Question Is OCing this 10700K even worth it for gaming?

moonbogg

Lifer
Jan 8, 2011
10,635
3,095
136
I bought it for better VR performance (which is gave) and most of those games are all janky and seem to run on a single core. The 10700K already boosts to 5-5.1Ghz for lightly threaded stuff anyway, so why the heck am I going to pump 1.4v into this thing trying to get all cores at 5-5.1? It looks like Intel already maxed these things out and only left 300Mhz on the table max for an all core OC. Sandy Bridge this is not. At least it's still fast though.
Another thing, I am doubting most of the posts I see about people bragging about their all core 5.1 OC with something ridiculous like 1.25v. They all talk like that and I think they're all full of crap.
 
  • Like
Reactions: Tlh97 and Gideon

ubern00b1

Member
Jul 6, 2017
42
35
91
You answered your own question with your first sentence...
I bought it for better VR performance (which is gave) and most of those games are all janky and seem to run on a single core
no need to overclock it balls to the wall if it satisfies your needs

As for the 2nd part of your question? who knows, maybe what is omitted is that they also have a $400+ z490 motherboard and a $200+ 240/360 AIO or better compared to an entry level $150 board and $50 air cooler? I mean, there's lots of variables in that comment which you can't just put down to vcore alone :smirk:
 

CP5670

Diamond Member
Jun 24, 2004
5,508
586
126
I set mine to 5ghz on all cores, which can often be achieved on stock volts (1.295v for me). By default, it may only boost to 4.7ghz since there is sometimes very minor usage on the other threads, enough to prevent it from boosting. Going from 5ghz to 5.1ghz requires a lot more volts and is not worth it.
 

scannall

Golden Member
Jan 1, 2012
1,944
1,638
136
At this point in time there really is no reason to overclock, except to sate a tinkerer's appetite. You won't see any perceptible difference on screen, in game etc. Just makes numbers a bit bigger in benchmarks. Quite a bit more meat in tuning your ram.
 

moonbogg

Lifer
Jan 8, 2011
10,635
3,095
136
I have more stability at 5.1 now by just leaving all the voltages on auto and letting it shove 1.39v through the thing. Had to turn the ring down from 48 to 47 so far. If it takes much more effort than this I'll just go back to stock, although I know what was said above about the chip turbo limiting to 4.7 if any additional activity is present, and let's be honest here; I can't have that. I won't stand for it and neither should YOU. Use case be damned. This chip is going to the moon and that's it. I mind up my made.

Also, whoever suggested I had a $50 cooler and potato board needs to understand that I spent $1600 on a triple rad water cooling rig. I will FRY this stupid chip into dust to get the most out of this loop, and I'll send this $270 board straight to hell along with it. Remember it.
 

BonzaiDuck

Lifer
Jun 30, 2004
15,699
1,448
126
I continue to OC my processors as an addiction, which arises and flourishes after a few rock-solid successes with earlier processor models.

And I'm still going to do it. But it doesn't pay back well for the effort. With NVME and Optane, in addition to greater total RAM, all the bottlenecks widen. A few hundred Mhz doesn't really mean diddly-squat, at that point.
 

Borealis7

Platinum Member
Oct 19, 2006
2,914
205
106
my 7700K, which i bought Jan 1st 2017, is the last K processor i will ever buy. OCing does not matter for games these days, better put your money on a better GPU.
 
  • Like
Reactions: will889

JoeRambo

Golden Member
Jun 13, 2013
1,814
2,105
136
I think overclocking as we know it from Celeron 300A days is gone for good. But that does not mean enthusiasts don't have what to do with their systems.

