Intel Haswell 4670k + 4770K Overclocking guide

[Mjolkspam]

x124 is typically related to ring bus instability

 

solutions

 

increase vccin

 

increase ring voltage

 

reduce ring multiplier

I havn't got the chance to test yet. I'll do it today probably. Will increase VCCIN to 2.0, maybe set ring multiplier to 34, so it doesn't run at 38 as Auto.

 

@KTM, i don't get BSOD because of BF4. I have other problems with that ^^  But good that you manage to fix it.

[Gene5is]

Like pretty much EVERYONE here said, thank you so much, your guide was exactly what i was looking for!.. its by far the best one ive seen this far!..

On to it: i was able to reach the following =>

1.33v vcore

1.33v cpu ring

1.95 VCCIN

@4.4Ghz on both core and ring bus...

Also memory @1866Mhz with 1.5v

First things first...

I did some IXTU tests for 10 hours and reach max temps of 78 degrees with custom WC.. Arent those temps too much for a custom wc?.. also im not new at wc and ive done quite a few builds but im mostly afraid that i have a bad chip..

The weirdest part came when after 10 hours of testing i stopped ixtu and let the pc cool down for a bit... then after an hour or something i started playing some csgo/wow etc and i got a random BSOD (WHEA_UNCORRECTABLE_ERROR)... i know this is hardware/oc related but after all the testing, do you think it could be OC related? Or is it maybe something else(also it didnt happened again YET)

PS: im typing from my phone right now sorry for typos and also sorry for the bad english!..

PS2: Ill post some screenshots once i get back at the computer.

And again, thank you for this awesome guide!.. Cheers from Asuncion-Paraguay

[ProKoN]

Like pretty much EVERYONE here said, thank you so much, your guide was exactly what i was looking for!.. its by far the best one ive seen this far!..

On to it: i was able to reach the following =>

1.33v vcore

1.33v cpu ring

1.95 VCCIN

@4.4Ghz on both core and ring bus...

Also memory @1866Mhz with 1.5v

First things first...

I did some IXTU tests for 10 hours and reach max temps of 78 degrees with custom WC.. Arent those temps too much for a custom wc?.. also im not new at wc and ive done quite a few builds but im mostly afraid that i have a bad chip..

The weirdest part came when after 10 hours of testing i stopped ixtu and let the pc cool down for a bit... then after an hour or something i started playing some csgo/wow etc and i got a random BSOD (WHEA_UNCORRECTABLE_ERROR)... i know this is hardware/oc related but after all the testing, do you think it could be OC related? Or is it maybe something else(also it didnt happened again YET)

PS: im typing from my phone right now sorry for typos and also sorry for the bad english!..

PS2: Ill post some screenshots once i get back at the computer.

And again, thank you for this awesome guide!.. Cheers from Asuncion-Paraguay

 

 

your temps are fine. Cooling Haswell past 1.3V is difficult

 

I would say your BSOD is most likely from  a unstable ring bus BSODX124

 

If it just happened once I wouldn't worry about it

 

if it happens again you may want to increase your ring voltage or drop the ring multiplier

 

nice OC

[Mjolkspam]

I have now tested increasing the VCCIN to atleast 2.0 to start with, i'm going slow up on everything.

 

So at 4.4Ghz with cpu ring on 3.4Ghz(manual) I used:  1.280V vcore, cpu ring at auto and vccin 2.0V.  1333Mhz with 1.5V.

As my last stress test i crashes just before the 8 hour mark, it's about the same time 15-20min before i reach to 8hours. And it's   0x00000124 (X124) I guess?  hal.dll is the driver that is crashing.

Atleast what it says in BlueScreenViewer.

 

I logged my temps this time. My higest peak was 71, average seem to be like 67-68. I'll go for 1.3V tonight and hope it wont crash 15min before the 8hour mark this time

 

CPU Ratio Mode, should it be Adaptive or Fixed?        CPU Core Voltage Mode,  Adaptive, Override or Auto?  I got an MSI motherboard.

