NF1GUR3D :.
ASUS MVE :: 3770K bench @ 5.2GHz (delidded) :: 2x4GB Mushkin 996990 @ 2400MHz CL10 @ 1.7V :: evga 670 2GB SLI @ 1280/7108 :: Mushkin Chronos 240GB SSD RAID0 (OS) :: WD RE4 2TB (storage) 2TB :: Hitachi Deskstar 5K300 2TB (backup USB3.0) :: ASUS VG248QE 24'' 144Hz monitor

My rig with pictures
"We are what we repeatedly do. Excellence, then, is not an act, but a habit." - Aristotle
18,703 Views
19 REPLIES 19

feniks
Level 11
never mind the question, I think I understand what happened here. It seems that my former board was overshooting vcore higher (when 100%/extreme selected) than this board hence the initial stability seemed at lower vcore values as entered in BIOS. it's also possible that MVE board has something quirky about Extreme LLC ... well anyways, I opted for using Ultra High LLC and stabilized 4.7GHz easily with vcore a reasonable level!

4.7GHz with 1.26v vcore


Uploaded with ImageShack.us

4.8GHz with 1.32v vcore


Uploaded with ImageShack.us

Really liking this board! Will see tomorrow how far it can take my 3770K ... especially once I start using offset vcore and even lower LLC settings I hope for some nice temps management under load 🙂
.: R3C0NF1GUR3D :.
ASUS MVE :: 3770K bench @ 5.2GHz (delidded) :: 2x4GB Mushkin 996990 @ 2400MHz CL10 @ 1.7V :: evga 670 2GB SLI @ 1280/7108 :: Mushkin Chronos 240GB SSD RAID0 (OS) :: WD RE4 2TB (storage) 2TB :: Hitachi Deskstar 5K300 2TB (backup USB3.0) :: ASUS VG248QE 24'' 144Hz monitor

My rig with pictures
"We are what we repeatedly do. Excellence, then, is not an act, but a habit." - Aristotle

apocalypzez
Level 10
Good one!!! For me at stock settings the vcore is 1.216~ :'(
ASUS Maximus IX APEX || Core i7 7700K || Corsair H150i Pro || EVGA RTX 2080 Ti B.E. || G.Skill Trident Z RGB F4-3000C15D-16GTZR 16GB (2x8GB) DDR4 @ 15-16-16-35-2N || ADATA SX8200 Pro 1TB x 2 (RAID0) || WD RED 3TB x 2 || Corsair AX1200i || NZXT H700i (without the smart hub) || Dell U3011 || Windows 10 Pro x64 || Logitech Z2300 THX Certified 2.1 & Corsair Vengeance 2000 ||

apocalypzez wrote:
Good one!!! For me at stock settings the vcore is 1.216~ :'(


it's not tragic. auto stock settings tend to overshoot vcore for the stock Turbo at 3.9GHz. I recall mine was somewhere around that value as well 😉

It's my second chip actually (from Intel RMA) as the initial one got degraded severely in only 8 weeks of mild use (called for higher vcore too). This one runs on low voltages so far, but temps are reaching high levels as well. I keep it under water (Raystorm block) together with GPU (Razor 670 block) in same loop and 2 rads (RX240 + EX360), with ICD TIM on CPU it seems more manageable this way.

However my biggest problem on former board was the need to use Extreme LLC (which was overshooting by up to +0.04V under load), because all lower LLC settings were undervolting badly and sagging under load.
This MVE seems to run perfect in lower LLC settings and I like it, e.g. Ultra High LLC seems near perfect (no overshoot, minimal vcore drop under load).
.: R3C0NF1GUR3D :.
ASUS MVE :: 3770K bench @ 5.2GHz (delidded) :: 2x4GB Mushkin 996990 @ 2400MHz CL10 @ 1.7V :: evga 670 2GB SLI @ 1280/7108 :: Mushkin Chronos 240GB SSD RAID0 (OS) :: WD RE4 2TB (storage) 2TB :: Hitachi Deskstar 5K300 2TB (backup USB3.0) :: ASUS VG248QE 24'' 144Hz monitor

My rig with pictures
"We are what we repeatedly do. Excellence, then, is not an act, but a habit." - Aristotle

HiVizMan
Level 40
Yes that is correct, a number of boards do over volt higher than what you set in BIOS. It is for that reason that folks tend to use a multi meter to see what is the actual voltages when they go extreme. 🙂
To help us help you - please provide as much information about your system and the problem as possible.

HiVizMan wrote:
Yes that is correct, a number of boards do over volt higher than what you set in BIOS. It is for that reason that folks tend to use a multi meter to see what is the actual voltages when they go extreme. 🙂


Yup, it varies from board to board, but all I've seen overshoot by some degree. Former one up to +0.04V under load (Extreme LLC), this one a tad less, but still Ultra High LLC settings seems much better (near perfect) 🙂

In future I want to try out vcore offsets (my former board didn't have this feature in BIOS) and 50% LLC and see what I can achieve that way, trying to minimize temps under load, otherwise I can't reach higher (stable) speeds past 4.8GHz.
.: R3C0NF1GUR3D :.
ASUS MVE :: 3770K bench @ 5.2GHz (delidded) :: 2x4GB Mushkin 996990 @ 2400MHz CL10 @ 1.7V :: evga 670 2GB SLI @ 1280/7108 :: Mushkin Chronos 240GB SSD RAID0 (OS) :: WD RE4 2TB (storage) 2TB :: Hitachi Deskstar 5K300 2TB (backup USB3.0) :: ASUS VG248QE 24'' 144Hz monitor