For any enthusiast Intel system the following need to happen:

1) All core OC to some sensible ( fixed ) clock at sensible voltage, you don't need/want to push things, just for example flat 5Ghz at sub 1.3V with reasonable vdroop.
2) Fixing uncore clocks to also sensible value of 4.X ghz and removing uncore downclocking by fixing low/high clocks in bios.
3) Removing CPU downclocking due to "power saving". You cannot afford to let your CPU take 10us to reach 5ghz from 1.2ghz, as that is millions of clocks lost, so High performance power plan and 100% minimal CPU clock is in order + sometimes in BIOS one needs to twiddle settings.
4) {3} does not mean skimping on power saving like C states, you just need to disable downclocking. It should show your "5"ghz all the time on all loads

5) The most important nowadays is memory tuning. Sky is the limit here and to have snappy system you need low latency. Obviously each memory access will miss at least once and despite naming shenigans by certain vendor caches are not infinite so even more memory misses will happen.
Shaving ~20% of latency is not uncommon giving how average system is set up by going from XMP to manually tuned and OCed memory. 45+ to ~36ns or so is not uncommon going to DDR4 4000Cl16ish OCs with handtuned secondaries and tertiaries.
But obviuosly this is the most time consuming activity and most risky as well and with CPU oc you gain 100mhz by tuning most multiplier and several voltages, where with memory you have dozens of "inputs" and there always looms a risk of destroying your Windows installation if you push things too far.
 
  • Like
  • Love
Reactions: BadThad and MangoX

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,686
136
Another thing, I am doubting most of the posts I see about people bragging about their all core 5.1 OC with something ridiculous like 1.25v. They all talk like that and I think they're all full of crap.
From Silicon Lottery
10700K @ 5.1GHz up to 8 cores, 5.2GHz up to 4 cores.
Vcore under heavy load @ 1.400V BIOS, 1.320V Socket Sense, 1.230V Die Sense

As of 8/18/20, the top 22% of tested 10700K fell within our 5.1GHz bin or greater.
 

CP5670

Diamond Member
Jun 24, 2004
5,508
586
126
5) The most important nowadays is memory tuning. Sky is the limit here and to have snappy system you need low latency. Obviously each memory access will miss at least once and despite naming shenigans by certain vendor caches are not infinite so even more memory misses will happen.
Shaving ~20% of latency is not uncommon giving how average system is set up by going from XMP to manually tuned and OCed memory. 45+ to ~36ns or so is not uncommon going to DDR4 4000Cl16ish OCs with handtuned secondaries and tertiaries.
But obviuosly this is the most time consuming activity and most risky as well and with CPU oc you gain 100mhz by tuning most multiplier and several voltages, where with memory you have dozens of "inputs" and there always looms a risk of destroying your Windows installation if you push things too far.

Does this really make any noticeable difference either? I read some articles that beyond 3200 on Intel and 3600 on AMD the difference is minimal.

I think the only OC that matters today (like you would actually notice it outside benchmarks) is the video card, and even that can take up gobs of extra power.
 

JoeRambo

Golden Member
Jun 13, 2013
1,814
2,105
136
Does this really make any noticeable difference either? I read some articles that beyond 3200 on Intel and 3600 on AMD the difference is minimal.

It depends where You look, for example scaling never ends in:

And it is not "handtuned" either, there is huge difference between XMP settings and manually tuned secondaries and tertiaries.

Some bencharks like Linpack continue to scale to infinity with memory tuning, Not uncommon to go from 350GFlops @5ghz to 500Gflops on 9900K just by tightening timings. And some like Cinebench do not care at all.
But to me the most important thing is smoothness, i even went so far to disable HT to avoid scheduler wasting my time, 10C of 5.1Ghz Skylake is sure damn plenty for desktop machine.
 
  • Like
Reactions: maluckey1

maluckey1

Senior member
Mar 15, 2018
331
144
86
It depends where You look, for example scaling never ends in:

And it is not "handtuned" either, there is huge difference between XMP settings and manually tuned secondaries and tertiaries.

Some bencharks like Linpack continue to scale to infinity with memory tuning, Not uncommon to go from 350GFlops @5ghz to 500Gflops on 9900K just by tightening timings. And some like Cinebench do not care at all.
But to me the most important thing is smoothness, i even went so far to disable HT to avoid scheduler wasting my time, 10C of 5.1Ghz Skylake is sure damn plenty for desktop machine.

Agreed, the 10900k and 10700k aren't really bottlenecks (at least for for now). Because of the persistent urban legend that RAM speed doesn't matter, many people still believe that anything more than 3600 is a waste....it's NOT. I went from 3600mhz @ 18-22-22-42-64 to 4133mhz with the same timings and raised FPS by 10 percent, with the absolute minimum frame rising almost 15 percent. In comparison, raising all-core clock to 5.1ghz from 4.4ghz gave me 2 FPS with my 10700k, and minimums were about 1 FPS better.

