Vdroop Vdrop Or something else? 2500K overclock

S

Stanfiem

Limp Gawd
Joined
Oct 21, 2011
Messages
364
My system is what is in my signature except for I am at 4.2ghz. I know this is not necessary but just to be safe I Play skyrim, CS:S, Batman AC, Oblivion, Fallout NV, on a 1080p 46 inch TV.

While overclocking and stress testing my system I have noticed a slight drop in volts that has gotten bigger since I have increased the multiplier and vcore.

Constant Values
Multi: x33
C1:auto
C3 and C6: auto
Os OC: disabled
RAM Freq.:1866
Ram Timing: 9,10,9,28, 2T
Ram Volt: 1.5 Bios reads 1.510
Chipset: Sandy Bridge Rev 09
Bios Version: F1 I have thought about flashing to F6 but have not read enough about it and have not had any problems with F1 thus far unless F6 corrects What I am seeing, be it Vdroop or Vdrop or something else, I am just going to stick with F1

Turbo x42
vcore: 1.305
Bios Vcore reading: 1.332
Multi-steps Load-Line: Disabled
Load Vcore (under stress): raises to 1.34 then drops to 1.284 to 1.296

Turbo:x43
Vcore: 1.310
Multi steps Load Line: disabled
Load Vcore: goes from a steady 1.296 at idle to 1.284 then it drops to 1.248 a couple of times and raises and settles at 1.284
When I raised the turbo to 43 on all 4 cores my cpu no longer drops volts and multi at idle

Turbo: x44
Vcore:1.385
Multi Steps Load Line: Disabled
Load Vcore: 1.308 to 1.332

Using OCCT's LInPak Idle I get 1.008 to .9 then under load my Vcore soots up to 1.340 then back down to 1.296 While running Linx I hover around 1.284 to 1.296 Is this normal or should I enable a level or so of Multi-step Load-Line.

I have thought about setting Load Line Calibration to level 1 or 2 to correct some of what I think is vdroop but was asking for some advice before I push this chip any further and do this. I remember reading somewhere about how enabling load line calibration was bad for sandy bridge chips but people were saying that it did not apply to the newer 2x000k chips. Has anyone ever heard this, and can comment on if this is true or not.

I ask because This is my first overclock. I have read up quite a bit and have a pretty good grasp of most of the "steps" or what goes into overclocking but would love some input. and for reference my temps are good during all of this i have yet to see a core go above 62 on serious load and that is just a max while gaming i get about 40 to 48 max.

Also Mods I posted this in the Intel Processor forum and realized it might be better in the overclocking section. Sorry
 
Last edited:
Also anyone find it tough to get an accurate reading on the VCore Every program I have says something different the closest to what I have set is the Gigabyte Monitor/ tuning software (used only for monitoring vcore)
 
The Vdroop on my board is pretty bad, I have to use LLC 1 to maintain my 46x10 with a +.030 DVID. Is LLC set to auto? I'd try standard at least to eliminate the possibility of the bios screwing with it.

As far as i've read LLC should be fine assuming it doesnt nuke the chip with an ass-ton of volts (1.7 lol)

LLC 2 spiked my voltage to 1.6 or so when I was trying some higher oc's, and couldn't get it stable with a negative DVID.... I havent tried that OCCT test before. I just prime with max power consumption. I also have to use the craptastic eazytune to get vcore readings.

My version of that board does not have a vcore setting, I have to use dvid instead. Not sure how vcore, dvid, and llc all jive together with the sandy bridge and its auto overvolting when you increase the multiplier.
 
Well I just looked at the graph I had and the voltages are way off with OCCT I think it measures the offset voltage or something because it was reading 1.3 volts then a raise to 1.4 volts when the volts are at 1.296 and drop to 1.260. So Disregard the graph that I posted earlier I am going to delete it.

I know that some drop is to be expected when going from idle to load I was more wondering if a .05 drop is bad.

