7700k 80c load at default clock, no room for OC?

Kryogen

Gawd
Joined
Feb 3, 2002
Messages
937
My 7700k is 80c under prime95 (AVX) load with a nh-d15s. idle is 30c. Goes back to idle temps within like 1 second of stopping prime.

if I raise it to 4800, it's fine, but it's 91 under load with a vcore around 1.35

Should I try to raise the multiplier, but keep the voltage below 1.3 and it's going to stop where it's going to stop with 1.3 max or something?

Opinions on that whole thing? My previous 2600k was oced from 3.4 to 4.4, and this new cpu looks like it's quite at its max right at 4.5 .... uh?

Does anyone know if you can leave it on adaptive but set a max vcore value in asus ROG bios?

Thanks
 
Last edited:
You will probably have to delid or get a better cooler if you want to push it any further. Which test is it? Small fft? What are your temperatures under normal load? P95 may be pushing temps unrealistically too far.
 
Um, my cpu is at idle most of the time... but when loading programs, its 55-60 or so.
 
Make sure you're not using the AVX enabled Prime95 tests or you'll get crazy high temps and are unrealistic. I suggest downloading and trying Asus Realbench Stress Test instead of P95 and see what your temps are. I bet high 70's to low 80's. I've got my 7700K at 4.8Ghz using a D15 as well and it's more than capable.
 
I am using the AVX prime95 yes.
Realbench gives me 68 max at stock.

Maybe get it to 4800 and try again?

I don't want to de-lid or switch to water cooling. I want to get a reasonable OC out of my system with that d15s cooler. I have a minimal amount of time+effort to put into this, and I don't want to go over the top and reduce the life of my cpu.

So, I am aiming at a reasonable OC with realistic temps for long term use.
 
Make sure you're not using the AVX enabled Prime95 tests or you'll get crazy high temps and are unrealistic. I suggest downloading and trying Asus Realbench Stress Test instead of P95 and see what your temps are. I bet high 70's to low 80's. I've got my 7700K at 4.8Ghz using a D15 as well and it's more than capable.

Realbench barely pushes temps over 65C with my setup, it doesn't seem to be worse case scenario at all.

AVX inclusion in apps is only going to increase over time. Seems like Prime 95 is a good stress test because it actually stresses the CPU...
 
I have a minimal amount of time+effort to put into this, and I don't want to go over the top and reduce the life of my cpu.
then settle for a lower OC and stay there. if you want higher youre going to need to put in the time and effort to get it there. from everything ive seen it seems 1.35v should be the max for a daily driver/247 on air. maybe [email protected] and call it a day. anybody trying to go higher is using water and 1.35+.

I and many others, even kyle, have stopped using prime95 due to way it overworks the avx causing unrealistic loads and temps and possible damage or shortened life. realbench DOES use avx but in a more realistic manor(+SSE4 and DXVA). if you absolutely need it "prime stable" and are worried about avx, then use 26.6 or older and test avx separately(i think kyle said use handbrake). I would use IBT for a couple hours, then realbench for a couple, handbrake for a couple if its encoding system and then games games games.

ps: your temps are correct for your setup without a de-lid, just higher than what you want.
 
how do you set a max voltage in asus bios, but still allow the cpu to draw less voltage if it needs to?

Basically, I want to see how much I can raise the multiplier while having a max of 1.3v
 
its labeled variable or offset or something like that. but most turn that off 'cause it can cause instability. ie when the voltage needs to be higher but doesn't rise quick enough and the cpu hangs. just set it to 1.3v that should get you to 4.8ish.
 
Ok, with 1.3 it failed to run at 4.6

Leaving it to auto, 4800, fails tests in prime 95 at 1.328 (sets it to 1.328 max and rund at 94 max, so I'm not upping voltage more without a delid or watercool, which I am not doing.....)
Soooo, I guess that voltage will be a max of Auto -> 1.328
That works in prime 95 full load with a max temp of 94 (full avx).

Did a test with realbench, temps are 78 max, with an idle of +-30, and voltage is auto to 1.296

Sooooo, I guess that it's going to be 4700 on auto, and that gives +- 1.3v under a realistic load with temps just under 80.
MEM is set at XMP 3200

Probably just going to leave it there, it works fine, voltage is reasonable, temps are ok, and it's on auto and no hassle.

