Overclock.net banner
Cancel
Create thread New Forums

Overclocking i7-10700k results, comparisons & discussions

Jump to Latest
348K views 889 replies 110 participants last post by  GuruSR  
P
#1 · (Edited)
I saw the i9 version of the thread so I thought I'd start the one dedicated to the i7.

So what are your results with this chip? ofcourse, proper results with aida64/rb/p95

msi unify z490 + nzxt x73 = currently sitting at 5.2ghz(48 cache) 1.295v LLC3

i guess it's a good chip, seen a lot of people struggling at higher voltages.

So what about you guys?
 
#635 ·
For 5.1 GHz need less then 1.4 volts

Inviato dal mio MI 9 utilizzando Tapatalk
 
Save
Reply Quote Rep+ Rep+
#636 ·
Again, why don't you specify what volts you are talking about, otherwise it is meaningless.

Silicon Lottery is 1.4v in BIOS, and with their LLC setting is 1.23v die sense under load.
 
Save
Reply Quote Rep+ Rep+
#641 ·
Depends on the load-line and droop at load. But probably.
 
  • Rep+
Reactions: TheBoom
Save
Reply Quote Rep+ Rep+
#643 ·
So I’ve been wondering about a certain aspect of overclocking these chips and LLC vs voltage.

Lower llc with higher voltage leads to more vdroop and better transients in stability testing right?

Unfortunately that also means higher idle and partial load voltages and that kind of applies to real world usage?

So why bother using lower llc when it leads to slightly better stability in stress tests but inadvertently leads to higher voltages and temperatures in real world usage?

I feel like I should just find the vmin at llc5 or higher and that actually gives me lower idle and partial load temps.
 
Save
Reply Quote Rep+ Rep+
#644 ·
Assuming you mean fixed vcore here.
You need a balance. But if the load is not a very heavy amp load, you may see a benefit from tightening the LLC and lowering the bios voltage.

Too low LLC is bad because you need to set your voltages very high at idle. I think many people are uncomfortable with a 1.45v idle voltage even if the vcore drops down to 1.20v at full load.
Games do not usually put a 200 amp load on your CPU. So for games, the transients are not important. They only start mattering in games when the LLC is too tight, example, LLC8 (no vdroop) may cause issues for sure. LLC7 needs to be tested. You would need to determine your "load vmin" (minimum voltage under load needed to be stable in a game) at LLC5, for example, which is very droopy on a Maximus 12 Apex/Extreme, and then compare it with LLC7 and determine if the vmin changes. If the vmin is no higher than 10mv at LLC7 in XYZ game, compared to LLC5, then there is no problem using LLC7, as long as you do enough testing. Avoid stability testing in stress tests--those will get nuked by too aggressive LLC. Just determine your minimum required load vcore for whatever you are using (e.g. games).

Of course you can also use a very droopy LLC and use adaptive/offset voltages too.
 
Save
Reply Quote Rep+ Rep+
#645 ·
why do you consider LLC as voltage when it is based on resistance?

Inviato dal mio MI 9 utilizzando Tapatalk
 
Save
Reply Quote Rep+ Rep+
#646 ·
It is voltage as a function of load current. It is characterized by its slope, which is resistance.
 
Save
Reply Quote Rep+ Rep+
J
#654 · (Edited)
I'm back, lols...
Returned the Scythe Fuma 2 and replaced with an Arctic 280 AIO 'cuz why not...AIO is set up in push pull config.

Dirty OC - 50 (everything else at auto) with a fixed voltage of 1.325v and power savings and turbo limits disabled gets me 1.285 VR VOUT and 1.350 VCore when running Prime95 small no avx in the mid 80s and POUT 179W.

Linpack X sees 1.290 VR Vout with same Vcore and POUT and temps in the mid 70s.

2x8GB RAM at stock 2133mhz.

Hoping Im on the right track.

EDIT:
1.300 Fixed got me 1.261 VR VOUT and 1.320 VCore with 80c and a POUT of 166W on Prime; On Linpack x I saw similar numbers thought temps a little lower and I scored 3 out of 3 on same residuals.

