what is the lowest volt you think for a stable 9700k at default 4.6 ghz all 8 cores,or 4.8ghz oc?

Joined
Sep 14, 2016
Messages
33
interested in how much volt the 9700k requires on average at the stock 4.6 ghz all 8 core turbo frequency? I know this can vary depending on cpu quality and motherboard volt settings/motherboard models etc , if i'm not mistaken? this would be with MCE off and multiplier set to 46-48, then lets say we can adjust the LLC and volt method like adaptive/auto/offset etc, what would be best way to get lowest and stable volt for 4.6 ghz all core turbo or small overclock to like 4.8 ghz max?



saw some users getting 4.8 ghz stable all cores at 1.2v for example, this is probably lucky, you think 1.2v for 4.6 ghz all 8 cores should be super easy and no problem? or can you go even lower?



btw I know the single core turbo is set at up to 4.9 ghz, this is irrelevant to me and i'm only interested in all 8 core load frequencies like gaming etc, aswell as other reasons I would like to keep the frequency at 4.6-4.8 ghz max, with as low volt as possible (case airflow/cooling not the best, the cpu cooler itself is also not the largest one to fit into the small case, etc)


appreciate any help and comments!
 
Wow.. you're asking for a lot of info!

At 4.6 the average 9700k needs about 1.2-1.25v at stock

BUT this is without factoring in AVX. If you mean reducing the avx offset to 0, 1.3v is more like what you'll hit. AVX is used in games.

MSI did an analysis of a sample lot of processors here: https://www.msi.com/blog/intel-9th-cpu-overclocking-5ghz-with-z390-motherboards

I'm not going to hotlink the image, but here it is: https://asset.msi.com/global/picture/news/2018/mb/intel-9th-3.jpg

They also did some binning

We binned some 9th Gen processor samples and came out with the frequency-voltage relation. We divided the samples into grade A, B, and C, according to the binning result. A is the best at overclocking, B is in between and C is not good. The graphs below show the percentage of each grade. As you can see, 20% of i9-9900K is good at overclocking.

https://asset.msi.com/global/picture/news/2018/mb/intel-9th-2.jpg


It is possible to hit 1.25 under avx loads @ 4.3-4.5ghz. but not many people have this granularity of testing
 
Wow.. you're asking for a lot of info!

At 4.6 the average 9700k needs about 1.2-1.25v at stock

BUT this is without factoring in AVX. If you mean reducing the avx offset to 0, 1.3v is more like what you'll hit. AVX is used in games.

MSI did an analysis of a sample lot of processors here: https://www.msi.com/blog/intel-9th-cpu-overclocking-5ghz-with-z390-motherboards

I'm not going to hotlink the image, but here it is: https://asset.msi.com/global/picture/news/2018/mb/intel-9th-3.jpg

They also did some binning



https://asset.msi.com/global/picture/news/2018/mb/intel-9th-2.jpg


It is possible to hit 1.25 under avx loads @ 4.3-4.5ghz. but not many people have this granularity of testing


thanks for answer, interesting stuff indeed for me! also i've looked around and seen several persons running 1.17-1.2 v for stock 4.6 ghz all 8 core turbos on their 9700ks,

did not ask them for avx offsets unfortunately but will try to do so and update,


btw, AVX being used in games, this is not the same as avx used in a prime95 stresstest also, correct? you are not going to spike your cpu to 100c just because the game is using avx instructions,
whereas it can easily go to 100c in a few seconds in a prime95 test with avx with inadequate cooling etc, right?


is there a list also somewhere with games using avx or can you atleast mention a few popular ones ? just so I can try finding some benchmark frequencies and temps on my own from those games etc !


cheers and thanks again so far !
 
Game wise, basically anything released in the last 5-6 years uses AVX. No it won’t peg the cores like prime does, but they are the same instructions

Assassins creed, deus ex, heck doom 2016, battlefield series, crysis 3, far cry 4-5..
 
Is AVX in games completely separate from the graphics API used? For example, are the AVX operations more efficient when rendering with DX11 than OpenGL?
 
Game wise, basically anything released in the last 5-6 years uses AVX. No it won’t peg the cores like prime does, but they are the same instructions

Assassins creed, deus ex, heck doom 2016, battlefield series, crysis 3, far cry 4-5..

yeah, gaming temps in those games even with avx, is faar away from cpu stresstests using avx instructions, from what I can gather anyway, and since I wont be doing any avx stresstesting but simply games,

I'd assume its pretty safe to not set an avx offset in this particular case? either way should not be a big issue from everything I looked at so far :S


but good to know none the less!
 
yeah, gaming temps in those games even with avx, is faar away from cpu stresstests using avx instructions, from what I can gather anyway, and since I wont be doing any avx stresstesting but simply games,

I'd assume its pretty safe to not set an avx offset in this particular case? either way should not be a big issue from everything I looked at so far :S

but good to know none the less!

You want your system stable, but if it thermally throttles in the worst-case scenario I'd argue - so be it.
Optimize your system for the workloads you really run.

Personally - I don't run an AVX offset. I just don't have anything besides the synthetic stress tests which cause my system any grief, and many apps are using some AVX. They just don't pathologically hammer it.
 
My 6700K is clocked stable without using Prime95 AVX workload and it hasnt failed me in any game.
fyi
 
I consider handbrake stable to be 90% stable. It is typically the heaviest workload I put on my computer, short of compiling which is a more transient load.

If it can run x265 or x264 for 2-3 hours, then it's stable enough for me.
 
I consider handbrake stable to be 90% stable. It is typically the heaviest workload I put on my computer, short of compiling which is a more transient load.

If it can run x265 or x264 for 2-3 hours, then it's stable enough for me.

I know where you are coming from, I use a lot of Adobe After Effects personally and use an older version that natively supports multicore processing so have a thread running for all cores for best possible rendering performance and rendering a 1 hr or so video can very easily lead to errors if the system is unstable but it can sometimes pass 1 hr prime without issues, likewise ofc gaming and other lighter tasks. So far it's been the case if the rendering is stable, just about anything else I do on the computer typically also is.
 
btw, AVX being used in games, this is not the same as avx used in a prime95 stresstest also,
AVX is AVX
Is the same instructions. The game coding might not be a stressfull as prime95 ven with AVX though

but you also haveto consider Prime95 is CPU stress test only.
Games stresses multiple component. They are not the same kind of tests


I consider handbrake stable to be 90% stable. It is typically the heaviest workload I put on my computer, short of compiling which is a more transient load.

If it can run x265 or x264 for 2-3 hours, then it's stable enough for me.
The issue here is that you dont know if you CPU is doing miscalculations as doing encoding does not verifying the results.
 
Maybe so, but usually a heavy AVX load like handbrake will destablise an overclock (and blue screen or black screen crash) if it isn’t right
 
Back
Top