Opinions?
Thanks
 
I kind of expected more because my 2600K gave me a 1ghz oc with temps like that.

Is that processor any better than the 2600k it replaces? I mean, I'm getting 4.7 instead of 4.4, is that really a better performer?
 
by itself clock for clock the 7700K is offering around 30% better performance, so yes it's an upgrade without take in consideration overall platform upgrades, on top of that you are adding another 300mhz, so yes, it will be definitively an upgrade.
 
I am using the AVX prime95 yes.
Realbench gives me 68 max at stock.

Maybe get it to 4800 and try again?

I don't want to de-lid or switch to water cooling. I want to get a reasonable OC out of my system with that d15s cooler. I have a minimal amount of time+effort to put into this, and I don't want to go over the top and reduce the life of my cpu.

So, I am aiming at a reasonable OC with realistic temps for long term use.

With great cooling, you will have more to gain from doing a proper relid job on it. When I relidded my 6700k on a custom loop, I not only got a 20c average drop, it also allowed me to redo the voltage/settings on the chip dropping required voltage for my 24/7 daily settings. I also run prime95, do not care to not use it either. Though krabs does have the avx divider to make it easier on you. If you stay with your current cooler, you won't have much success with it because of the shit TIM between the IHS... ie. relid is almost mandatory.
 
UP TO 30% is the key part. in [H]'s testing only one test(blender) showed a 30% increase. it entirely depends on what youre using the system for. for normal every day use and gaming there is very little difference.
 
UP TO 30% is the key part. in [H]'s testing only one test(blender) showed a 30% increase. it entirely depends on what youre using the system for. for normal every day use and gaming there is very little difference.

[H] CPU test are not the best to rely on in my opinion, too much synthetic and no realworld gaming so for me is useless (again, for me), I found some scenarios (gaming which its what I do the most with my personal machine) where my skylake at same 4.8ghz shown bigger than 30% my old 3770K. In some games as GTA V or fallout 4 the performance was simply dramatic due faster RAM, however that should be included on the overall landscape of an "upgrade" and Ivy Bridge by itself was already 5% - 10% better than sandy bridge, in every day task, under an SSD yes, you are right that nobody is gona tell a difference, but once you factor better chipset with newer features and drive support (nvme, M2, etc) things can be quite different from a six year old platform to a newer one.
 
if you don't take stock in [H]'s review that fine but I did cover the point aboot new features. new features is the biggest advantage. with a good gpu , in gpu limited games, you wont see much of a difference. in cpu limited games, yes, you will see a bigger improvement but it all comes down to use. IF you are using the new features or are cpu limited you'll see a bigger gain.
 
I read the review, but oh well. Mobo was 5 years old, that gave me a switch to M2, better mobo, faster ddr4, more sata 6 ports, ... so in the end, at least now I have a new 2017 platform that should be good for another 5 years.
And wow that thing is quiet with asus fan optimization. I love it.
 
or maybe I'll just wait a bit more to see if delid becomes "the thing", and then I'll buy a delid kit that has everything to do the job, and maybe I'll delid the damn thing myself also....

Liquid metal isnt an issue directly on the chip? It doesnt short circuit anything?
 
or maybe I'll just wait a bit more to see if delid becomes "the thing", and then I'll buy a delid kit that has everything to do the job, and maybe I'll delid the damn thing myself also....

Liquid metal isnt an issue directly on the chip? It doesnt short circuit anything?

Delisting is already a thing, people have been doing it regularly. It really took off with the 6700K having bad thermals.

The CPU die isn't electrically active, don't get liquid metal everywhere and you will be fine.
 
How much is a delid tool, and is there a complete written guide somewhere?

What kind of silicon, procedure, what tool, what liquid metal, etc...
Maybe I'll do it, that doesn't look too hard. And If I can lower temps 20-30C, that might allow me a reasonable OC at 1.35v, which I can't really do right now.
 
How much is a delid tool, and is there a complete written guide somewhere?

What kind of silicon, procedure, what tool, what liquid metal, etc...
Maybe I'll do it, that doesn't look too hard. And If I can lower temps 20-30C, that might allow me a reasonable OC at 1.35v, which I can't really do right now.

how have you missed all the [H] articles aboot this? go to the front page and click "CPU" at the top, cant miss them.
something I thought aboot with these 3d printed tools, there are tons of ppl online that provide 3d printing and prototyping. you send them the file and money and they print and ship it. someone without a 3d printer could use one of those services. haven't used em but heard they are pretty reasonable.
 
yup, yup(that's what [H] used) and idk. there doesn't seem to be anywhere cheaper than amazon but that it ridiculous for shipping!
 