EDIT 2:
1.275v Fixed gets me 1.273 VR VOUT at Idle; 1.238-1.240 VR VOUT under load; low 70s; and POUT 154W doing 3 runs of LinX....BSOD on third cycle of LinX

EDIT 3:
1.285v Fixed gets me 1,282 VR VOUT idle; 1.248 VR VOUT using LinpackX, low 70s; POUT 166W, but residuals not the same.

So now I am wondering what to do next...
 
D
#655 ·
I have just signed up to overclock. i have the 10700ka (avengers edition) the maximuss xII formula rog with the asus rog 3 fan lc 360. with 3200 ram on the xmp mode with all core set to by core. had prime 95 running for 20 mins and hitting stable 65-70 degrees so quite happy only changing the by core blck to 51. had stable 5.122 all round. its a bit late and many a beers right now to keep messing but keep to goand read the posts before and see what this beast compares to.

I'm Nick old school it tech, its been a very long time since i have had something this good that I own to play with. ive played with and have worked with 10k stock market guys pc's oh ****ing daym ive drooled a few times.

any how be keen to see what it can do with your help and more then keen to start helping others who are still learning.
 
J
#656 ·
Hello guys,

I am not really an experienced overclocker, however, I have overclocked all my CPUs since I own a PC. I recently updated from my 4790k because a good friend of mine wanted a cheap system and I wanted a new one anyways.

Here's the deal: for the love of god I can't figure out how to limit the idle cpu core voltage. My board allows for fixed, adaptive or override modes regarding vcore. Fixed works fine and as expected. However, If somehow possible I'd like the cpu to reduce voltage when it's not need.

Therefore, I tried the adaptive setting. But whatever I do I can't seem to be able to find out how I limit the idle voltage. I shoots up to 1.36V all the time. Under load I can ifluence the voltage just fine.

After hours (!) of google and trying out setting I still don't get it.

Can you guys give me a hint?

Thank you!
 
#657 ·
JazVM said:
Hello guys,

I am not really an experienced overclocker, however, I have overclocked all my CPUs since I own a PC. I recently updated from my 4790k because a good friend of mine wanted a cheap system and I wanted a new one anyways.

Here's the deal: for the love of god I can't figure out how to limit the idle cpu core voltage. My board allows for fixed, adaptive or override modes regarding vcore. Fixed works fine and as expected. However, If somehow possible I'd like the cpu to reduce voltage when it's not need.

Therefore, I tried the adaptive setting. But whatever I do I can't seem to be able to find out how I limit the idle voltage. I shoots up to 1.36V all the time. Under load I can ifluence the voltage just fine.

After hours (!) of google and trying out setting I still don't get it.

Can you guys give me a hint?

Thank you!
Click to expand...
What Motherboard? You need adaptive + EIST (SpeedStep) enabled and/or Speedshift enabled in BIOS.
 
Save
Reply Quote Rep+ Rep+
T
#659 · (Edited)
New owner of i7-10700KF on MSI MEG Unify Z490 maotherboard , Patriot 4400Mhz 16Gb RAM , Kraken x73 360mm AIO
Followed this guide to overclock :

I am sitting now on 5.1Ghz all cores , 0 offset AVX, 4.5Ghz Ring , 1.4Vcore - adaptive , LLC6 , CPU SA = 1.35V ; CPU IO = 1.28V ; DRAM = 1.5 V , XMP profile 4266Mhz 16-16-18-36

All stable in Windows 10, AIDA64 test, CinebechR20.

BUT, crashing in Prime95 , Blender ...


Any hints how can I get it stable on Benchmarks ? (at 5.1Ghz obvioulsy)
 
Save
Reply Quote Rep+ Rep+
T
#661 ·
You were right . Changed to LLC4, Override .
Blender went through (CPU temp 96-99deg ! -- I will delid CPU and direct cool tomorrow ...) .
Vcore= 1.392V during test.
Test result : 2min 45 s -> bmw blender

Now, for daily use , I should go back to "Adaptive" leaving LLC4, shouldn't I ?
 