Aaron: Does your voltage dispaly differently when you are in the bios and when you read it in your OS because mine reads 1.33 in bios and 1.296 at idle.

OCCT uses Linpak similar to linx

What is the difference between LLC and Multi-Step Load-Line (or are they the same just different titles.
 
Do you have any of the C1 C3/C6 or EIST powersaving on? If so it will definately be lower at idle in the OS.

The bios voltage I see is the default voltage for whatever multiplier I have set, plus ( or minus ) dvid. It drops to around 1.0 at idle due to my powersaving features I have on in the bios ( c1 and eist )

Again, since your board lets you set the vcore manually I dont know how that interacts with the auto voltage settings that are set by the multiplier.

My board doesnt have an option for multi step line load.... if I were to take a random guess at it, id imagine it fine tunes the amount of LLC applied depending on the load? I dunno, you'll have to ask someone with that specific board.
 
All of the c states are set to auto. Occasionally it will drop to 1.6ghz then back up to 4.29ghz for a second or 2 I am guessing this is because it is set to auto and not enabled. I am going to go back into the bios and enable some c states and see what that does. I just read in the manual that Multi-step load-line Has 10 levels of LLC so it is basically LLC
 
Last edited:
Bout the only thing I leave on auto are the extended ram settings.... there is so little documentation on what auto does, I find it better to set it to something so you at least know where you are.
 
Dont forget to set all your power settings in windows (assuming win 7) to maximum to eliminate the OS screwing with stuff.
 
I checked all of the Power settings and all are set to max performance. I enabled EIST but it still only idles at about 4.1 with now voltage drop. Well I have been messing with the voltages and trying to get the chip stable. I have reached 44x 1.36 vcore and I can run prime 95 linx and OCCT but when I run Intel Burn test on maximum it freezes after a couple of seconds with a 0x124 BSOD. But If I run the test on anything under 6000 MB in the custom test It is stable for 5 minutes (the most I have tested) I find this weird. I tried down clocking my memory to 1600 1.505 volts and this did not help. So I have it set at 43 1.32 for now until I can figure this out.
 
Last edited:
Well I woke up this morning and was back at it trying to get it stable. I am still going to stress test some more but I ran intel burn test for 20 minutes, linX for 20 minutes and am going to run Prim95 for about 6 or 7 hours once I see If I can drop the Vcore. I was getting the 0x124 BSOD and just kept raising the Vcore but it was not helping So i lowered it a bit and raised the QPI/VTT and I got Intel Burn Test's "Maximum" to run stable. I wonder why it would only fail If I used the Max amount of RAM? Im going to try for 45 and then stop a bit

Turbo x44
vcore: 1.385
QPI/VTT: 1.060
Multi-steps Load-Line: Disabled
Load Vcore: Prim95 the voltage drops to 1.332 (-0.048) and in Intel Burn test it stays at about 1.308 (-.072)
The voltage still drops a bit. I am guessing that this is a normal drop but I dont know I am still going to leave Load Line calibration off for right now.
 
Last edited:
I wonder why it would only fail If I used the Max amount of RAM?
Click to expand...

I'm not 100% positive on this, as i've only been playing with the sandy bridge processors for 3 weeks now myself....but, I would guess it fails when you use the max amount of ram because it is beating up the memory controller on the cpu....more bandwidth as compared to smaller amounts of ram? Dont quote me on that, but so far that is my understanding of what I think is going on with these chips.

As far as getting your OC squared away, i'd probably keep the ram clocked at 1333 (make sure you tighten up the timings and lower the voltage as required by your ram at that speed) ~ everything I have read so far says that ram speeds over 1333 dont do a hell of alot for these type chips. Being able to run 1.5 vdimm instead of 1.6+ as required by the higher speeds may even help to keep core temps down.

