i5 12600KF Prime95 P and E core Overclock on 240mm AIO with MSI Z690 torpedo EK X

[@chronoreverse]

My 12600k needs 1.230V for 5GHz/4GHz/43 all-core small FFTs P95. Heat is... incredible. After running it for a bit to confirm stability, I set TVB so that power viruses immediately kick the multiplier down while still allowing easy stuff like Cinebench the full frequency.

[@Ben-ld1qi]

That's a great idea for TVB, by how many multipliers do u set it?

[@chronoreverse]

@@Ben-ld1qi There are two temperature thresholds so I set -1 for 88c and -2 for 95c. It adds some oscillations in frequency but seems to do the trick.

[@Ben-ld1qi]

@@chronoreverse Thanks that's a really good idea, I'll implement that. The reason I asked is because I only need the CPU/RAM (DDR4 S8B) overclocked for gaming benefits in 1080p/240hz. I was using Linpack to test stability of the memory timings but Linpack ends up shooting cores to 97~100°c, so I'll probably set a decrease multipliers for when it reaches 80°c+. When gaming tho, CPU is in the 60's celsius overclocked on a budget air cooler, so I'm happy about that.

[@Schroinx]

My 12600K does 5 GHz undervolted with -0,1 v and air cooling - noctua but still. I'll check with prime to see if that makes more heat than Intels OC utility. Board is Asrock Z690M-ITXax.

[@franciscocornelio5776]

I thought it wasnt possible to oc with air cooler. What are your temps? I have also aircooling with 2 fans 12 heathpipes

[@AA-kx5es]

Great vid as always. Thank you

[@zayedsays4847]

Great video, I'm running : 12600K Asus Prime Z690M Plus D4 H100i V2 AIO Got a stable all core 5.2Ghz Pcore 4.1 Ecore at 1.27v - LLC 5. Max temps I've gotten in Cinebench 23 was 85c, gaming max 60c, idle 30c

[@VNguyen97]

can i have your profile OC

[@kunstderfugue]

My analogy for raising the voltage is that it's like increasing the pressure on a hose, more water is going to flow out necessarily

[@givemeajackson]

I read "horse" and was very confused

[@kunstderfugue]

@@givemeajackson unfortunately the results of raising a horse's pressure really fast are far harder to clean off the walls and ceiling

[@BBWahoo]

@@kunstderfugue and mouth 🥵

[@DrKrFfXx-0]

Does raising the CPU Switching Frequency have stability benefits?

[@otavioribeiro2872]

VIDEO ABOUT THIS PLZ

[@ActuallyHardcoreOverclocking]

depends on the motherboard.

[@NightFlash24]

I saw marginal, if any benefits on a low end z370 board, but a significant increase in VRM temps, so take care with it

[@davemichael1859]

Switching frequency affects power rail ripple at the expense of efficiency

[@johndough4871]

The 12600KF I have is also super impressive. Glad to see that I have a good fallback if my 14600K craps out on me.

[@denismolander]

why you didn't use higher LLC? For lower cpu voltage

[@t1xer45]

Thanks for the great video, I did as you do. Is there a chance that you would help me lower the timing frame? currently working on profile 2 ddr4 14-16-16-36 (Gear1) series have 4000mhz cl16 you can even lower them on timings and increase the mhz

[@SayeR88]

i5 247w power consumption 😳😳

[@-FOXX]

got 5.6g (2), and 4.1all at 1.485 adaptive, avx stable p95 all core 12600k, 8100ish r20

[@-FOXX]

mode 4 MSI, runs 5.2g in p95 (1.31v vcc sense) smallfft Anything else make sense to do? Can't go over 4.0ring no matter voltage/set ring

[@BBWahoo]

RIP CHIP

[@connordobbin-bh3nc]

Mine must be different then. Can't even do anything and I'm missing settings

[@eleftheriosgiotas3179]

Does this or the Tomahawk DDR4 variant have SMD dimm slots? For example Aorus pro has them SMD on both DDR4 and DDR5

[@philippkepplinger]

I have the Tomahawk DDR4 and it does not have SMD dimm slots.

[@ActuallyHardcoreOverclocking]

the Aorus Pro DDR4 uses through hole dimm slots. Only DDR5 boards have SMT slots.

