Voltage for 4670k overclocking

[Mysticode]

I'm trying to hit a modest 4.3ghz overclock on my 4670k and Asus Maximus VI Hero motherboard, however I keep getting error 124 Bsod in under 30 minutes of running prime95 small test. This is an undervoltage error as far as I know.

I got things stable at 1.25v with 2 hours or so of prime95, however official ROG asus guides are saying this should work for 4.5ghz fine, let alone 4.3ghz....

Is there any sure fire way to test i don't have a completely shitty chip?

Any help is appreciated.

Thanks

[Ryan Leech]

Don't go above 1.38v, no matter the situation.

[cjeffcoatjr]

Dont use Prime 95 with Haswell.

Don't go above 1.38v

That to

[Mysticode]

Why no Prime95? I read it's fine for basic stress testing of even Haswell.

[Hecknologist]

apparently it pushes them too hard, most people say aida64 is better.

when it comes to voltages with 115X socket I never go past 1.3v for everyday use.

[Mysticode]

I'm not aiming for over 1.3v. Just want to get some advice on what to set, not what not not to set.

[Sernyhough]

I'm not aiming for over 1.3v. Just want to get some advice on what to set, not what not not to set.

Prime 95 uses more voltage than your board allows. creating higher temps.

[Mysticode]

Following a quick and dirty guide from Linus, I'm trying 47x.
1.2v would not boot, no bsod just hard freeze.
1.3v booted with bsod error 124.
1.31v booted with bsod error 124.
1.32v the same.
1.33v the same.
1.34v the same.
1.35v the same.


Abandoning 47x, going to 43x at 1.25v:

43x:

1.25v boots fine. Aida64 "System Stability Test" for Stress CPU currently running...

[Applepienation]

Looks like you may have gotten a lemon. I'd be a bit surprised if it overclocked that badly though.

[Mysticode]

Well the first thing I did was load BIOS defaults, just in case I had something set wrong somewhere...

 

Aida64 is running right now, but I am only seeing average 60c in temperatures on my cores. With Prime95 small running, I was getting closer to 80c  with this same 43x @ 1.25v. Is this thing really stressing my CPU? Should I try to crank up to 45x and run Aida64 again? I have only been testing with Prime95, and people are saying that it send too many voltage usage to the CPU causing abnormal instability in testing...

[Kuzma]

AIDA64 > Prime95.

[airdeano]

I'm trying to hit a modest 4.3ghz overclock on my 4670k and Asus Maximus VI Hero motherboard, however I keep getting error 124 Bsod in under 30 minutes of running prime95 small test. This is an undervoltage error as far as I know.

I got things stable at 1.25v with 2 hours or so of prime95, however official ROG asus guides are saying this should work for 4.5ghz fine, let alone 4.3ghz....

Is there any sure fire way to test i don't have a completely shitty chip?

Any help is appreciated.

Thanks

 

 

TTL has a great way to overclock.

 

 

this way you are not all over the place with no strategy.

[Mysticode]

Watching this now, thanks. Was following Linus video because he was using an rog board and haswell, while Tom is using gigabyte and ivy bridge.

[airdeano]

Watching this now, thanks. Was following Linus video because he was using an rog board and haswell, while Tom is using gigabyte and ivy bridge.

 

start low and slow.. there is no fire to get to and you'll find out the proper steps to

get an overclock to stick and be stable. this method is the same on all platforms.

mobo/UEFI may have different labels, but same results. on haswell no need to

use any LLC (load line calibration).

 

 

go to 3:13:00 and what how he is using haswell overclocking.