If someone wants to jump in and STFU NOOB me please feel free :)

All chips are different, me personally i wouldnt be afraid to push it to 1.4v if you are going for max overclock, assuming temperatures are under control (under 80c at 100% cpu load)
 
The Only time I see temps above 60 are when running Intel Burn Test Every other Stress test I see max temps in the upper 40's mid 50's with a max of about 57. That makes since that the "problem" relies on the memory controller in the chip.. If I am not doing anything in the background I can run the max amount of memory allowed but If i have firefox and alot of programs then I crash. Anyone know any values that I can increase decrease to get this more stable.
I think I may downclock my RAM to 1333 and see how low I can get the volts Where would I find the suggested timings for my ram Right now I have them at stock and all values at auto except the ones below (these are stock I just set them here to make sure.


Stock is:
XMP: Profile 1
RAM Freq.:1866
Ram Timing: 9,10,9,28, 2T
Ram Volt: 1.5
 
Last edited:
If your board is anything like mine, I HAVE to use LLC to get a stable oc beyond about 4.4 or 4.5. Well, I guess I wouldnt have to if I was willing to use a .100 dvid offset, but that is going to raise the idle vcore more than i'd like it.

Don't be afraid of using some LLC, just start of slow and see what it gives ya. When you get a huge drop in voltage when the proc ramps up to 100% its going to cause stability problems. LLC is there to prevent that.
 
If you are having memory related issues (and sounds like you may be) you can try raising the voltage for VCCIO and VCCSA and see if that helps. I believe VCCIO is the memory controller, but I've seen some conflicting info on which of those two settings actually matters more.

Also, some moderate value for LLC can help (edit: as Aaron_ATX mentioned), and you can also try enabling Internal PLL Overvoltage, although that is only normally needed above 4.6 or so.
 
Forceman said:
If you are having memory related issues (and sounds like you may be) you can try raising the voltage for VCCIO and VCCSA and see if that helps. I believe VCCIO is the memory controller, but I've seen some conflicting info on which of those two settings actually matters more.
Click to expand...
thanks I will try that out I keep getting the same codes 0x124 (QPI/VTT, Vcore +/-) and 0x101(Vcore +/-) I will google this but Do you Know off hand what the stock settings for these values are?

If your board is anything like mine, I HAVE to use LLC to get a stable oc beyond about 4.4 or 4.5. Well, I guess I wouldnt have to if I was willing to use a .100 dvid offset, but that is going to raise the idle vcore more than i'd like it.

Don't be afraid of using some LLC, just start of slow and see what it gives ya. When you get a huge drop in voltage when the proc ramps up to 100% its going to cause stability problems. LLC is there to prevent that.
Click to expand...
Ok..You have convinced me I am going to try level 1 first then try level 2 if 1 does nothing. If that does not work Im going to start upping VCCIO and/or VCCSA.

On another note I have enabled EIST and the all 3 c states disabled and I am still not getting a downthrottle at idle. Any one know what may be causing this I have Turbo Boost enabled and my multi set to x33 with the turbo cores all set to 44
 
Last edited:
I think you can find the stock settings under eazytune, if its not in your bios. In eazytune its under the tuner tab, then the advance tab on the right, and then on the left there will be a tab for voltages.

Pretty sure my bios shows me what the stock settings are tho as well.
 
I did some digging and I found that the VCCIO is QPI/VTT in my Bios (1.05 stock) and the VCCSA is the system agent voltage (.92) stock.

Thanks Aaron, In the past, I was not sure if those settings were the actual stock settings or if it was what the bios read them as because In my Bios It does not display the stock settings it displays real time values of the desired value.
 
Last edited:
On another note I have enabled EIST and the all 3 c states disabled and I am still not getting a downthrottle at idle
Click to expand...

I * believe * that EIST is the windows portion of mult/vid control. If I understand what Forceman mentioned in another post where I asked that question at least....and if you have the windows power settings all at 100% / performance im not sure EIST does much even if its enabled. Not 100% sure on that tho.

