Question for 8700k owners: is 5.3ghz possible on air cooling?

[Indus]

I'm curious.. how many 8700k will hit 5.3ghz or 5.2ghz on air cooling? 212 evo or noctua or something similar.

Is it even possible?

What are the chances?

 

[daveybrat]

On good air cooling i think most will hit 5GHz easily at a reasonable voltage. You can hit higher but without delidding or raising the voltage a lot i'd stick with 5GHz.

 

[UsandThem]

No, the 212 EVO isn't for a heavily overclocked CPU. A person would want a much beefier air cooler (or liquid cooling).

https://techreport.com/blog/32661/just-how-hot-is-coffee-lake

That result still suggests a Hyper 212 Evo-class cooler probably isn't sufficient for holding the overclocked i7-8700K in check, given how little headroom it offers.

5.0 Ghz would probably be as high as the 212 EVO would safely take a person, and I don't think if it was my CPU, I'd let it run that warm. According to Silicone Lottery, 22% of the CPUs they test hit 5.2 Ghz, but that's also with them being delidded.

https://siliconlottery.com/collections/coffeelake/products/8700k52g?variant=47048295116

 

[ehume]

I tested out a bunch of heatsinks in a review for Overclockers.com. I included only six-heatpipe heatsinks: the Megahalems, the Genesis (both the Prolimatech heatsinks were tested with two Blue Vortex 14cm fans), and Noctua NH-D14, NH-U14S and the NH-D15. On LinX 0.6.5 they all averaged in the 80's or lower, spiking 88C to 99C -- I was using Linpack on 12 threads with AVX2. Given all of that -- I was admittedly pushing these heatsinks, trying to break them -- with a normal load I could expect to push my chip a little, maybe even to 5.1 or 5.2 GHz. But I was using 1.37 Vcore, which is high. When I backed off a little I could back off on the Vcore a whole lot, which suggests that at 5 GHz, this chip is probably looking at the top of its range.

If you are thinking of going to 5.2 or 5.3 GHz, I wouldn't go there. Maybe in a year the process will be improved enough, but you can see that air cooling -- and these are the best -- is at its bleeding edge doing 5 GHz right now. Maybe with a golden chip you will do better, but what are the odds of that?

 

[Indus]

Thanks ehume for a detailed answer.

I jumped on Ryzen because it ran cooler and it does give me 120-144 fps in pubg np, but I was still curious about the other side since I have a 165hz monitor now.

However I am not going to do water cooling.. as I don't consider it safe. I'll probably wait till the process has matured to a 60-70C temp and not 88-99C.

 

[IEC]

Thanks ehume for a detailed answer.

I jumped on Ryzen because it ran cooler and it does give me 120-144 fps in pubg np, but I was still curious about the other side since I have a 165hz monitor now.

However I am not going to do water cooling.. as I don't consider it safe. I'll probably wait till the process has matured to a 60-70C temp and not 88-99C.

I am getting 80C running BOINC on a 4.7GHz all core overclock at -50mV offset, no AVX offset on a decent air cooler (Scythe Fuma Rev. B with push/pull fans). I also have 6x 120mm Scythe Gentle Typhoon AP14 fans at full blast in a Corsair Air 540 case = insane airflow. Without delidding, forget 5GHz, much less higher than that.

 

[TheELF]

Thanks ehume for a detailed answer.

I jumped on Ryzen because it ran cooler and it does give me 120-144 fps in pubg np, but I was still curious about the other side since I have a 165hz monitor now.

However I am not going to do water cooling.. as I don't consider it safe. I'll probably wait till the process has matured to a 60-70C temp and not 88-99C.

You do realize that intel AT THE SAME CLOCKS performs at least 10% better right?
So at ~4Ghz it would outperform ryzen and be super cool.
In fact even the 3.6Ghz quad core without HT, i3-8100 is about 10% faster then any ryzen,a 8700k would be running in idle and very cool and still outperform any ryzen...
https://www.techspot.com/article/1532-pubg-cpu-benchmarks/

 

[Indus]

https://www.techspot.com/article/1532-pubg-cpu-benchmarks/

I don't think that graph is accurate anymore since I'm actually playing 1440p and getting the 102 min/ 120-144 average with my ryzen and I too play on Very Low.

It could also be that my overclocked video card is pushing it 120-144.. but whatever the reason.. I'm definitely not averaging 100 which the graph said but I was curious since I'm not averaging 165 either which I would like.

 

[ehume]

Indus is right to point at the high spikes, but that was with Linpack + AVX2. I think that 80C represents a more typical heavy load. As for the wonderful Fuma (we use one here) it is probably enough for regular loads, but I'd recommend holding your OC to 4.9 GHz.

 

[IEC]

I don't think that graph is accurate anymore since I'm actually playing 1440p and getting the 102 min/ 120-144 average with my ryzen and I too play on Very Low.

It could also be that my overclocked video card is pushing it 120-144.. but whatever the reason.. I'm definitely not averaging 100 which the graph said but I was curious since I'm not averaging 165 either which I would like.