[Mysticode]

Tested with OCCT Linpack, 64 Bits enabled, AVX Capable Linpax enabled, all cores:

 

43x @ 1.25v 30min test (78C max) : Pass.
(system saw it at 1.264v, LLC set to auto) 
43x @ 1.24v 10min test (84C max) : Fail. Stopping at 10 min due to high temps
(system saw it at 1.264v, LLC set to auto)

 

43x @ 1.24v 30min test (82C max) : Fail. BSOD error 101 at 2 min (increase vcore) 
(system saw it at 1.264v, LLC set max 8) 
43x @ 1.25v 10min test (85C max) : Fail. Stopping at 2 min due to high temps
(system saw it at 1.264v, LLC set max 8)

 

 

Tested with OCCT Linpack, 64 Bits enabled, AVX Capable Linpax DISABLED, all cores:

 

43x @ 1.25v 30min test (72C max) : Pass.
(system saw it at 1.264v, LLC set to 8) 

 

 

 

*ram set to 1600mhz for base benchmarks
*H100i in push

*I do not think LLC is doing much for Haswell in these tests, as my vcore is only going up under load, not down

[airdeano]

 

Tested with OCCT Linpack, 64 Bits enabled, AVX Capable Linpax enabled, all cores:

 

43x @ 1.25v 30min test (78C max) : Pass.

(system saw it at 1.264v, LLC set to auto) 

 

43x @ 1.24v 10min test (84C max) : Fail. Stopping at 10 min due to high temps

(system saw it at 1.264v, LLC set to auto)

 

43x @ 1.24v 30min test (82C max) : Fail. BSOD error 101 at 2 min (increase vcore) 

(system saw it at 1.264v, LLC set max 8) 

43x @ 1.25v 10min test (85C max) : Fail. Stopping at 2 min due to high temps

(system saw it at 1.264v, LLC set max 8)

 

Tested with OCCT Linpack, 64 Bits enabled, AVX Capable Linpax DISABLED, all cores:

 

43x @ 1.25v 30min test (72C max) : Pass.

(system saw it at 1.264v, LLC set to 8) 

 

 

 

*ram set to 1600mhz for base benchmarks

*H100i in push

*I do not think LLC is doing much for Haswell in these tests, as my vcore is only going up under load, not down

 

 

43@ 1.25 and grill it. so far seem to be the passing grade.

[Mysticode]

Grill it? Like, put AVX Capable Linpack back on and see if it BSOD or not? Should I be worried if it's going over 80c? Tom said 80c is a fair threshold for Ivybridge, so I assume Haswell is the same?

[airdeano]

85° is (should be OCCT default). once stable on AVX, your stable pretty much anywhere.

like TTL said, once in real-world, you won't be hitttin' those AVX temps, just for testing.

[Mysticode]

Yeah I havn't hit 85c so far, but I have hit 84c. Would you yourself feel ok sticking with 85c for 24/7 usage?

[airdeano]

you'll never be at those levels in real-world.

 

but yes.

[Mysticode]

Ah very true, good point. I do a lot of crypto mining however. Guess I should fire that up and see what temps I hit.

Edit: ran some gpu and cpu mining, temps on cpu hit 84c again but just now I got a bsod with error 101. Putting up voltage to 1.26v and turning off LCC (as earlier mention that it doesn't matter for haswell). Running OCCT with avx enabled.

Edit 2: 1.26v caused another bsod error 101 running OCCT with AVX enabled. Going to 1.265v in bios and seeing if OCCT will do minimum 30 minutes with avx. Two minutes in and OCCT has failed due to temps getting to 86c. I feel if I knock out of avx mode, it won't be an accurate test of the 24/7 load that I'm throwing at this PC as I am doing cpu based crypto mining.

Edit 3: Running at 1.265v, cpu miner is stressing the cpu at 100% load for about 15 min now, max temp is 75c which is pretty good, starting opening Chrome and a few tabs in, bsod error 101. Have to hard reset my PC. At this point I don't think it's a basic vcore voltage issue. Any one have any advanced steps I can try out here?

[Opsilon]

Ah very true, good point. I do a lot of crypto mining however. Guess I should fire that up and see what temps I hit.

Edit: ran some gpu and cpu mining, temps on cpu hit 84c again but just now I got a bsod with error 101. Putting up voltage to 1.26v and turning off LCC (as earlier mention that it doesn't matter for haswell). Running OCCT with avx enabled.