M
 

CP5670

Diamond Member
Jun 24, 2004
5,508
586
126
What games or settings did you see that in? 15% is quite significant, but most of the benchmarks on this are done at unrealistic settings (like 1080p on a top GPU in that article above), so it's hard to tell what you would actually notice in a real situation. It only matters if it's below the monitor's refresh rate.
 

moonbogg

Lifer
Jan 8, 2011
10,635
3,095
136
I tried OCing based on core loading. So far at only 1.35v fixed I have:

If 4 cores loaded, they do 5.2Ghz
If 5 or 6 cores loaded, they do 5.1Ghz
If 7 or 8 cores loaded, they do 5.0Ghz.

Seems like a good balance and makes sense that this could work better than trying to force all cores under XTREME LOAD to 5.1Ghz. Only 10 runs of Intel Burn Test at very high and some cinebench and gaming so far, so we'll see how it goes. Fixed has given me way more stability at less voltage than anything else. I'm not used to that. I always go dynamic, but this time may be different.

EDIT: Does fixed Vcore wreck a CPU faster than dynamic if voltage is the same? I've heard yes. I've heard no. I've heard it's more current dependent as far as electro-migration and degradation goes. I don't know.
 
Last edited:

AdamK47

Lifer
Oct 9, 1999
15,169
2,830
126
Agreed, the 10900k and 10700k aren't really bottlenecks (at least for for now). Because of the persistent urban legend that RAM speed doesn't matter, many people still believe that anything more than 3600 is a waste....it's NOT. I went from 3600mhz @ 18-22-22-42-64 to 4133mhz with the same timings and raised FPS by 10 percent, with the absolute minimum frame rising almost 15 percent. In comparison, raising all-core clock to 5.1ghz from 4.4ghz gave me 2 FPS with my 10700k, and minimums were about 1 FPS better.

M
I went from 3800 to 4400 memory and the difference is certainly noticable with my 10900K.
 

maluckey1

Senior member
Mar 15, 2018
331
144
86
I tried OCing based on core loading. So far at only 1.35v fixed I have:

If 4 cores loaded, they do 5.2Ghz
If 5 or 6 cores loaded, they do 5.1Ghz
If 7 or 8 cores loaded, they do 5.0Ghz.

Seems like a good balance and makes sense that this could work better than trying to force all cores under XTREME LOAD to 5.1Ghz. Only 10 runs of Intel Burn Test at very high and some cinebench and gaming so far, so we'll see how it goes. Fixed has given me way more stability at less voltage than anything else. I'm not used to that. I always go dynamic, but this time may be different.

EDIT: Does fixed Vcore wreck a CPU faster than dynamic if voltage is the same? I've heard yes. I've heard no. I've heard it's more current dependent as far as electro-migration and degradation goes. I don't know.

It's more the temperature + voltage that kills CPU cores (not just one or the other). On a 10900k, 1.35v isn't going to affect anything if you're running cool. Pick what frequency/voltage/temperature works best for you (cooler is better) and then fiddle around with ring ratio. Some games like fast ratio, while some don't care. It depends on how much is going on in the game, versus your overall hardware setup and game settings

M
 
  • Like
Reactions: Tlh97 and moonbogg

maluckey1

Senior member
Mar 15, 2018
331
144
86
What games or settings did you see that in? 15% is quite significant, but most of the benchmarks on this are done at unrealistic settings (like 1080p on a top GPU in that article above), so it's hard to tell what you would actually notice in a real situation. It only matters if it's below the monitor's refresh rate.

You're right about below the refresh rate....I WISH this could hit 165 mhz. Fact is, I can't do that at 1440P gaming in most new games. Not with an RTX 2070

Cyberpunk with lots of eye candy at 1440p has a LOT going on with RT, reflections and shadows set high. With my lowly (formerly glorious) EVGA RTX 2070 XC Ultra, even with DLSS on balanced this setup struggles to keep above 54-58 FPS. Crank the RAM throughput and reduce RAM latency (and to a lesser extent you can also raise the CPU ring ratio) and you can boost that to playable 58-68 FPS with absolute minimum FPS also rising with it.
 