[ProKoN]

ratio mode = Dynamic

 

core voltage mode = Adaptive

 

 

Watch the last video

[Mjolkspam]

ratio mode = Dynamic

 

core voltage mode = Adaptive

 

 

Watch the last video

Dynamic?  Adaptive for then i guess.    (*EDIT* I'm just stupid, i did read wrong...)

I was asking because i didn't realy understand.  That video in my eyes telles what i should use when i have the stable OC, not what i should use while overclocking?

I'm Swe and bad at english, so some things will get messy for me when i read or listen

 

**EDIT**  I'll do an edit on this one.   I have now manage to run it without an crash for 8hours 20min. 

 

vcore  1.29  (Adaptive)

vccin   2.1

@ 4.4Ghz    (Dynamic)  -  Ring Ratio @ auto(3.8Ghz)  -  Memory 1333Mhz  1.5V

https://hostr.co/file/rDFRoYpR9Yq2/4.4Ghz-1.29V---vccin-2.1V.jpg

 

Still at 72 on the temps. Looks fine.     So where should I move now?  Start lower my vcore as i increased the vccin to 2.1?  Or should i start to get the ring stable with this clock?

[KTM]

Hmm.. I still get BSOD sporadically while playing BF4 for a while. Not every time though.

Every time x124/hall.dll and ntoskrnl.exe

Passed all stress tests..

[Gene5is]

So, as promised, here is a screenshot from what i could achieve...

 

Any tips on how to lower a little my Vcore voltage would be MUCH apreciated!

 

Thanks again for your awesome guide and all the help you gave me!

 

Cheers!

 

 

PS: How can i post this picture in a bigger format? since the forum is not allowing me to post it on 1920x1080

 

EDIT: Yup, tinypic is way better than photobucket.. thanks!

[ProKoN]

Dynamic?  Adaptive for then i guess.    (*EDIT* I'm just stupid, i did read wrong...)

I was asking because i didn't realy understand.  That video in my eyes telles what i should use when i have the stable OC, not what i should use while overclocking?

I'm Swe and bad at english, so some things will get messy for me when i read or listen

 

**EDIT**  I'll do an edit on this one.   I have now manage to run it without an crash for 8hours 20min. 

 

vcore  1.29  (Adaptive)

vccin   2.1

@ 4.4Ghz    (Dynamic)  -  Ring Ratio @ auto(3.8Ghz)  -  Memory 1333Mhz  1.5V

https://hostr.co/file/rDFRoYpR9Yq2/4.4Ghz-1.29V---vccin-2.1V.jpg

 

Still at 72 on the temps. Looks fine.     So where should I move now?  Start lower my vcore as i increased the vccin to 2.1?  Or should i start to get the ring stable with this clock?

 

well you know your cpu is stable at 1.29-1.3V , just because you kicked up the vccin does not mean your going to get any drop in vcore sorry, try to lower the vcore if you want but i think your already there.

 

i would work on the ring bus

 

go ahead and work on your ring bus

Hmm.. I still get BSOD sporadically while playing BF4 for a while. Not every time though.

Every time x124/hall.dll and ntoskrnl.exe

Passed all stress tests..

 

BF4 has been running like a donkey for me lately.

 

x124 is most likely due to unstable ring bus.

 

try to drop it a multiplier and see if that helps.

 

no one likes a x124 bsod while gaming!

So, as promised, here is a screenshot from what i could achieve...

 

Any tips on how to lower a little my Vcore voltage would be MUCH apreciated!

 

Thanks again for your awesome guide and all the help you gave me!

 

Cheers!

 

 

PS: How can i post this picture in a bigger format? since the forum is not allowing me to post it on 1920x1080

 

 

sorry man, i wish there was someway to get the Vcore lower but thats life!

 

i have been using tinypic to upload my screenies, site works well i think

 

http://tinypic.com/

[Mjolkspam]

@ ProKon, just tested 1.28. Crashed, didn't think it would work but worth a try.

