Any alternative to load line calibration

[an annoying pirate]

Hi so i am struggling to get my i5 6600k to 4.4 ghz. Idk if my cpu a bad overclocker but 4.1ghz run stable at 1.28 but 4.2ghz only goes stable at 1.34vcore Its like if i completly failed the silicon lottery

And my ♥♥♥♥♥♥ msi z170a gaming m5 motherboard doesnt even have llc WHEN IT WAS ADVERTISED AS A GOOD OVERCLOCK BOARD
Like ffs the cheaper m3 or the m7 have LLC, ♥♥♥♥♥♥♥♥ rip off

(actualy it does have llc but its only a yes/no option that is completly useless)

Anyway is there any other setting i can use to reduce vdroop and have less voltage during idle

Cooling not a issue though i want to stay lower than 1.38 on the vcore during idle

EDIT : jesus christ i started a flame war

[tacoshy]

Sphere の投稿を引用：

microcracking research papers to verify?

Microcracking occurs to different materials expending due to heat at different speeds. That's also what causes the normal CPU degeneration and one reason Intel doesn't Solder their CPU's anymore.

With changing voltage and clocks temperature change constantly faster then with fixed settings causing more microfractions. It's a normal process.

[Washell]

Sphere の投稿を引用：

Silicon doesn't obey the old P = VI Semi-conductors have a completely separate behaviour in response to voltage and amperes. Voltage is NOT linear in response to heat when talking about CPUs.

I know the formula is a little more complicated and involves clockspeed too. I also know that voltage is still a part of the formula and a lower voltage means a lower end result.

[Kaihekoa]

an annoying pirate の投稿を引用：

Thing is the llc in that mobo is only a yes/no toggle and both doesnt really reduce vdroops. While the cheaper m3 at least as some llc mode and my issue here is huge voltage drop during load causing instability

And how exactly do you know the LLC option isn't reducing vdroop? You don't have an overclocking board, you have a gaming board, so just do the best you can with your hardware. Next time you upgrade, get a good overclocking motherboard instead. Asus Maximus Hero or Apex are always good for OCers.

[_I_]

intel z board is an overclocking board
if its stable dont worry about it

but 1.28v seems high for 6600k at 4.4ghz
my 4670k is 100% stable at 4.4ghz and 1.27v

if you disable the c states in bios it should not underclock or go to idle states and keep the cpu at the multi and voltage you set

basic oc rules
stable and temps high, lower core voltage
stable and temps low, raise cpu multi
unstable and temps low, raise vcore
unstable and temps high, go back to last stable and stop

[tacoshy]

_I_ の投稿を引用：

intel z board is an overclocking board
if its stable dont worry about it

just because it is a Z-Motherboard doesnt make it an overclocking motherboard. The Z-CHipset only unlocks more options but so do some of the H-Motherboards. To be an OC-MoBo you need more things then just unlocked settings. Befor I now going to lsit everything I just goign to mention Mosfet Phases.
The Gaming M5 indeed is not an OC Motherboard because some of the settings are hidden and the support for OCing despite unlocked settings are limited incl. multiple levels of LLC.

_I_ の投稿を引用：

but 1.28v seems high for 6600k at 4.4ghz
my 4670k is 100% stable at 4.4ghz and 1.27v

If you read a single guide you know the term silicion lottery and now why this comparission is somehow stupid besides the obvios rookie mistake of comparing voltages and clockrates of 2 different architectures.

_I_ の投稿を引用：

basic oc rules
stable and temps high, lower core voltage
stable and temps low, raise cpu multi
unstable and temps low, raise vcore
unstable and temps high, go back to last stable and stop

just no! terrible advice tho I understand your way of thinking here but thats not remotly how you OC rightly.

1st: you lower voltage always as low as possible to reduce temps which you might mean. Otehrwise you dont lower core voltage just to get lwoer temps as it also risk instability. But I come back to that later.

2nd: CPU Multiplier nearly has no influence on temps compared to voltage. also to raise CPU Multipleir you also have to raise voltage which you do first but lets come back later to this.

3rd: I can agree with that one

4th: well that the way of giving up. But changing voltage of course you need but you also can start to counter instability with other settings like AVX Offset, Uncore...



How to really OC fast like ppl that do that on a regularily basis and bin CPU's do:

1st: Raise Voltage and test with "Prime 95 v26.6 FTT 1344K" for at least 20 minutes
->if temp > 80-85C then lower voltage
->if temp < 80-85C then rais voltage
->if temp = 80-85C continue with step 2

2nd: raise CPU Clock Multiplier
->if passing "Prime 95 v26.6 FTT 1344K" then raise multiplier
->if failing "Prime 95 v26.6 FTT 1344K" then lowering multiplier
->if found highest stable multiplier then advance to step 3

3rd: raise AVX offset by 1 and raise CPU clock multiplier by 1
->if passing "Prime 95 v26.6 FTT 1344K" then raise CPU clock multiplier by 1
->if failing "Prime 95 v26.6 FTT 1344K" then raise AVX offset by 1
->if found the highest stable CPU multiplier with an AVX offset as low as possible continue to step 4

4th: lower voltage
->if passing lower further
->if failing raise it back to last settings and your done

[an annoying pirate]

_I_ の投稿を引用：

intel z board is an overclocking board
if its stable dont worry about it

but 1.28v seems high for 6600k at 4.4ghz
my 4670k is 100% stable at 4.4ghz and 1.27v

1: not the same architecture
2: 1.28V for me is 4.1ghz not 4.4

[TehSpoopyKitteh]

tacoshy の投稿を引用：

Sphere の投稿を引用：

microcracking research papers to verify?

Microcracking occurs to different materials expending due to heat at different speeds. That's also what causes the normal CPU degeneration and one reason Intel doesn't Solder their CPU's anymore.