  • Rep+
Reactions: prhbtd
Save
Reply Quote Rep+ Rep+
P
#662 · (Edited)
No override is the best, vcore maxsafe 1.5v by msi engineers so 1.45v no problem. But you should find lowest vcore with which ghz you like. I prefer 5 ghz with Llc 6 vcore 1.38 for low temperatures because my games are not cpu bonded. I use custom 2080ti hybrid from asus 2080ti turbo blower type and daily use 2100 mhz aprox 55 celcius. So it depens on you if you wont need high ghz for daily use you dont need to create extra heat. You can not kill 10700k chip below 1.5.
 
T
#663 ·
To be honest, for everyday use and games , I did switch back to 1.4V , LLC6 , adaptive , on 5.1Ghz. This way, for example in games I have , like for Battlefield V , around 25% CPU usage, CPU temperatures around 50-60deg , and more importantly, Vcore is going down to 1.35-1.37V . So I feel safer in terms of CPU voltage and still having a fast 5.1Ghz all cores in action.

With override and LLC4 as you suggest, Vcore is kept higher, and indeed no crashes on benchmarks . But for everyday use there is nothing to stress CPU like a heavy CPU benchmark, not games nor video editing/productivity software.

I will see tomorrow if direct die cooling will change the ecuation
 
Save
Reply Quote Rep+ Rep+
#665 ·
You can’t kill a chip below 1.5v sure, but you will definitely degrade it sooner. Anything above 1.4v under high loads will.

Also you probably need sub ambient cooling to cool 1.5v unless you live in Iceland or the likes.
 
Save
Reply Quote Rep+ Rep+
T
#666 ·
Ok , I delidded the i7-10700k today and applied Liquid Metal - AIO cooler direct die.
The results are very good ... temperatures went down significantly . For e.g. , doing Blender before at 5.1Ghz LLC4 1,4V override temp was hitting 99deg.
Now , for same config , Bleder runs with CPU 76deg. .... Text Font Screenshot Technology Software
Computer hardware Technology Motherboard Electronic device Computer component


Moreover, better cooling allowed me to step up to 5.2Ghz stable.
I've done some benchmarks at 5.2Ghz CPU, 1.405V Vcore Override, LLC4, 4.5Ghz Ring, .. CPU temp max was 79 deg.

So, now for everyday use I am on 5.2Ghz fixed all cored, 1.405V vcore, Adaptive, LLC5, 4.5Ghz ring. The Vcore stays all the time under 1.4V , between 1.351V - 1.397V .
Being a Intel 14nm CPU , as Builzoid said, such voltage should still give a long life for the CPU (I hope :)).

Tried to change 5.3Ghz with same setting , but BSOD in Windows
I am kind of sure I could go there with more voltage , but it's not worth....
 
Save
Reply Quote Rep+ Rep+
T
#667 · (Edited)
After watching some Youtube videos , here is the conclusion :

In games, Intel on 8 cores is still ahead AMD Ryzen 5800X !!!
I mean i7-10700k @ 5.2Ghz all cores has better FPS in 1080p games compared with Ryzen 5800X 8 cores. Moreover , if you try to overclock Ryzen 5800X , say to 4.7Ghz water cooled , the FPS is going down , not up .... same single core score R20 score is going down, from 625 to 608 .... (Youtuber : Hardware Unboxed)

That means, on 8 cores Intel is still THE BEST GAMING CPU !
 
Save
Reply Quote Rep+ Rep+
T
#670 ·
Hi All,

I'm currently tinkering with my CPU and I could use a little help with understanding a few settings and results. I'm trying to somewhat follow Buildzoid's guide, my results are quite different though.
I'm using the following settings:


CPU Base Clock: 100MHz

Enhanced Multi-Core Performance: Disabled

CPU Clock Ratio: 50

Ring Ratio: 46

AVX Offset: 0

Extreme Memory Profile: Profile1 (Corsair Vengeance LP 3200)

Vcore Voltage Mode: Override Vcore

Internal CPU Vcore: 1.350V

Internal CPU Vcore Offset: Auto

DRAM Voltage: 1.200V

CPU VCCIO 1.100V

CPU System Agent Voltage: 1.200V

CPU Vcore Loadline Calibration: Standard

CPU Vcore Protection: 400.0mV

VXAG Protection: 400.0mV

CPU Vcore Current Protection: Extreme

Idle temperatures are below 30C (ambient is 23C), low 90's under heavy load (Linpack Xtreme/Prime 95 with AVX2). All that on a H150i Pro XT (Balanced pump, fans at 1800RPM).