[@eleftheriosgiotas3179]

right.. I mistook that

[@Daniel-zr4pk]

damn man, ty, i tried 50 multi on P-Cores with fixed voltage 1.210v and it worked like a charm. i guess my cpu is better bin because i run 4.1 on the E cores and 4.3 on the ring. my avx-2 down-ing is -1 multi. its stable at prime95 small ffts but i have my cpu temp limited at 90C so it throttles a bit. i also think i have a faulty temperature sensor cuz i repasted the 3 times and i always get 15C more on Core #3 than Core#2. my paste was evenly spread last time and it was still temperature bugged. my cooler is really shitty so im not trying higher voltage than that. i got noctua 45mm heatsink with 2 fans. very bad cooler for oc, idk what i was thinking.... i was wondering why are you playing with this fluctuating voltage? why not go with fixed value 24/7 like me. what would the difference be? i dont believe it will affect longevity cuz my 2500K is still very healthy and not degraded at all and its a 10y+ old cpu soo.. i'd be interested in hearing why you prefer floating voltage to fixed voltage.

[@Daniel-zr4pk]

also something else i noticed. my Wattage is much less than yours. mine Package wattage capped at 173W. and yours is more than 200 with lesser clock? and your vrm temps are double mine?

[@mash9262]

hello good sir! my temps are also like yours. my core 3 is always heating way more than the others even after i replaced the termal paste multiple times, and with different brands (Noctua NT-H1 and Thermal Grizzlie Kryonaut - same results)

[@Daniel-zr4pk]

@@mash9262 very weird, what could be the problem? I still havent solved it

[@franckdruart8120]

yo im a bit late but a cpu delid might fix specific core high temps, i dont think ur sensor are off

[@Daniel-zr4pk]

@@franckdruart8120 rofl why would i delid my cpu?

[@rileyjohnson5997]

Hey could I copy the bios settings you used and be fine with my i5?

[@ole7736]

On Gigglebyte you can disable all power limits with one switch as well.

[@razerface35]

What is that switch called?

[@ole7736]

@@razerface35 I think it is called Power Limits, in the Advanced CPU Configuration section.

[@Mxgtumgoten]

Why you don t have a mora420? Should be baseline today

[@sagerdood]

Z690 vrm is overkill because were going to be using raptor lake on the too....

[@stigi3849]

Buildzoid and I ARE best friends :)

[@rasoulkhandan9334]

Hi dear friend Do you use the Prime95 program only to test the system stress or to calculate the pi number, please guide me. Thanks

[@baumstamp5989]

which of the mobos allows non-K bclk OCing, is there an exhaustive list? i could only find der bauer video showing two super expensive boards. do i need to pay $200-300 for a board to be able to do BCLK ?? i am so torn with the 12000 generation as to what to buy because it really seems nobody cares anymore about systematically testing all possible stuff with all possible configs anymore. nor documenting it...

[@Mom19]

Just look up the board you want if it has an external clock generator. If it hasn't, no BCLK on non K.

[@PolskiJaszczomb]

@@Mom19 Other issue is, locked 12 gen CPUs also have locked System Agent control, so no actual RAM overclocking, as you're stuck with 0.95V on VSA.

[@Mom19]

@@PolskiJaszczomb that is true yes, but the question was about BCLK just for the CPU I suppose :O

[@PolskiJaszczomb]

@@Mom19 Well, the extra payment for that BCLK OC equals to getting an unlocked CPU, so there's no real reason to opt for BCLK overclocking, as external clock gens only exist on the most expensive boards. Intel level of trolling.

[@Mom19]

@@PolskiJaszczomb Yeah I agree, certain things are currently not worth it to switch to anyways. PCIe Gen 5 is cool, but no benefit well not much of a benefit from gen4 so far either, maybe with direct storage by then. DDR5 is also not really any improvement and certainly not worth the big extra so far. Not to mention current Kits will be probably garbage in a year or so. Im very happy with my 5800x for my VR system. Its all fun and cool, but it's not that worth switching to aslong as your current system can still keep up. Save your money :D

[@robinkonig5828]

does anybody know if you can BCLK oc locked rocket lake cpus like the 11400? i can not find anything online on this topic

[@ActuallyHardcoreOverclocking]

no you can't

[@manuelbardina]

Why is aio appreciably worse than custom? Is it the usage of aluminimum instead of copper, the limited flow from the comparatively anemic pump, is it just the smaller contact patch on the cpu block, or the less engineered fins/microfins/etc. ?

[@jdogdarkness]

I'd say it's a little bit of all that. (Varies by model of course) but primarily it's the amount of fluid that is in the loop. It provides more mass to take handle higher thermal load. Also with that, it takes longer to reach heat soak. (Max thermal load/density/transfer). Limiting factors are surface area(the fins and patch contact), thermal mass(the water and build materials), and the fans(which also tie into those other standards)

[@jdogdarkness]

Obviously I'm not an engineer but I do remember some science from school and the plethora of videos I've watched on the matter lol. Jayz2centz and GamersNexus have talked about it before if u wanted to find out more.