I'll start on the ring bus tonight.

 

**Big EDIT**  I did just try 1:1 ratio, 4400Mhz ring ratio with 1.29V, vccin 2.2V. It crashed a minute or two into the stress test. BUT, i didn't get an BSOD..

I got something i havn't got since i played Bad Company 2, everything just locked with a sound loop (first time i was watching Ring Bus video), second time it locked aswell, no sound loop tho

as i didn't have any sound going.  

 

Is that normal for an crash?  Shouldn't it be BSOD and restart?

[ProKoN]

@ ProKon, just tested 1.28. Crashed, didn't think it would work but worth a try.

I'll start on the ring bus tonight.

 

**Big EDIT**  I did just try 1:1 ratio, 4400Mhz ring ratio with 1.29V, vccin 2.2V. It crashed a minute or two into the stress test. BUT, i didn't get an BSOD..

I got something i havn't got since i played Bad Company 2, everything just locked with a sound loop (first time i was watching Ring Bus video), second time it locked aswell, no sound loop tho

as i didn't have any sound going.  

 

Is that normal for an crash?  Shouldn't it be BSOD and restart?

 

 

crash is a crash

 

crashes comes in many forms.

[Mjolkspam]

crash is a crash

 

crashes comes in many forms.

So true.

 

I lowered my ring ratio to 4200Mhz with 1.29V, vccin @ 2.0.  Crashed after 5 hours or so.  I got a BSOD 0x0000009c. I even got an Bug Check String (MACHINE_CHECK_EXCEPTION) this time.

I'll try with vccin back at 2.1V again at 4200Mhz. If that doesn't work, i'll lower it to 4.1 or 4.0.

 

Because as i understand it, i shouldn't go higher than 1.29V on the ring bus?  As I use 1.29V on the vcore.

[Harlin]

First post, so please be gentle...

 

I also just want to say thanks for the videos and information, much appreciated by noobs like me

 

 

So, I seem to have reached a (stable but uninspiring) ceiling with the below settings:

Memory 1333Mhz @ 1.5v

Vcore 1.3

Vrin External Override 1.9v

Multiplier x 42

 

Temps running IETU max out at around 72, so I appear to have more head room, yet, I  cannot get stable over x42. I followed the guide, but although I can make it to desktop, at x44, I get a BSOD (124 etc) shortly after running IETU. I'm a little stuck for my next steps. I'd be happy with 4.4, but seem to be stumped as to what to do next. Just to say I did as suggested load optimized defaults, so I've not changed any settings other than the vcore, memory speed/voltage, and VRIN.

 

Any tips would be most welcome

 

 

[ProKoN]

So true.

 

I lowered my ring ratio to 4200Mhz with 1.29V, vccin @ 2.0.  Crashed after 5 hours or so.  I got a BSOD 0x0000009c. I even got an Bug Check String (MACHINE_CHECK_EXCEPTION) this time.

I'll try with vccin back at 2.1V again at 4200Mhz. If that doesn't work, i'll lower it to 4.1 or 4.0.

 

Because as i understand it, i shouldn't go higher than 1.29V on the ring bus?  As I use 1.29V on the vcore.

 

It is ok if your ring Voltage exceeds Vcore voltage

 

 

First post, so please be gentle...

 

I also just want to say thanks for the videos and information, much appreciated by noobs like me

 

 

So, I seem to have reached a (stable but uninspiring) ceiling with the below settings:

Memory 1333Mhz @ 1.5v

Vcore 1.3

Vrin External Override 1.9v

Multiplier x 42

 

Temps running IETU max out at around 72, so I appear to have more head room, yet, I  cannot get stable over x42. I followed the guide, but although I can make it to desktop, at x44, I get a BSOD (124 etc) shortly after running IETU. I'm a little stuck for my next steps. I'd be happy with 4.4, but seem to be stumped as to what to do next. Just to say I did as suggested load optimized defaults, so I've not changed any settings other than the vcore, memory speed/voltage, and VRIN.