Now the interesting thing is, the BIOS Vcore is set to 1.350, but HWInfo's VR Out shows 1.270-1.280V while idling, and roughly the same (1.260-1.290V) while running stress tests (occasionally jumping to 1.340V for a split second). Vcore readings are between 0.700V and 1.356V (occasionally jumping to 1.420V for a second) respectively.

Running Prime95 small FFT gives a non fatal l0 cache error every single time at the 30-31 minute mark, no matter what settings I fiddle with.
Is there anything I should change, further adjust? I'd like to achieve a stable overclock at a lower voltage if possible.


Any advice, help is much appreciated.
 
#673 ·
What frequency and OC method?
 
Save
Reply Quote Rep+ Rep+
#677 ·
That looks like a typical VRVOUT at load for this chip at 5 GHz,
 
  • Rep+
Reactions: JMak00
Save
Reply Quote Rep+ Rep+
J
#678 · (Edited)
Ok, thanks...pardon the next question...what does that mean? In other words, is it safe running 1.30 - 1.35VRVOUT on a daily oc?

EDIT: At idle, VCore is bouncing between 0.744 to 1.344; while VRVVOUT is bouncing around 1.300 to 1.345.

EDIT 2: I ran 4 hours of Cinebench last with a 1.450 VCore set up as described above and a 4x8 set of Patriot sticks at 4000 16-16-16-38 without any whea errors or bsod and temps didn't go above 85c (arctic liquid freezer 360).
 
N
#682 · (Edited)
Hello!

Long time overclocker here, finally happy with my results.

It's been a journey; Mobo that died and took 1mo for repair, radiator fans that stopped working, a CPU waterblock that I overtightened (breaking an ear off)...

Here's the setup;

W10
Custom Watercooling Loop - 360mm + 240mm radiators with good fans
MSI Unify Motherboard* - Second one!
10700KF
GSkill 3200 CL14 Memory, 4x8 sticks - Samsung B Die!

The Results;

5.1ghz - Stable on latest A30 Bios. Voltage offset +035mv. HWiNFO reports 1.388-1.90v. Stable as a rock.
4.8ghz Cache
4.0ghz Memory at CL14/15/15/36. 1.52v. VCCIO 1.22v / VCCSA 1.36v

Observations;

Had a real booger of a time getting this chip stable. If I enable Cstates, I get cache errors. Even at stock settings. I could boot into 5.2ghz all core (fixed) with 1.4v (fixed). But unstable. Tried for 5.1ghz, same voltage, no luck (cache errors). 5ghz, same voltage, stable. Got down to 1.38v (again, fixed), but that's the limit of stability. 1.37v causes cache errors. But memory was stable at 4266 CL16, Cache was stable at 4.7, and 5.0 all core without AVX offset, ran really good!

Decided to keep pushing. Downloaded the new, updated bios version 7C71vA3. Got luck with Auto Voltage + offset, 5.1 stable! Can't get ram back to 4266 CL16, BUT, can run 4000 CL15! So, I'll take it.

Any other voltage mode (fixed, adaptive, etc) gives cache errors, even if set the same.

Hope some of you find my post useful!
 
#683 ·
Anyone with Asus and the new 0901 bios tried out TVB?

Seems pretty cool that we can do it now on a 10700k. They don't say what the auto boost +1 or +2 temperature thresholds are though.

If you do it manually it seems backwards with a negative offset instead.
 
Save
Reply Quote Rep+ Rep+
#684 ·
TheBoom said:
Anyone with Asus and the new 0901 bios tried out TVB?

Seems pretty cool that we can do it now on a 10700k. They don't say what the auto boost +1 or +2 temperature thresholds are though.