[@manuelbardina]

@@jdogdarkness they all seem like they'd contribute, but I'd be interested in seeing someone snip the tubes, use some barbs to convert components to G1/4, and mix and match with real open loop parts, to show which particular part bottlenecks the aio most vs open loop

[@jdogdarkness]

@@manuelbardina that could be interesting. There is a dude that like designs fans and does other things relating to PC thermals that would be the best bet. Gamers Nexus disasemles and analyzes constructions of AIOs down to the fins. But Jayztwocentz is prob the biggest custom loop guy on KZbin. know about Linus Tech Tips right? Just throwing out some things you might find interesting.

[@Daniel-zr4pk]

my 12600kf is at 1.210v 4.8 pcores 4.1 ecores 4.3 ring, ram 7.5ns 4000cl15 b-die. Prime95 small ffts perfectly stable and ycruncher 2.5b avx2 perfectly stable.

[@Daniel-zr4pk]

two days ago i succeeded with stable oc. it was unstable before that because of the ram.

[@Daniel-zr4pk]

ps.. hmmmm you got 5.0 ghz on 1.210v too, maybe i can try this later

[@PolskiJaszczomb]

Nice barebone RAM kit :d You should go and ask Arctic for a Liquid Freezer II 420 mm :v

[@NightFlash24]

Even the 280 or 360 is easily capable of heavy overclocking. I have a 240 on my 9700k running 1.33 vcore, Prime95 @ 190 watts and 75c core temps. Radiator is only mildly warm to the touch - with fans at 60%. Thick rads ftw. Of course i had to lap the IHS, because i ran into contact / heat transfer issues well before the limits of the puny little 240.

[@PolskiJaszczomb]

@@NightFlash24 I've got 1st revision LF II 360mm, cooled delided and sanded 5775c to 76C @165W in P95, 4.2GHz 1.368-1.405V (FIVR FTW...). Now I've got 12700k coming and I really hope its IHS is not a concave mess, I'd rather not have to sand it down, but oh well, not that it's gonna die if it didn't come dead x). I'd never opt for 240mm AiO over D15, no benefits and more hassle + potential pump failure. 280mm and bigger is the way.

[@joshuabennett723]

Can do 5.0 on the p and 4.2 on e cores on my 12600kf with a tower cooler pegged at 100c running cinebench though 😂

[@barishyeniden1478]

With all these settings you are just trying to reach 5.0GHz, instead of messing with all these settings just activate ''Enhanced Turbo,'' under load all cores will automatically switch to single core speed i.e. 4.9 GHz, on desktop or small tasks it will drop to minimum speeds etc.

I have the Edge WiFi DDR4 with 12700kf. Can I use your settings to hit 5.0ghz for daily? THanks!

[@Mom19]

Just try it out and then adjust from there

@@Mom19 I should mention I have the Liquid Freezer II 280 but my temps got crazy when I ran cinebench. I'm thinking it's probably the way it's mounted since I jerry-rigged the mounting for it. I scavenged some of COrsair's mounting bracket. lol. Anyway, Artic sent me the lga 1700 bracket yesterday. I'm hoping my temps will improve with it. :)

[@Mom19]

@ it could be very well mounting, yes. Maybe airflow too, make sure your pump isn't pulling air either :)

[@jackiekuang7603]

@ I was getting very high temps even with a LGA 1700 bracket, did the washer mod on my MSI Z690 Pro-A and dropped a further 5C. I also delidded my 12700K since even with the washer mod and 280mm AIO at max, I was smashing 97C. The delidding brought temps down further so now my max is 89C on cinebench for 1 hour. Hopefully the new mounting bracket works out for you as unfortunately for me I needed to do a lot of work to get mines tamed.

@@jackiekuang7603 oh god. I'm not prepared to delid. lol

[@christopherjackson2157]

As I understand it the relationship between frequency and power draw should be fairly linear. So 2.5 to 5.0 would double it but 5.0 to 5.5 would increase it around 10 percent? So its more or less a non-issue in most overclocks ? Someone correct me if I'm wrong

[@Torbjorn.Lindgren]