Honestly, I'd turn all that stuff off while you try and get your overclock set. It will be a little easier to track voltages. I didn't do that the first time around and I wrote down about 10 pages of notes with every little change I made...once you get the clock and voltages down you can turn them back on (I use C1 and EIST, no C3/C6 ~ I think C3/C6 did something crazy to my bios once when I put the machine to sleep ~ cant prove it tho )
 
Stanfiem said:
On another note I have enabled EIST and the all 3 c states disabled and I am still not getting a downthrottle at idle. Any one know what may be causing this I have Turbo Boost enabled and my multi set to x33 with the turbo cores all set to 44
Click to expand...

It's all tied up with Windows power management in ways I'm not certain of, but you'll need EIST and C1E enabled to get proper downclocking and downvolting. Having one or the other enabled can cause other behavior - but I'm not certain of the exact results. C3/C6 are deeper sleep states.
 
Thanks for all of the help guys. I wish I would have been overclocking sooner If I knew it was this fun. I am learning tons in the process. I will disable the c states and EIST for the time being. Then I will Enable C1E and EIST together. o and see if I can get my chip stable on lower volts. Or should I wait for this and Just keep going until I am stable at, lets say 4.7, then try and lower volts and other things.

Should I downclock my memory to 1333 or wait to see if I get a RAM related BSOD before i start messing with the stock speeds. Right now they are at stock 1.5v 1866 9 10 9 28 2t
 
If it were me I would set the ram at 1333. Im unsure of the effect higher speeds of ram has on the cpu memory controller. Tighten the timings up accordingly, (i'd imagine you could set them at T1 command rate also) just to reduce the possibility of the ram effecting your OC. Once you establish a CPU overclock then move the ram where you want it and see what effect that has.

I did exactly that, once I was mostly satisfied on my cpu overclock, I just turned my ram back to 1600, and I havent had any problems. Admittedly I have not 12 hour primed or any of the hardcore stability tests. I just play BF3 for a few hours, Batman, Skyrim, or whatever game, and havent had a crash yet.

But for leet overclocker status you are going to want to 12 hour prime it :)
 
SonDa5 said:
I have a different MB so not sure if my MB settings will help but it may. Here is a thread for my MB that I used to help me get a nice 4.6 GHZ Turbo Boost OC.

http://forums.overclockers.co.uk/showthread.php?t=18340310
Click to expand...

Thanks I will check that out

Aaron_ATX said:
If it were me I would set the ram at 1333. Im unsure of the effect higher speeds of ram has on the cpu memory controller. Tighten the timings up accordingly, (i'd imagine you could set them at T1 command rate also) just to reduce the possibility of the ram effecting your OC. Once you establish a CPU overclock then move the ram where you want it and see what effect that has.
Click to expand...
I will set the ram to 1333 and then run stress tests with it set to that. When you say tightn them up accordingly what do you mean exactly right now they are at 9,10,9,28,2t

I did exactly that, once I was mostly satisfied on my cpu overclock, I just turned my ram back to 1600, and I havent had any problems. Admittedly I have not 12 hour primed or any of the hardcore stability tests. I just play BF3 for a few hours, Batman, Skyrim, or whatever game, and havent had a crash yet.

But for leet overclocker status you are going to want to 12 hour prime it
Click to expand...

Once I get everything in the overclock set to where I want it I am going to stress using prime95 or intel or linx Im not sure for 12 to 24 hours. Just to make sure everything is ok temp wise.

I have everything stable at 44ghz when I run realtemp easytune6, AVG, Afterburner and IBT. If I have Multiple programs in the background and browse the internet/ use itunes the computer freezes and I get a 0x124 BSOD. If I have firefox already open then the test runs fine. I am thinking I should not worry about if I fail with resource heavy programs like Itunes and Firefox running because Prime95 runs fine with these programs running I am typing this while listening to Bassnectar Blast Off right now and running prime 95 right now. Is this a safe assumption? Or should I be anal and get the computer stable while running everything I can including prime 95 intel burn and everything else
 
I will set the ram to 1333 and then run stress tests with it set to that. When you say tightn them up accordingly what do you mean exactly right now they are at 9,10,9,28,2t
Click to expand...