If you do it manually it seems backwards with a negative offset instead.
Click to expand...
Yes. What is does is allow you to shift down the multiplier by one or two clicks based on a temperature across the per core multiplier profile (or all cores). It doesn't boost anything. It is very difficult to test for stability since you have to test not only with certain number cores active, but as a function of temperature as well. I have a 52-51 per core profile setup for it. For example, to test a 4 active core count, I use prime95 with 4 threads., but 6 cores with the 4 same profile I want since there will always be other cores doing OS activity. Then you have to vary temperature. I don't have a lot of confidence it is completely stable or optimal.

You can see what the auto-boost (which just increases your multiplier ands sets a fixed profile) temperatures are in BIOS once you set it and reboot to BIOS.
 
Save
Reply Quote Rep+ Rep+
#685 ·
BTW you can also see and change the setting via the Intel Xtreme Tuning utility or with the windows utility Shamino posted in the Asus Rog forum. Both can be used to adjust real-time in Windows. Here is what I have mine set to in BIOS displayed using the latter:

2466541
 
  • Rep+
Reactions: TheBoom
Save
Reply Quote Rep+ Rep+
#686 ·
Thanks for the info, sorry for the late reply as I was having issues yet again with my memory oc after updating to 0901.

I think it’s possible to properly test for stability, but you have set it manually and sync all cores.

Or just test the boost clock manually with a set voltage I guess.
 
Save
Reply Quote Rep+ Rep+
#693 ·
I wanted to post some info on my setup. I actually bought 2 10700k chips from Amazon recently. They were on a great deal and wanted to get a better chance of silicon lottery.

I have been fine tuning over the last few days to squeeze every ounce of performance out of my chip that I kept (I returned the other one).

Soo here goes.

I originally had an Asus Strix z490 ITX board after a lot of people recommended it. It is a great board... but I gotta say... not the best for my situation. I was not able to get the best out of my 2 chips I had. I often needed much higher voltage than the MSI Unify z490 ITX board so I would highly suggest moving over to MSI Unify if you are not seeing much gains out of your 10th gen chip. I can only really speak to my 10700k chips though.

To save me from writing the same thing over and over, AC in my post is All Core. and all voltage I post will be VCORE as reported by HWINF64 unless stated otherwise.

I have a slew of stability testing I do and only move onto the next test if it passes the previous test with 0 crashes, BSODs and of course 0 HWINFO64 hardware errors:

Test 1: Prime95 custom, avx disabled with ffts 4k-128k at 1min each for 2 hours.
Test 2: Real Bench Stress Test for 2 hours on 16gb option.
Test 3. Hardcore Multitasking Simultaneous Test:
-Real Bench Benchmark (set to infinite run)
-Cinebench R23 (tests for AVX as well) multithread (custom option to run for 360 min)
-1 window of Chrome streaming a 4K Youtube video (usually earth 4k 7 hour video)
-Netflix App with 4k movie streaming
-1 website with constant data streaming (I use investing.com and click on the Crypto currency page, it is constantly streaming live data from tones of different exchanges).

I do this for 4 hours.

Test 4:
I run the exact same test as test 3 but this time for 8 hours to test a full workday of stability. I also turn on my webcam and leave that on a window by itself.

If it passes test 4 with zero hardware errors in HWINFO64 I certify it stable for daily use. I did this because I use to run each of the programs separately and would notice a random error throughout the day sometimes but all the individual tests would pass fine. It wasn't until I started doing Test 3 that those errors would reveal themselves later. So I highly suggest doing something similar if you want to truly test daily driver stability.


Anyways... now that you know what I consider stable... let's talk about the chips.


2 chips, both 10700k. One was made in China, the other Vietnam.


ASUS 5.1ghz all core at llc 4, ring at 46:

Chip 1 was stable 1.415v on the Asus board. Chip 2 was stable at 1.375v.
I wasn't satisfied so I purchased the MSI board, started testing both chips. I purchased it because it seemed like ASUS went cheap and had much less power stages. MSI had like 8 or 9 so I figured the power supplied would likely be much better.