With changing voltage and clocks temperature change constantly faster then with fixed settings causing more microfractions. It's a normal process.

Microfracturing is not at all normal. If you are microgracturing it usually means you've gone beyond the TDP of the CPU.

If expansion occurs, you've also reached beyond TDP.

[tacoshy]

not starting again what TDP means and that there is no TDP limit... Just try to find out why CPU's degenerate over time and getting weaker and slow.

[an annoying pirate]

Sphere の投稿を引用：

Like I said not a single research paper shown. Please leave the engineering to the actual engineers. Also Prime 95 yeah no you shouldn't use it at all.

why? prime95 is fine

[rezo]

I think you just got unlucky with the silicon lottery. I imagine newer chips overclock much better than older ones, and my i5 4690k runs at 4.7 GHz at 1.37v with a AIO cooler and gets 700 on cinebench and 106% on passmark, which is a significant improvement from the 500/84% it gets stock.

Unlucky

[Washell]

The Spoopy Kitteh の投稿を引用：

Microfracturing is not at all normal.

It is.

Summary
The majority of failures in electronics are caused by thermo-mechanical loads and solder fatigue
is the major failure mechanism. The CTE mismatch between the board, component and attach
materials creates stresses in the solder and the plating material. Experimental data for solder
fatigue predictions and basic models can be used to predict solder fatigue for surface mount
components. Board designers can change component placement and board laminate material to
alleviate fatigue since component level design changes are usually not an option. The board
laminate design also affects PTH reliability. The board designer influences PTH reliability by
modifying drill diameters, laminate material, and plating parameters. Solder and PTH fatigue are
just two of the many effects of thermo-mechanical loads but they can be predicted and
prevented.

https://cdn2.hubspot.net/hubfs/1871852/DfR_Solutions_Website/Resources-Archived/White-Papers/Reliability/Temperature-Cycling-and-Fatigue-in-Electronics1.pdf?t=1489786101638

They work around it for critical components used for industries like aviation, but consumer parts are just designed to last their usable life. No point in designing a consumer CPU to last 20 years and quadrupling the price.

[tacoshy]

Sphere の投稿を引用：

Honestly I run a heavy F@H to test an OC. A CPU can draw far more power than what a motherboard can actually handle on Prime95. So unless you want to turn your cpu into a heating element don't use it. F@H runs really complex solving algorithms for protein folds which utilize the OpenMP library and it can make usage out of every cpu instruction including AVX ( Advanced Vector Extendions).

SO you have no idea what Prime 95 is for and why every hardcore OC'ler in the world use it to test stability.
Prime 95 is exactly designed for that emulating the very worst case scenario. SO that you know that your PC can handle that test because then you also have the garuntee that it can handle everything else you throw at it. Also the heat is a ncie thing as you know you stay below 90C in that test then you also know that your PC will never get close to that heat and that you never have to worry about overheating issues.

And a CPU cant pull more power then a Motherboard canhandle. If you stake your an engineer then I stake you should also know why.

[tacoshy]

Sphere の投稿を引用：

Synthetic benchmarks really tell you nothing about how a system will react given different workloads. Tacoshy there is so many variables involved nothing is even remotely that simple. This is where you are blatantly wrong thermal stress is not the only thing that can kill a cpu when OC'd not even close.

http://www.lavalys.com/community/blog/2010/04/top-7-myths-about-overclocking/

Right there in the list (heat is not the only worry)

Linustechtips explicitly warned against the use of Prime95.

ANd Prime 95 is not a Benchmark as you not getting a comparisson rating or any other value to check and compare with also it has different FTT's for etsting different things.

your link has shown nothing which we havent already said or not know. Also as we said, there multiple limits you should not cross but one limit that will happen is the ehat barrier. Of course you can try to to input the limit of 2V which not going to be applied unless disableing safety features and going for LN2 unlocked limits.

Also as we stated befor OCing is simple it doesnt rquire much as multiple did it here easily. But thats basic and easy OCing. If you hardcore OC and not just change multiplier a Vcore but start to fien tune then it going to be hard or if you try to break records. For example we have Monk here sitting in the Hall of Fame for a long time. I do really high OC's for air overclocking but I also !!break my CPU in term of longlivity!! as I know I can go beyond what other ppl should do and would reach under normal circumstances since I know that my CPU gets thrown out in 6-9 month anyway.
Also you have international OC Champions like Der8auer and 8Pack for which the first breaks record on a daily basis which states that OCing is easy and using Prime 95.

So Linux Forum or OC Champions (Btw Der8auer is an hardware engineer)... which is the more reliable source?

[tacoshy]

OCing pointless? wow... so getting 20-30% more eprformance and therefor also worklaods as renderign for example getting faster aswell as direct influence on fps is worthless... just wow.

since 80' or 90's IPC only having 2-5% improvement? Tellt hat to Ryzen comapred to FX or other generations... besides that since a user cant change the IPC the can change the clockrate. Why is clockrate imnportant? Look at what IPC stands for.

[tacoshy]

Sphere の投稿を引用：

As I said numerous websites annoying pirate just enable EIST.

I can OC higher with EIST disabled as other too. I want to see your numerous websites regarding to that topic but if you search with google the first websites all say to disable it:

http://www.dungeoner.com/en/overclocking-and-intels-power-management-settings-eist-c-states-turbo-boost/
http://www.overclock.net/forum/5-intel-cpus/1042491-enhanced-halt-c1e-off.html
http://www.overclockers.com/forums/showthread.php/634578-EIST-C-state-Thermal-Management

shall I continue? Seriosly you should learn how to really OC and maybe look into dedictaed forum or ask ppl that do that more or less on a daily basis.