Usually when you run ram at speeds slower than what they are rated for, the timings can be changed to lower latencys, aka tightened up, faster as in response time, not faster as in more bandwidth...same with the voltages, typically lowered a little. Run CPU-Z and under the SPD tab there should be several sets of ram timings that are detected for slower speeds. Actually you should just be able to set the ram to 1333 in the bios and basically let it auto detect the timings and it more than likely will get it close enough to correct. ~ Again not sure this is your problem, it just may help eliminate a variable.


have everything stable at 44ghz when I run realtemp easytune6, AVG, Afterburner and IBT. If I have Multiple programs in the background and browse the internet/ use itunes the computer freezes and I get a 0x124 BSOD.
Click to expand...

Do you have C1 and EIST on when it x124's or x101s? If so, i'd bet the clocks and voltages are dropping when the processor is under light load, and then when the multi bounces back up again, but the vdroop is just enough to cause a BSOD. You may want to add either a little more dvid, or really another level of LLC is probably better....assuming it is crashing as I just described.
 
It only crashes when I open up other programs while running the test. I am stable now When I started this thread I could not get intel stable at all. Now with a little bit of QPI/VTT increase I am stable in everything. I have been trying to get the computer to post at 45 I have gone all they way up to 1.412 and LLC level 3 but I can not get it to post am I missing something or Do you think I have hit a wall?
 
Ahh, ok yeah I missread what programs were running when you were bsoding

If you have an option for CPU PLL overvoltage, might be time to turn that on and give it some voltage, if you havent already.

It wont even post at x45? Hmm, you may be getting close to the end for the chip/mobo. One of the main beefs with my version of this board is that it only has as 4 pin power connector to the motherboard, where the better clocking boards have an 8 pin.....not sure if yours is the same way.
 
Mine has an 8pin connector. I got the ud4 because it was supposed to be a good overclocker It has 16 phase power but i guess in the real world that does not mean much. Who knows, maybe I am just bad at overclocking. Should I try flashing the bios to a more recent version I am running f1 and there is an f6. I am going to google my board and the new bios and see.

I can now get the computer to post but it will not boot up into the OS. It does not even leave the post screen where it reads the CPU clock and all the sata ports. When I normally start the comp. The multi is displayed first then it changes to the turbo (if I have this enabled.)
 
Last edited:
I enabled PLL overvoltage like you said and it booted right up into the OS I am having to push my voltages up a bit. I keep getting a 0x124 BSOD immediately while running prime 95.

I can get into the os on 1.9 QPI/VTT: 1.060 but no go on prime. I have been systematically upping my vcore and my QPI/VTT, and sytem agent voltage because on x44 I would get the x124 bsod and upping my QPI/VTT helped make it stable but before more vcore helped. Should I be increasing one at a time then if upping the vcore fails when i reach 1.500 or so i back it back down and start increasing QPI/VTT.

Multi x45
vcore: 1.425
QPI/VTT: 1.080
System agent voltage: .930
Intel CPU PLL overvoltage: enabled
Multi-steps Load-Line: Level 1
 
With my MB (MSI Z68 GD65 G3) I only need about 1.37v for 4.5GHz. I think you may have too much vcore voltage and insufficient cooling.

I always update my MB BIOS. If your MB BIOS has an updated version running I would install it.

Also might want to set your RAM to 1033mhz speed settings with 1.35v and 9-9-9-27 CAS timings to take off some of the load on the CPU IMC.

Not all memory will OC well with higher speeds and low CAS timings.

Once you figure out the max speed of your CPU then you can work on getting your memory back up to stock speed or over clock it if possible.

Seems like most 2500ks will do 4.8GHZ and on a good MB they will do up to 5GHZ and exceptional "golden" 2500ks will go 5.2GHZ +.
 