MSI 5.1ghz all core at llc 3, ring 46:

Chip 1 stable at 1.36v, Chip 2 stable at 1.295v!!!! I was pretty much jumping after Chip 2 passed Test 4 and I saw 0 errors. I couldn't believe it so I re-ran the test again and by the afternoon, again 0 errors.

I decided to go for gold and pushed Chip 2 to 5.2ghz AC... after 2 days of testing, it was stable at 1.375v. I did need to drop LLC to 1 for it to be stable. I wasn't OK with that as a daily voltage as i am sure the transient spikes were well above 1.45v but largely unknown. I would get to ~1.39 vcore under load with LLC 1 though. I thought about running it at 1.395 vcore with LLC 3 (almost no droop) but it wouldn't pass. I tried 1.4cvore with LLC 3 and it did but I am not a fan of high vcore, I'd rather have lower voltage during off-load times and slightly higher voltage under load. I know this might be different for some others but... he do you.

I decided I didn't like the 1.375v at 5.2 AC as a daily so I decided to tune my 5.1ghz OC.

I ended up with this setup setup for multipliers.

Core 0 - x53
Core 1 - x53
Core 2 - x51 (weakest core)
Core 3 - x52
Core 4 - x52
Core 5 - x53
Core 6 - x53
Core 7 - x53

1.320-1.325vcore idle LLC1 1.344vcore under load.
1.3v VCCIO
1.3v VCCSA

Under load, the system will effectively only do 5.1 AC but while not under load, you consistently get the cores hitting their max almost ever 2-3 seconds.

I also need to mention, I was able to get my vcore lower by disabling EIST, C-states, Speed Step. I do have Turbo on.

I did not enter an AVX offset as I don't do anything that really uses AVX hardcore. And Cinebench R23 does test for medium AVX usage so, good enough for me.

I am running all of this in my NZXT H1 with the stock 140mm AIO, temps don't get above 90c on the hottest core. Package temps under load at 88c. Idle, they drop down to 25c. This case is small and runs fairly hot so I am impressed with it.

I am thinking of deliding the 10700k, it really is a great piece of silicon that I got so I really want to try and push it to 5.2 AC at something much lower. I tested my system on a cold night where temps were in the single digits 5c... got the system to be stable at 1.335vcore on LLC 3, temps never got above 81 sitting outside in that cold air, so if I could get my temps down 7c-10c I think I could really push this to 5.2ghz AC at a comfortable vcore slightly higher than my current setup.

Let me know thoughts and if you have any questions!
 
#694 ·
TorontozFinest said:
I wanted to post some info on my setup. I actually bought 2 10700k chips from Amazon recently. They were on a great deal and wanted to get a better chance of silicon lottery.

I have been fine tuning over the last few days to squeeze every ounce of performance out of my chip that I kept (I returned the other one).

Soo here goes.

I originally had an Asus Strix z490 ITX board after a lot of people recommended it. It is a great board... but I gotta say... not the best for my situation. I was not able to get the best out of my 2 chips I had. I often needed much higher voltage than the MSI Unify z490 ITX board so I would highly suggest moving over to MSI Unify if you are not seeing much gains out of your 10th gen chip. I can only really speak to my 10700k chips though.

To save me from writing the same thing over and over, AC in my post is All Core. and all voltage I post will be VCORE as reported by HWINF64 unless stated otherwise.

I have a slew of stability testing I do and only move onto the next test if it passes the previous test with 0 crashes, BSODs and of course 0 HWINFO64 hardware errors:

Test 1: Prime95 custom, avx disabled with ffts 4k-128k at 1min each for 2 hours.
Test 2: Real Bench Stress Test for 2 hours on 16gb option.
Test 3. Hardcore Multitasking Simultaneous Test:
-Real Bench Benchmark (set to infinite run)
-Cinebench R23 (tests for AVX as well) multithread (custom option to run for 360 min)
-1 window of Chrome streaming a 4K Youtube video (usually earth 4k 7 hour video)
-Netflix App with 4k movie streaming
-1 website with constant data streaming (I use investing.com and click on the Crypto currency page, it is constantly streaming live data from tones of different exchanges).