Only if you assume the same voltage and 2.5 GHz definitely shouldn't need as high voltage as 5.0 GHz - this is well into the territory where extra voltage should be expected to be needed. Power draw at a given frequency is quadratic with voltage!, so even if it only needs say a 20% increase in voltage for this jump that would mean the power draw went up with almost a factor of 3 instead of double (2.0 * 1.2^2 => 2.88x power draw). This is why be default processor/BIOS have a voltage curve mapping from frequency to required voltage. For desktop OC it's not uncommon to just go with a static voltage and static frequency (simpler and for a desktop usually fine), but it doesn't change that the voltage necessary for a processor to be stable goes up as the peak frequency goes up. The amount of extra voltage required tends to get steeper as you get near the processors frequency limit and 5+ GHz certainly qualifies for that at ambient so I would expect 5.0 -> 5.5 GHz to require quite a bit extra voltage so the power draw likely goes way up rather than merely 10%. But I don't have enough experience with Alder Lake overclocking to even guess how much extra voltage would be needed, other than suspecting the power usage is going to be some variant on "a lot more". Yes, the i9-12900KS is rumored to have a 5.5GHz max boost but the reality is that for all-core loads it's going to be heavily limited by the MTP, despite that increasing from 241W to 260W. I expect it'll boost a bit higher on single or few-core loads though. This is for ambient but the same basic principles applies for sub-ambient too (phase change, LN2 or LHe) but the cold reduces the voltages required for a given frequency to be stable (and the colder the more it shifts down), this helps a lot thanks to that quadratic factor.

[@christopherjackson2157]

@@Torbjorn.Lindgren awesome answer. Thanks for the assistance!

[@Bobzillaaaful]

37.8k max fan speed lol,gotta love hwinfo bugging out like that

[@kwentin6317]

pour une fréquence de 5Ghz la tension est beaucoup trop élevé en plus la valeur max de temperature est trop élevé.... elle sert à protéger son processeur

[@futmut1000]

Arctic 420mm aio is so cheap these days, you just need enough space to fit it, also prime95 is just useless, you wont ever run into that type of load on normal usage, on prime95 you might get 95c but on gaming 65c, im on 5.1ghz 1.33v and it overheats on every stress test but runs perfect on games at a decent 70c

[@espi8997]

But not many cases have that 420 AIO support

[@anoryme8959]

Le 12600 kf très stable à 5.2ghz j'ai réussi à le monter à 5.75ghz mais pas stable. 5.4ghz atteint.

[@SkyChady]

Sounds like Ernie from Sesame Street.

[@onikonobi]

Lol ok im not the only one 😅

[@jdogdarkness]

Hey why is it that the utility statistics can go above 100%? Anybody have an explanation for that? Thanks!

[@meltingfaces10]

Does the 12600k have significantly worse silicon quality than the 12900k? That much voltage for only 5Ghz is kinda bad. For comparison my 12900k maxes out at 1.28v with TVB voltage optimization on at 5.2Ghz and pulls max 260w with stockfish. It's not like my CPU is a golden sample either, it only has a SP88 rating.

[@ActuallyHardcoreOverclocking]

an overall SP of 88 is above average for 12900Ks and since the overall SP is a combination of the P and E core SP your P cores could actually really good. Also 260W power draw in Prime 95 SFFT sounds abnormally low.

[@meltingfaces10]

@@ActuallyHardcoreOverclocking The breakdown is SP98/68 P/E, so the P cores are pretty good. However, there are a number of variables that could have a significant effect. I'm using adaptive mode + advanced VF offsets (-0.175v@48x,-0.075v@53x). I've matched the AC/DC LL values (36/36) to the VRM load line (Mode 6@800Khz). And I'm using washer mod with a delid (though it didn't increase undervolting headroom). Using a Unify-X btw.

[@ActuallyHardcoreOverclocking]

@@meltingfaces10 what stress test are you runnining? Also 98 SP Pcores are really good.

[@meltingfaces10]

@@ActuallyHardcoreOverclocking Stockfish and y-cruncher. The ASUS Strix D4 board I have can't undervolt as aggressively as the Unify-X, but I don't have access to the necessary equipment to explain why.

[@ActuallyHardcoreOverclocking]

@@meltingfaces10 Y-cruncher isn't very hot and if you have AVX512 enabled the CPU runs even cooler than with AVX256.

[@dackel4253]

could you do a small comparison video on how much radiator thickness/fam rhickness matter? Like compsring different setups with the phanteks T30 and the noctua NF-A12 and 27mm and 38mm radiators (or 30mm). Thats a very specific wish, but i think it would be very interesting. Anyways, i really like your videos, your do your channel name all honor!

[@shadow-ht5gk]

My god my 12600k is horrible compared to this… need 1.36 for 5ghz…

[@crispysilicon]

First

[@jrhowrey]

Has anyone told you how wasteful it is to just run prime95... The amount of power wasted for doing what amounts to nothing is kind of sad. You're not gaming, you're just letting it waste power.

[@gustavalbers3238]

Compare it with other types of useless power consumption (Crypto Mining) and you'll see it's not much.

[@PolskiJaszczomb]

Some people just live and waste air or bytes in internet. So?

[@givemeajackson]

How on earth did you get here?

[@elesinalex]

Has anyone told you how wasteful is your comment is? Sending the data to KZbin and showing it to other people also uses power you know

[@_monti142]

are you paying for his power? if no disconnect your keyboard and keep it disconnected