 

Any tips would be most welcome

your chip is running like a dog

 

couple suggestions

 

1- try to increase Vrin 2.1-2.2V

 

2-Drop uncore ratio to 30x (it may be turboing to 40x) or increase uncore voltage to 1.3V

[Mjolkspam]

I'm using spoilers to minimize the post    I edit this post instead of doing a new one.

I'll do a new one sooner or later, just felt like spam doing a new with only one post under me

 

Part 1:

Good.    I did try 1:1 ratio again, but without succes tho. Increased the vcore 1.32V, uncore to 1.32 aswell and vccin at 2.2V,  but nope. 1:1 ratio wont be achieved with 4.4Ghz cpu clock, probably gonna need higher voltage.

 

So i'm gonna start lower it 'again'. I'm born this way, i don't give up until i'm sure it does not work  

 

So I tested 4.4Ghz cpu, 1.29V, VCCIN 2.2V. "The Ring" was at 4300Mhz with 1.29V aswell. Crashed after about 1hour.  I'm going to run 4200Mhz again tonight with the same Voltage, 

but this time when i'm trying 4200Mhz ring bus, i'll have 2.2V on the vccin instead of 2.0V.

 

"Sorry if i talk much in the thread" ^^ Just want to tell where i am and what i'm going to do. I'm always checking this thread before I start the stress test when going to bed.

Edit 1:

I just ran into a BSOD on a 8hour stable OC   I was  4.4Ghz, vcore 1.29V, vccin 2.2V while  "The Ring" was at 3400Mhz with Auto voltage. Mem 1333Mhz, 1.5V.

Got the standard X124 bsod, so i guess my ring isn't stable?

 

When i got this last bsod i was streaming NHL game (St.Louis vs. Chicago) using Chrome, Mumble was in the background.  Nothing heavy running.

So at this point.. I have no clue what i should do!  Where should i continue?   My guess is that i should start over with the CPU stability, increasing voltage to 1.3V.

But that also feels wrong as i seem to have a unstable Ring?

Edit 2:

Again... About 5-10min after the Edit above!   I had increased the voltage to 1.3V  Everything else is the same.  I did the same things, X124 bsod yet again!

What should I do at this point?   I increased the vcore to 1.31V for now, I'll do a test with that tonight and see what happens.

Edit 3:

So I have stress tested it at:

 

4.4Ghz - vcore 1.32V

vccin @ 2.2V

 

Ring bus @ 3400Mhz - Auto voltage  |  (Ring bus @ Auto is 3800Mhz, so I changed it down)

Memory @ 1333Mhz - 1.5V

 

https://hostr.co/file/goK3TiuXYgIP/4.4Ghz-1.32V---vccin-2.2V.jpg

 

This was stable for 12+ hours (I stopped the test myself as i needed the computer).  My temps was at it's peak 77, which is good for my CPU cooler to be honest.

Tonight i'm thinking of trying everything at the same but i'm gonna lower the vcore down to 1.31V to see if i can get it stable. And continue even further if I can.

 

I don't think I can manage lower than 1.3V tho, as I have been getting X124 BSOD at 1.29V for some simple tasks. So weird it can handle 8+ hours stress test, but it can't stand some streaming + mumble etc =/

And the sad part is, it's not the cpu that is unstable, it's the ring bus..  Which i have at 3400Mhz.  I won't give up tho.

 

Some help of what i could do would not be wrong at this point tho.  // Mjolkspam

 

Edit 4:

 

Oh, i thought i wasn't gonna touch the uncore volt until i was starting with the ring bus  So i left that at auto and set the ratio for the ring at 34 (3400Mhz).

 

Anyways, I increased the "The Ring" this time. 4000Mhz with 1.32V (Everything same as the edit above). Before I cancelled the stress test,

it was active 10+ hours. Temps where the "same", 78 this time at it's peak.