The 3d printed tool does not support the cpu enough imo compared to a proper tool.

How much am I going to gain by delidding anyway? 2-300 mhz or so? Is that worth it?
 
The 3d printed tool does not support the cpu enough imo compared to a proper tool.

How much am I going to gain by delidding anyway? 2-300 mhz or so? Is that worth it?

You will lower the temperature more than you will raise the overclock. These CPU will happily run along at pretty much anything under 100C, just depends on what you are comfortable with.

After seeing the temps, I redid my entire cooling setup once I got my 7700K. I don't even plan to overclock :eek:
 
so not too many users here delidded their 7700k?

the only reasonable option looks like the rockit tool imo, with the repo guide, and then some liquid metal and rtv?
looking at 100$ or so CAD I guess.

Does the LM stay good all the time or is there ever a need to reapply?
Would like a 5.2 chip instead of 4.7
 
so not too many users here delidded their 7700k?

the only reasonable option looks like the rockit tool imo, with the repo guide, and then some liquid metal and rtv?
looking at 100$ or so CAD I guess.

Does the LM stay good all the time or is there ever a need to reapply?
Would like a 5.2 chip instead of 4.7

You should try pushing the chip now to see how high you can get. If you have a golden chip and it runs at 5GHz on air, might be worth removing the lid. You may find you max out at 4.9 (or whatever) at which point a lid job won't be getting you to 5.2 anyway.
 
Well the noctua performs as well as the corsair h100 anyway so I dont get the big deal about air?
 
Well the noctua performs as well as the corsair h100 anyway so I dont get the big deal about air?

IMO it comes down to misinformation. People think that because it is water cooling, it must be better.

Heck, people put aftermarket HS's on non K chips. The stock Intel one is fine if you are not overclocking.
 
The d15s cools much better than my h80 while also being much more silent.

Ok i will probably order the rockit kit for delid and relid, amd report with liquid ultra mod what oc i am able to get snd the temp diff.
 
IMO it comes down to misinformation. People think that because it is water cooling, it must be better.

Heck, people put aftermarket HS's on non K chips. The stock Intel one is fine if you are not overclocking.
I agree about the misinformation. The CLC advertising deliberately hypes the 'water' part and glosses over the ultra low quality of CLC components.
But I do not agree with stock coolers being fine. Any half-decent $30-40 cooler will be much quieter and deliver lower temps than stock coolers do. I have 5 intel towers here and none have stock coolers. I don't even own a stock cooler. Only 3 of them are overclocked and those overclocks are not extreme. i7 920 @ 4.0GHz, old Athlon 620 @ 3.5Ghz (I think) and i7 980X @ 4.3 Ghz.
 
I agree about the misinformation. The CLC advertising deliberately hypes the 'water' part and glosses over the ultra low quality of CLC components.
But I do not agree with stock coolers being fine. Any half-decent $30-40 cooler will be much quieter and deliver lower temps than stock coolers do. I have 5 intel towers here and none have stock coolers. I don't even own a stock cooler. Only 3 of them are overclocked and those overclocks are not extreme. i7 920 @ 4.0GHz, old Athlon 620 @ 3.5Ghz (I think) and i7 980X @ 4.3 Ghz.

You have to realize, we are the very tiny minority when it comes to computers.
The overwhelming majority of systems run the stock Intel cooling. Of course an aftermarket cooler gives much better temps, but modern Intel processors are happy to chug along at 85 deg all day long.

Do I run stock cooling? No, I have a custom water loop. That puts me in an even smaller sub-set of a very small sub-set of cooling options.

Most people don't give a crap about computers as long as they can load Facebook and Tumblr.
 
Indeed computer builders are a small segment of computer industry, but I'm not sure how small our group is. I don't think it is a 'tiny minority'. Maybe the people around me are not representive of general public. I like to think they are a reasonable representation of public in general. In my group about 50% are using home built systems. Of the 50% that are not the majority are using notebook/laptop systems. About half of the homebuilt are older gamer systems re-cycled to parents or grand parents .. people who don't need cutting edge systems .. most of these are like you say, more interested in Facebook, Google, etc. Then of course we have the tablet and phone people who, are the real Facebook and Tumblr crowd.