CP5670

Diamond Member
Jun 24, 2004
5,508
586
126
Interesting, I might have to try out a memory OC or get new memory. I have some mid-tier G.skill Trident CL16 3200 memory that I just keep at stock. In Cyberpunk, I get 80-110fps at 4K with medium/high settings and no RT. For the most part, only older games get a flat 120fps at all times (the refresh rate).
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
You're right about below the refresh rate....I WISH this could hit 165 mhz. Fact is, I can't do that at 1440P gaming in most new games. Not with an RTX 2070

Cyberpunk with lots of eye candy at 1440p has a LOT going on with RT, reflections and shadows set high. With my lowly (formerly glorious) EVGA RTX 2070 XC Ultra, even with DLSS on balanced this setup struggles to keep above 54-58 FPS. Crank the RAM throughput and reduce RAM latency (and to a lesser extent you can also raise the CPU ring ratio) and you can boost that to playable 58-68 FPS with absolute minimum FPS also rising with it.
We're quite a long way from displays hitting 165 MHz refresh rate 🙄😶😶😶
 

maluckey1

Senior member
Mar 15, 2018
331
144
86
We're quite a long way from displays hitting 165 MHz refresh rate 🙄😶😶😶

The monitor is willing...but not with my GPU...even if I had an RTX 3090, the framerates would only be in the high 80's FPS with everything else in my system set the same. It's a shame that by the time I can actually find and afford an RTX 3080, it'll be 2023 and the card will be obsolete and the game a memory.

M
 

CP5670

Diamond Member
Jun 24, 2004
5,508
586
126
I doubt Cyberpunk will get a constant 165fps on any hardware for many years. The game is just not that well optimized and the framerate sometimes tanks for no reason, especially in indoor areas.
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
The monitor is willing...but not with my GPU...even if I had an RTX 3090, the framerates would only be in the high 80's FPS with everything else in my system set the same. It's a shame that by the time I can actually find and afford an RTX 3080, it'll be 2023 and the card will be obsolete and the game a memory.

M
I was just kidding because you wrote MHz instead of Hz :)
 

moonbogg

Lifer
Jan 8, 2011
10,635
3,095
136
One of the reasons I like VR right now is because it reminds me of the golden age of PC gaming when there was actually room for improvement in the critical areas of resolution, performance and basic game design. It allowed you to appreciate an upgrade because it delivered a new experience to you. Gaming on a flat panel lost that feeling pretty much as soon as I got a 1440p monitor. Screen door effect was basically gone and pixelated imagery was never coming back (unless it was a chosen art style). Diminishing returns were here and the last upgrade that meant anything substantial to my gaming experience was going from a Core 2 Duo to Sandy Bridge.
When I got this CPU, the gameplay in VR was massively smoother. The VR imagery is good but pixelated and it's hard to see enemies across the map at times because of the pixelation. Some people think this is bad, but I think it's good. When I put on my headset it doesn't feel like I'm playing in a futuristic environment, but instead feels like I stepped into a time machine that brought me back to an era that I never thought I'd experience again; a time when things were new and just getting started. It's the only place where gaming feels exciting again, new again, and has that quality of moving forward rather than sitting still.
 

CP5670

Diamond Member
Jun 24, 2004
5,508
586
126
My thoughts as well. VR feels new and fresh the way regular PC gaming did 20 years ago. The pixellation can be improved with more supersampling (if you wanted to), but needs a top video card to maintain 90/120fps across most games.
 

KentState

Diamond Member
Oct 19, 2001
8,397
393
126
The monitor is willing...but not with my GPU...even if I had an RTX 3090, the framerates would only be in the high 80's FPS with everything else in my system set the same. It's a shame that by the time I can actually find and afford an RTX 3080, it'll be 2023 and the card will be obsolete and the game a memory.

M

I actually built my HTPC around the idea of playing CP2077 on it but the game is such a disappointment that I haven't turned it on since the week of release. Even more frustrating was that an RTX3080 could not get the game to 60fps at ultra with DLSS enabled.