 

https://hostr.co/file/QBS9a86BJp4M/4.4Ghz-vcore-1.32V-vccin-2.2V---ring-4000Mhz-1.32V.jpg

 

I didn't go for 4400Mhz on the ring as I actually tried that before. Crashed after 1 hour or so if i'm not wrong.  I'm going to try it again, but it's a long shot for me at these voltages I think.

It's either that or 4200Mhz tonight. 

 

I even think i'm gonna try to lower the vcore to 1.3 or 1.31.  I havn't tried 1.31V i think, have been much to do a few days now. So i can't remember if I missed that.

But what the heck, I got good temps at 1.32V, maybe I should just stay there

 

One thing tho, If i'm not able to get 1:1 ratio at 4.4Ghz with 1.32V, should I be happy with 4000-4200Mhz on the ring? Or should I try increased the uncore voltage and see

if i can manage the temps to maybe get a stable 1:1 ratio? What do you think ProKon?

[Harlin]

It is ok if your ring Voltage exceeds Vcore voltage

 

your chip is running like a dog

 

couple suggestions

 

1- try to increase Vrin 2.1-2.2V

 

2-Drop uncore ratio to 30x (it may be turboing to 40x) or increase uncore voltage to 1.3V

 

Thanks for the reply.

I might try the Vrin today, but I've also put in an RMA, as there seems to be quite a lot of 'poor' Haswell performers out there (over and above the Haswell limitations) 

Batch code is L331C352 bought from Overclockers in the UK, and they are normally quite good about these things

 

I'll try the Vrin and let you know. Thanks for the help.

[Harlin]

Righto.

 

So managed to get 15m IETU, so at least 'bootable' in W7 @ 4.3hz

Temps were 76c max during the test, so I'm going to give it a more rigorous thrashing tonight. If I manage to get it stable, what would be the next steps to make it to 4.4hz?

VCCIN is 1.9, and I have LLC set to High.

 

 

I'm planning on reseating the CPU tonight, with some improved TIM.

 

 

Many thanks for your help so far!

 

Current Settings below

[underclock]

Hi there.

 

First off, thanks for the great info !

 

Now to this "work in  progress".........

I deceided to go for 4.6 GHz overclock

 

I got to:

4.6GHz at 1.370v

PIV =2.00V

Processor Cache Ratio = 40 (Will be dialing it in a bit later)

Ram is not running on XMS , just set volts to 1.5 and the clock to 2000

Some fine tuning still to be done.

 

The overclock passed the CineBench , So I ran a quick 1h Intel Extreme Tuning.

Suprisingly the temperatures stayed under 78°C , the liquid cooling seemed to do well.

Then I played a game for 5 hours without any problems.

 

I will be running a 8h stress test later tonight.

 

The print-screen should have all the relative information.

 

 

 

[aurello]

Hey guys, need your advice.

 

this is my setting, 4,5Ghz @ 1,375 Adaptive 0.025, RingBus 4300 @ 1,375 Adaptive 0,035, VCCIN 1,9

 

and that's my stable setting, i can't get trough to 4,6Ghz without BSOD on XTU stress test.

 

any advice?

 

[ProKoN]

Hey guys, need your advice.

 

this is my setting, 4,5Ghz @ 1,375 Adaptive 0.025, RingBus 4300 @ 1,375 Adaptive 0,035, VCCIN 1,9

 

and that's my stable setting, i can't get trough to 4,6Ghz without BSOD on XTU stress test.

 

any advice?

 

 

 

 

Hello Friend welcome to the forum

 

Your temps are way to HOT .

 

Your CPU is actually throttling under load

 

First step is to work on getting the CPU stable with acceptable temps. 85c or under is acceptable. 100c is not

 

get a better cooler or drop your voltage and settle for 4.3-4.4GHz

[ProKoN]

 

So i'm gonna start lower it 'again'. I'm born this way, i don't give up until i'm sure it does not work  

 

 

 

keep updating your post because I like to read it .