Normally at 4.5 you wouldn't need to mess with anything other than Vcore. VCCIO (might be QPI/VTT on your board) is for the IMC, so that could help with memory errors, but not sure what VCCSA benefits.

Try a higher setting for LLC - you have 1.4+ set, but with Vdroop you might be only getting 1.3 something under load.
 
Under Load I am getting about 1.334 minimum but hover around 1.394. However upping my LLC helped me get it stable.

The reason i have QPI/VTT (VCCIO) on is because I could not get IBT or Prim95 max ram tests to be stable with out raising this. I am going to try to lower it a bit and lower System Agent Voltage (VCCSA) and the vcore a bit and see how stable I am.

I am running prime95 blend right now and It has been stable for about 20 minutes so that is good. These are the settings I am at right now.

Multi: 45
Vcore 1.435
QPI/VTT:1.1
system agent: .925
PLL Overvolt: enabled
LLC: level 3
 
Well I am now stable in Prime95 for 20 minutes and intel for 3 passes at
x46
1.450vcore
1.1 QPI/VTT
LLC: level 4

My temps are at max 85 so I am not going to push the multi any further unless I can fine tune the voltages and get the temps down a bit. After i fine tune a bit I think I might try my hand at offset OC because it seems like the idle states work better if you use Offset voltages instead of Manual.
 
Stanfiem said:
No the intel max is 1.52 so i am good.
Click to expand...



For your speed I don't think you need that much voltage.


I'm running IBT stable with 46x multiplier on my 2500k with 1.336v.
CPUZ
http://valid.canardpc.com/show_oc.php?id=2211985

I know not all cpus are the same but I think you are pushing too much voltage into your cpu and it is increasing the temperature. I'd try lowering voltage. If you can't get your voltage down then you may have reached the limits of that CPU or the mother board is not very good at over clocking. In any case I wouldn't put that much voltage on the CPU if not necessary. IMO the lower the voltage the better.
 
Stanfiem said:
No the intel max is 1.52 so i am good.
Click to expand...

Yeah, I agree, 1.5 is a bit high, at full load, LLC on levl 1 and PLL off, full load for me is 1.332v @ 4.5 GHZ. Keep in mind my OC is meant for a 24/7 operations and temps top out at about 160F.

1g3omo.jpg
 
Stanfiem said:
Well I am now stable in Prime95 for 20 minutes and intel for 3 passes at
x46
1.450vcore
1.1 QPI/VTT
LLC: level 4

My temps are at max 85 so I am not going to push the multi any further unless I can fine tune the voltages and get the temps down a bit. After i fine tune a bit I think I might try my hand at offset OC because it seems like the idle states work better if you use Offset voltages instead of Manual.
Click to expand...

Voltages and temps seem kind of high to me for daily usage. It is possible that you have a lower clocking sample. I'd try backing down to 44x and start over. No Pll overvoltage enabled in BIOS and go back to "auto" on your VCCIO and VCCSA. Set your memory to your kits default timings and voltage...i.e. [email protected] or [email protected].

What motherboard are you running Gigabyte xxx-UD4? How many levels of LLC is offered in BIOS for your board? 7 or 8? I would set LLC to start with at the level that will give you the least overvoltage with the least vdrop... i.e. BIOS set @ 1.325v... Windows CPUZ idle @ 1.330v and Windows load @ 1.315v. You'll have to test the different levels to find which one does this the best.

You should be able to run a moderate overclock without having to jack up voltages a great deal. All chips are different your chip could have a sweet spot at around 43x 44x 45x, some are 44x 45x 46x others are 45x 46x 47x etc. I happened to get lucky (silicon-wise) and have a nice low voltage 2500K sample that can do 48x 49x 50x. Try to see what the max multi is for your sample with vcore in the 1.30/1.325/1.35v range. You can always try to push a little higher with a bit more vcore later depending on cooling and temps.
 