I do this for 4 hours.

Test 4:
I run the exact same test as test 3 but this time for 8 hours to test a full workday of stability. I also turn on my webcam and leave that on a window by itself.

If it passes test 4 with zero hardware errors in HWINFO64 I certify it stable for daily use. I did this because I use to run each of the programs separately and would notice a random error throughout the day sometimes but all the individual tests would pass fine. It wasn't until I started doing Test 3 that those errors would reveal themselves later. So I highly suggest doing something similar if you want to truly test daily driver stability.


Anyways... now that you know what I consider stable... let's talk about the chips.


2 chips, both 10700k. One was made in China, the other Vietnam.


ASUS 5.1ghz all core at llc 4, ring at 46:

Chip 1 was stable 1.415v on the Asus board. Chip 2 was stable at 1.375v.
I wasn't satisfied so I purchased the MSI board, started testing both chips. I purchased it because it seemed like ASUS went cheap and had much less power stages. MSI had like 8 or 9 so I figured the power supplied would likely be much better.

MSI 5.1ghz all core at llc 3, ring 46:

Chip 1 stable at 1.36v, Chip 2 stable at 1.295v!!!! I was pretty much jumping after Chip 2 passed Test 4 and I saw 0 errors. I couldn't believe it so I re-ran the test again and by the afternoon, again 0 errors.

I decided to go for gold and pushed Chip 2 to 5.2ghz AC... after 2 days of testing, it was stable at 1.375v. I did need to drop LLC to 1 for it to be stable. I wasn't OK with that as a daily voltage as i am sure the transient spikes were well above 1.45v but largely unknown. I would get to ~1.39 vcore under load with LLC 1 though. I thought about running it at 1.395 vcore with LLC 3 (almost no droop) but it wouldn't pass. I tried 1.4cvore with LLC 3 and it did but I am not a fan of high vcore, I'd rather have lower voltage during off-load times and slightly higher voltage under load. I know this might be different for some others but... he do you.

I decided I didn't like the 1.375v at 5.2 AC as a daily so I decided to tune my 5.1ghz OC.

I ended up with this setup setup for multipliers.

Core 0 - x53
Core 1 - x53
Core 2 - x51 (weakest core)
Core 3 - x52
Core 4 - x52
Core 5 - x53
Core 6 - x53
Core 7 - x53

1.320-1.325vcore idle LLC1 1.344vcore under load.
1.3v VCCIO
1.3v VCCSA

Under load, the system will effectively only do 5.1 AC but while not under load, you consistently get the cores hitting their max almost ever 2-3 seconds.

I also need to mention, I was able to get my vcore lower by disabling EIST, C-states, Speed Step. I do have Turbo on.

I did not enter an AVX offset as I don't do anything that really uses AVX hardcore. And Cinebench R23 does test for medium AVX usage so, good enough for me.

I am running all of this in my NZXT H1 with the stock 140mm AIO, temps don't get above 90c on the hottest core. Package temps under load at 88c. Idle, they drop down to 25c. This case is small and runs fairly hot so I am impressed with it.

I am thinking of deliding the 10700k, it really is a great piece of silicon that I got so I really want to try and push it to 5.2 AC at something much lower. I tested my system on a cold night where temps were in the single digits 5c... got the system to be stable at 1.335vcore on LLC 3, temps never got above 81 sitting outside in that cold air, so if I could get my temps down 7c-10c I think I could really push this to 5.2ghz AC at a comfortable vcore slightly higher than my current setup.

Let me know thoughts and if you have any questions!
Click to expand...
Can you confirm the better performing chip was the one made in Vietnam? i have just received mine and it is made in Vietnam with the Marvel packaging, hoping i got lucky!
 
Overclock.net
posts
28M
members
562K
Since
2004
The premier forum for overclocking experts and enthusiasts. Discuss hardware optimization, custom builds, benchmarking, cooling solutions, and pushing the boundaries of computing performance. From beginner guides to extreme overclocking, join our technical community to master system tuning.
Full Forum Listing
Explore Our Forums
Intel CPUs Hardware News NVIDIA PC Gaming Water Cooling