 

you should really include the uncore voltage in all of your edits. "auto"  uncore voltage is not good

 

x124 while watchin hockey eh, its happened to me a few times. I think  the flash player  can be a little bugger sometimes. I like to run a hd youtube video while i stress test.

 

1.31V on the core is fine. keep uncore voltage at 1.3ish as well

[aurello]

Hello Friend welcome to the forum

 

Your temps are way to HOT .

 

Your CPU is actually throttling under load

 

First step is to work on getting the CPU stable with acceptable temps. 85c or under is acceptable. 100c is not

 

get a better cooler or drop your voltage and settle for 4.3-4.4GHz

 

Yes temps are hot.

 

So how you make better cpu throttling?

 

i can get cpu stable on 4.3 @1.26v and max temps 80c

while 4.4 @1.28v and max temps 85c-90c

 

my challenge is to get stable 4.5 and up, while maintain the max temps below 90.

at 4.5Ghz, my average temps are 50 while playing games or working, never jump above 80.  that's only on stress test.

 

Am I get bad chip? as i have to keep it 1.4v for 4.5Ghz, its high voltage jump from 4.4Ghz@1.28v

and i tempted to delid my chip.

 

Btw, i'm using NH-D14 for my cooler, and got enough air on my casing.

[morbid_bean]

Ok  So I Decided I would try to aim for a Tight Stable 4.0Ghz.  After a good week and half of stress testing and following the first 3 videos I managed to achieve the Below for a good 11 Hours

 

Multiplier: 40

Cache multiplier: 39

Vcore Voltage: 1.148

Cache Vcore Voltage: 1.138

VCCIN: 1.910

Ram - 1333mhz

 

Temps never went above 69 on load.

 

I then wanted to go in for the kill and try to see if It could handle a full 24 Hour Stress test..  Based on from what I could piece together, it lasted at least 19-20 hours before It got a BSOD.

 

Can It be safe to say that I am "Stable" because I ran it past the 8 hour minimum. ?

If I wanted to aim for a stable 24 Hour Test, what value would be the first to to start pumping up? Eg. Voltage? Cache Voltage? VCCIN? 

[Jerakl]

I'll need to use this soon, once I get my NH-D15.

 

Also, @ProKoN , I live in Sherwood Park

[ProKoN]

Ok  So I Decided I would try to aim for a Tight Stable 4.0Ghz.  After a good week and half of stress testing and following the first 3 videos I managed to achieve the Below for a good 11 Hours

 

Multiplier: 40

Cache multiplier: 39

Vcore Voltage: 1.148

Cache Vcore Voltage: 1.138

VCCIN: 1.910

Ram - 1333mhz

 

Temps never went above 69 on load.

 

I then wanted to go in for the kill and try to see if It could handle a full 24 Hour Stress test..  Based on from what I could piece together, it lasted at least 19-20 hours before It got a BSOD.

 

Can It be safe to say that I am "Stable" because I ran it past the 8 hour minimum. ?

If I wanted to aim for a stable 24 Hour Test, what value would be the first to to start pumping up? Eg. Voltage? Cache Voltage? VCCIN? 

 

 

I would say your fairly stable overall. Chances of you crashing while gaming, editing or doing normal tasks is fairly slim.

 

I would just run it as is. if you crash at all during a normal application give your chip a slight bump in vcore first,  .05 - .1V

 

if it crashes a second time do the same for the ring voltage.

 

vccin should be fine.

 

unless you run all tests at 24 hours and isolate the variables individually its difficult to say what exactly needs to be tweaked

 

I like to give er 18-24 hours my self also If it goes 18 hours I know its fairly solid. Xtu is not the most hardcore stress test but its fairly representative of the type of stress modern applications apply.

 

Nice LoVo overclock. most stock turbo voltage is 1.150V (on the i7, i5 is a bit lower).  you got a lower than stock voltage and all cores at 40x. well done. are you on stock cooler?