PolRoger90 said:
Voltages and temps seem kind of high to me for daily usage. It is possible that you have a lower clocking sample. I'd try backing down to 44x and start over. No Pll overvoltage enabled in BIOS and go back to "auto" on your VCCIO and VCCSA. Set your memory to your kits default timings and voltage...i.e. [email protected] or [email protected].
Click to expand...
These are all good suggestions but I am at all stock timings and volts. I will catch you up. Thanks for your input BTW. The X46 was me just pushing the chip to see how far it could go and I hit a temperature barrier x46 I was not planning on running my chip at those settings 24/7.
In order for me to get to x42- x44 I needed the following voltages to get stable with No LLC enabled

Constant Values

All Stock settings
RAM Freq.:1866
Ram Timing: 9,10,9,28, 2T
Ram Volt: 1.5 Bios reads 1.510
Chipset: Sandy Bridge Rev 09
Bios Version: F1
VCCSA (system agent): .92

x42
vcore: 1.310
LLC: disabled

x43
vcore: 1.305
LLC: disabled

x44
vcore: 1.380
LLC: disabled
QPI/VTT (VCCO): 1.060 (with out this I could not get the system past x43)

I then had to enable CPU PLL Overvoltage in order to Post at 45 and I had to enable LLC to 1 to get stable
x45
vcore: 1.485
LLC: level 1

When I set the LLC to level 4 I could get the chip stable at a lower voltage
x45
Vcore: 1.405
LLC: Level 4

x46
Vcore: 1.435
LLC: level 4


What motherboard are you running Gigabyte xxx-UD4 How many levels of LLC is offered in BIOS for your board? 7 or 8? I would set LLC to start with at the level that will give you the least overvoltage with the least vdrop... i.e. BIOS set @ 1.325v... Windows CPUZ idle @ 1.330v and Windows load @ 1.315v. You'll have to test the different levels to find which one does this the best.
Click to expand...
Ga-z68xp-ud4(rest of comp is in my sig). Are you saying that I should try to keep the vdrop to within .010v with every mulit at every voltage? Because I was getting keeping the chip stable with the highest vdrop possible bust still keeping volts below 1.52. When I started overclocking I was not using LLC so at 42, 43, 44, I did not need LLC so I did not use it. at 45 I used Level 4 But there are 10 levels

You should be able to run a moderate overclock without having to jack up voltages a great deal. All chips are different your chip could have a sweet spot at around 43x 44x 45x, some are 44x 45x 46x others are 45x 46x 47x etc. I happened to get lucky (silicon-wise) and have a nice low voltage 2500K sample that can do 48x 49x 50x. Try to see what the max multi is for your sample with vcore in the 1.30/1.325/1.35v range. You can always try to push a little higher with a bit more vcore later depending on cooling and temps.
Click to expand...
Like I said before The voltages I have listed are the minimum voltages I am able to use in order to be stable.I am now trying to get stable at x45 and see what kind of volts are needed using Dynamic Vcore for the low idle benefits. I am now stable at

turbo:43
EIST:enabled
C1E:enabled
DVID: +.010
QPI/VTT (VCCO): 1.055
Vcore: normal (1.345)
 
With Giga's 10 levels of LLC... I believe level 5 or 6 might be the closest to what I was describing while level 4 will offer a bit more droop under load. You can run with no LLC but it just means your chip will receive more droop under load and thus you'll need to set a higher vcore level in BIOS to compensate for vdroop/vdrop. What you are looking to find out when you overclock is the actual vcore load requirements your chip needs for a given multi...43x, 44x, 45x. under load in Windows via something like CPUZ or HWmonitor.

Your chip may have a somewhat weak imc... so while trying to find out your chips vcore requirements when overclcocked... I would back off on your ram speed and run either 1333Mhz or 1600Mhz dram speed. You should then be able to back off some on the vccio that you are needing to run 43x and higher. You can always come back and overclock your ram later. With SB Intel officially supports 1333 Mhz dram speed so running 1600Mhz and higher dram speeds is actually overclocking the chip's IMC.
 
You must log in or register to reply here.
Back
Top