My rig with pictures
"We are what we repeatedly do. Excellence, then, is not an act, but a habit." - Aristotle

YJK
Level 7
@Feniks .... dude tell me ur batch number of ur processor , every chip of ivy is not a overclocker chip , i had 2 bad chips of 3570k . but iam switching to 3770k , im looking for a good batch processor , i have a idea little bit of batch numbers ..... for ivy stability 4.6ghz is recomended from intel , after that , its ur luck and also depend on ur vcore
[B]Guns Are Neat Little Things, Aren't They? They Can Kill Extraordinary People With Very Little Effort.


YJK wrote:
@Feniks .... dude tell me ur batch number of ur processor , every chip of ivy is not a overclocker chip , i had 2 bad chips of 3570k . but iam switching to 3770k , im looking for a good batch processor , i have a idea little bit of batch numbers ..... for ivy stability 4.6ghz is recomended from intel , after that , its ur luck and also depend on ur vcore


hehe, thanks 🙂
it's not a golden chip by any means, just above average (better than my former one). it took a bit more vcore to full stabilize 4.7 & 4.8GHz under Intel Burn test 2.54 running Maximum stress mode (using nearly all of my 16GB of RAM) and so the final rock solid stable numbers are at:
a) 4.7GHz with 1.275v vcore with Ultra High LLC (500KHz CPU VRM setting), PLL enabled at 1.8v
b) 4.8GHz with 1.325v vcore with Ultra High LLC (500KHz CPU VRM setting), PLL enabled at 1.85v
c) 4.9GHz seems to call for something between 1.38v up to 1.40v vcore with Ultra High LLC and produced too high temps to stress test it (2 cores hitting thermal limits in minutes with my water cooling at 1.40v vcore), but perhaps I haven't tried hard enough to optimize it yet and maybe IBT stress testing is too taxing for this speed, possibly Prime95 would generate less temps under load, but would take much longer to stabilize instead (I have no patience for Prime95...).

The batch is a new Costa Rica (it's from Intel RMA, my original one crapped out): 3224B210

Here's the golden batch series if you can find it 🙂 - those batches are said to do 5GHz at 1.20v vcore on water cooling with not much trouble if you can dissipate the heat:
3218B...
mentioned here multiple times:
http://hwbot.org/forum/showthread.php?t=45387&page=34

also, it's not only vcore that generates the heat, it's also the higher frequency which makes it run hotter, easy to test at fixed vcore and lower speeds. when using higher speeds at same fixed vcore (if stable of course), you will see it heating up more under stress testers.
.: R3C0NF1GUR3D :.
ASUS MVE :: 3770K bench @ 5.2GHz (delidded) :: 2x4GB Mushkin 996990 @ 2400MHz CL10 @ 1.7V :: evga 670 2GB SLI @ 1280/7108 :: Mushkin Chronos 240GB SSD RAID0 (OS) :: WD RE4 2TB (storage) 2TB :: Hitachi Deskstar 5K300 2TB (backup USB3.0) :: ASUS VG248QE 24'' 144Hz monitor

My rig with pictures
"We are what we repeatedly do. Excellence, then, is not an act, but a habit." - Aristotle

YJK wrote:
@Feniks .... dude tell me ur batch number of ur processor , every chip of ivy is not a overclocker chip , i had 2 bad chips of 3570k . but iam switching to 3770k , im looking for a good batch processor , i have a idea little bit of batch numbers ..... for ivy stability 4.6ghz is recomended from intel , after that , its ur luck and also depend on ur vcore


The current "best Batch number" as found by guys binning them on HWbot and several pro clockers I know are Costa Week 18 as reported by Feniks and Week 29 or Malay Week 20. (Its actually 18C that is the top but B are good too)

My week 18 does 5.0ghz 4c8t on 1.22-1.24v on air and my 29 is even lower volts. The 29 is a tested 6.9ghz chip on LN2.

8 Pack wrote:
The current "best Batch number" as found by guys binning them on HWbot and several pro clockers I know are Costa Week 18 as reported by Feniks and Week 29 or Malay Week 20. (Its actually 18C that is the top but B are good too)

My week 18 does 5.0ghz 4c8t on 1.22-1.24v on air and my 29 is even lower volts. The 29 is a tested 6.9ghz chip on LN2.


heh, I was never super lucky in Intel's lottery 😉 usually my first chip of any model is a disaster and the replacement is slightly above average (nothing crazy) 😉

I wish I had a chip that could do 5GHz @ 1.22v vcore LOL!
.: R3C0NF1GUR3D :.
ASUS MVE :: 3770K bench @ 5.2GHz (delidded) :: 2x4GB Mushkin 996990 @ 2400MHz CL10 @ 1.7V :: evga 670 2GB SLI @ 1280/7108 :: Mushkin Chronos 240GB SSD RAID0 (OS) :: WD RE4 2TB (storage) 2TB :: Hitachi Deskstar 5K300 2TB (backup USB3.0) :: ASUS VG248QE 24'' 144Hz monitor

My rig with pictures
"We are what we repeatedly do. Excellence, then, is not an act, but a habit." - Aristotle