It would be interesting to see a detailed data about computer owners.
 
Apropos PC builders:


Back in the days when those cream colored boxes where all new and less than 1 promille of all people had one it was US that spread the word, got people fascinated, thrilled and full of desire to have the SAME as we have. I could count up dozens of people ( mostly men but also a few women ) who I got into gaming on PC foremost and later when PC's conquered the general households far and wide it was still US they called for reference, come by to fix things, install a printer..etc..US GAMERS, hardware enthusiasts that can spend hours to fix a DLL error whereas they dont even know whats broken and why we grab our hair, ask for an ashtray and another coffee and say" soon, SOON, I am ALMOST there..." kind of thing.

That alone got me to a point where business people came by and asked for help, they veven offered a thing called "CASH" :D

As I am lazy by nature it needed the money to come when they called and not when I felt like it but over time I got into it and am doing it now 15y+.


If there hadn't been US gamers it would have taken considerably longer to spread the PC's across all households and SOHO's. People did not now what to buy, how to set it up, maintain it and sometimes fix it too when needed. Many many of us dont get their hours paid, brothers, sisters, kids, wify, nephews and uncles...etc.. they all call US to fix their PC's, setup their router and Internet, ask what to buy next etc..

We are thanked not often enough, we do it since we love it and we know how helpless they are, forgetting what we sacrifice and just do it.


I lately got 5 PC's back from a company I service, Win7 on it, Dell...not bad if you dont have anything. I made them all new with Win10, installed the usualy stuff, added mouse and keyboard and made 5 families happy for free. It was easy for me, couple hours and 5 PCs are done but a huge step forward for those who got one.



to come to doyll's question:

From all people I know in person, no one that does not have IT as profession knows much about computing, they use them and that is basically it. I have been to hundreds of homes fixing PC's, most have no clue about PC's routers and how this all makes sense, basically walking in the dark hoping not to hit a pole !!!

Some of my gaming friends know enough to install some drivers the basic way but would not be capable of fixing a broken NV install etc.. not speaking of installing an OS and Bios settings. I have 1 friend who just got into PC's again and I think him, he could also become one of us, but it takes years to learn all the small stuff and time to work on many many different systems and OS's.


There are more good car mechanics than Computer "meachanics" ! That is for sure.

To become one of us you have to master the english language, no matter what your Passport reads. Many disqualify at this hurdle despite they had everything to get there. Most of the net is english, dont speak it and you exclude yourself from like 90% of IT-content.



I never intented to become what I became, it was a hobby foremost and than a need ( as I knew nobody to ask and fix it, I had to learn it myself ). I have had a few apprentices, most were plain useless, only 1 of them became an Admin at 1&1, all others didnt have what it needs, dont ask me what it is, but if you dont have it Computers arent yours, that I learned over the years.

Some people have an attitude towards certain things that will void their way into understanding computers. Limited curiosity is one of them that I can name. You gotta be like a cat ! Ask the questions, look behind, do it again and watch it again, LEARN.



That is my understanding about "US"...and all others who operate computers :D
 
Apropos PC builders:

*snip*

That is my understanding about "US"...and all others who operate computers :D
Both enightening and fun to read. :)

No idea where on Earth you are, but maybe where you are is part of the difference in peoples. Here about half have at least one person in household who 'fixes' thing rather then buy new or call repair-man. About half of theasr are quite good at fixing, .. other half are 'Tim the Tool Man' .. know enough to make are real mess of things. This is rural UK (not that 'rural' anymore) and also rural America / Canada. Maybe it has something to do living in low population density open country were cattle out-number people and it's many miles to town adn ofthen much further to major repair people .. meaning our parents and their parents and their parents parents parents all grew up in a world where if we don't fix it ourselves they didn't have it to use and had to do without.. I do have to admit that most kids today do not seem to care about how something works. They just want it to work .. if it does not they want a new one .. right this instant!;)
 
I tried using Prime95 before when I had a Haswell processor, but I find that using the Skylake/Kabylake processors, it doesn't bode well. Plus, it requires a lot of time to finish. What I do recommend is using the Asus Realbench program and do 5 benchmark tests in succession. I find that this simulates "real-world" use as opposed to running a synthetic stress test like Prime95.
 
Back
Top