Edit 2: 1.26v caused another bsod error 101 running OCCT with AVX enabled. Going to 1.265v in bios and seeing if OCCT will do minimum 30 minutes with avx. Two minutes in and OCCT has failed due to temps getting to 86c. I feel if I knock out of avx mode, it won't be an accurate test of the 24/7 load that I'm throwing at this PC as I am doing cpu based crypto mining.

Edit 3: Running at 1.265v, cpu miner is stressing the cpu at 100% load for about 15 min now, max temp is 75c which is pretty good, starting opening Chrome and a few tabs in, bsod error 101. Have to hard reset my PC. At this point I don't think it's a basic vcore voltage issue. Any one have any advanced steps I can try out here?

Try this, input voltage to 1.2, test the CPU at 4.2 in aida 64, for at least 6 hours, this is gaming stable, anything over this is awesome, but 6 hours at least.

 

Ok i know people will jump and scream on me for this, but here goes:

1. Instal Asus AI suite III

2. 

watch this video and JJ will explain a lot, just be patient

3. when you do this it will show you the approximate max overclock on your CPU ( 1,275 is the max voltage )

4. after that try manually dial the values into the bios and test them for at least 6 hours ( manual voltage when testing pls )

[deathjester]

The official ROG guide is a joke.

 

1) they are using adaptive voltage. The voltage they punch in is a fairy tale.The only time you should be using adaptive is after you found max clock on override (static voltage). You shouldn't use adaptive to stress test either. To be honest? Adaptive is damn near worthless on Haswell.

 

2) Asus tells you that you need adaptive to have a Haswell downclock and down volt to save power. This is false. Dynamic does this and your windows power settings do this. 

 

3) The 1 to 1 stuff he utters about ratio is complete nonsense. It doesn't mater what the cache is set at. Yeah it is a SMALL increase for every 100mhz higher then 3500 you go, but many chips need cache set to 3500 to hit higher clocks. Cache means so little compared to clock speed on Haswell it isn't funny.

 

Many people get stuck at 4.2 on Haswell, just like many people get stuck at 4.5 on AMD. All these chips aren't clocking to 4.5 and 5ghz. AMD now sells their higher bins as 9xxx. A 4.2 is a 700 mhz OC on Haswell and 4.5 is a 500mhz oc on AMD. The 4.2 is BETTER. Both chips can be "bad" at overclocking. These are not Sandy Bridges.

 

Half the guides on Haswell OC'ind are a joke, and most of the reviews and "stats" are using ES (engineering sample chips) if you look at the CPU ID in the review. Retail Haswell's don't oc like ES chips, just like r9 290x retail cards didn't match highly binned ES samples given to reviewers. Both companies resorted to "cheating" to promote their product.

 

When people spread these fairy tales about all chips overclocking high, it makes people feel bad when they get stuck at 4.2 ghz or 4.5ghz or whatever. It happens. It happens a lot. You can raise VCCIN, lower cache to try to get 4.4 but it still is a no go on MANY chips.

 

http://www.tomshardware.com/forum/id-1722630/intel-god-quick-dirty-guide-4ghz-haswell.html

 

Do no use the above guide unless you are on water or a dual tower air cooler MINIMUM. VCCIN adds heat. Near 1.3 on air is simply too damn hot. You still might get to 4.4 but if you get stuck at 4.2? It happens. Has nothing to do with the motherboard. People get stuck at 4.2 on Rog boards all the time. MB means jack on z87 cus of on die VRM. Worst I have read or seen is 4.0. Never seen below 4.2 with water and up to about 1.3v, but 4.0 is probably legit on air.

 

Unless you are playing Guild Wars 2 or Flight simulator X? 4.2 on a Haswell is freakin fast. REALLY fast. It is still fast in those games as well and will blow away anything but high clocked Sandy, Ivy (and you could lose big on the silicon lottery on ivy also), a higher OC is just nicer, but in no way guaranteed. 

[Mysticode]

What settings in AIDA?

[Mysticode]

Went for 42x, memory at 1600mhz, all other settings auto, including vcore.

Running Prime95 small for five minutes now, one of my cores are consistently hitting 85c. Should I be concerned? Average auto vcore setting as reported by Aida64 cpuid is 1.275v.