Considering how much has been changed in PUBG since December, those benchmarks are out of date. It's also the worst game to use as a benchmark because it uniformly runs worse than it should for the visuals whether you have a GTX 1050 or a GTX 1080 Ti.

I notice exactly ZERO difference while playing PUBG between my 8700K and my 1800X at 1440p and 4K.

 

[TheELF]

I notice exactly ZERO difference while playing PUBG between my 8700K and my 1800X at 1440p and 4K.

Can you provide us with usage percentage for both? Does the 8700k even turbo up?

 

[IEC]

Can you provide us with usage percentage for both? Does the 8700k even turbo up?

My 8700K is running at 4.7GHz all core, no AVX offset, -50mV vCore offset (OC'd, BOINC stable). It never goes below 4700 MHz.

 

[BonzaiDuck]

I tested out a bunch of heatsinks in a review for Overclockers.com. I included only six-heatpipe heatsinks: the Megahalems, the Genesis (both the Prolimatech heatsinks were tested with two Blue Vortex 14cm fans), and Noctua NH-D14, NH-U14S and the NH-D15. On LinX 0.6.5 they all averaged in the 80's or lower, spiking 88C to 99C -- I was using Linpack on 12 threads with AVX2. Given all of that -- I was admittedly pushing these heatsinks, trying to break them -- with a normal load I could expect to push my chip a little, maybe even to 5.1 or 5.2 GHz. But I was using 1.37 Vcore, which is high. When I backed off a little I could back off on the Vcore a whole lot, which suggests that at 5 GHz, this chip is probably looking at the top of its range.

If you are thinking of going to 5.2 or 5.3 GHz, I wouldn't go there. Maybe in a year the process will be improved enough, but you can see that air cooling -- and these are the best -- is at its bleeding edge doing 5 GHz right now. Maybe with a golden chip you will do better, but what are the odds of that?

I may garner only a minority consensus of opinion, but I've decided for myself. For any Skylake, Kaby Lake or Coffee Lake that isn't an "X" series chip, I wouldn't hesitate to either send the CPU to Silicon Lottery or buy a binned chip from SL and have them re-lid it with CLU.

Last time around, the $50 re-lid service bought me 12C-worth of temperature reduction.

I'd do this whether I choose air or water. If you choose air, you should be able to match many dual-fan AiO water-coolers applied to a stock, unmolested retail CPU. If you choose water, you could simply tell them to delid the processor and send the IHS to you without reinstalling it. Then -- be careful how you tighten the fittings for the waterblock. In that case, your temperature improvement should be profound.

My i7-6700K was such a re-lidded chip, and it's been running almost non-stop for good part of 18 months with a ThermalRight LG Macho cooler.

Personally, for these chips, I'd consider the $50 re-lid service as a mandatory preliminary step. You could also buy the de-lid tool and do it yourself. My sense of it regarding Silicon Lottery: they have performed this step thousands of times, and they know how to get the gap between IHS and the die just right. And they'll test the result to report the temperature improvement before sending it to you.

 

[IllogicalGlory]

You do realize that intel AT THE SAME CLOCKS performs at least 10% better right?
So at ~4Ghz it would outperform ryzen and be super cool.
In fact even the 3.6Ghz quad core without HT, i3-8100 is about 10% faster then any ryzen,a 8700k would be running in idle and very cool and still outperform any ryzen...
https://www.techspot.com/article/1532-pubg-cpu-benchmarks/

What an earth-shattering difference. What exactly offended you so much about the post you quoted? It never said Ryzen was equal or faster, in fact, implicit in that post was the conceit that the 8700K was faster. They were just wondering if it was worth upgrading to.

Anyway, in my experience, without delidding you aren't guaranteed even 4.7GHz on an 8700K. Don't expect more than 4.9GHz with a delid. Certainly some can do more, but not all will.

 

[Yuriman]

Delidding isn't that hard. You can get kits on eBay for use in a vice, though I still prefer to do it with a thin razor blade. Takes about 5 minutes.

 

[IRobot23]

You do realize that intel AT THE SAME CLOCKS performs at least 10% better right?
So at ~4Ghz it would outperform ryzen and be super cool.
In fact even the 3.6Ghz quad core without HT, i3-8100 is about 10% faster then any ryzen,a 8700k would be running in idle and very cool and still outperform any ryzen...
https://www.techspot.com/article/1532-pubg-cpu-benchmarks/

before you link his benchmarks again check adoredTv and what he discovered. End conclusion was actually useless, cause there is not a lot AMD fanboys.



You cannot use the word fanboys here. This is now twice.


esquared
Anandtech Forum Director

 

[TheELF]

What an earth-shattering difference. What exactly offended you so much about the post you quoted? It never said Ryzen was equal or faster, in fact, implicit in that post was the conceit that the 8700K was faster. They were just wondering if it was worth upgrading to.

Anyway, in my experience, without delidding you aren't guaranteed even 4.7GHz on an 8700K. Don't expect more than 4.9GHz with a delid. Certainly some can do more, but not all will.

Did you even read the right post?
Instead of running the 8700k at stock or even underclocking it to ~4Ghz (both of which options would give him a cool and fast(er) CPU), his logic dictated that he better get a slower CPU because it runs cooler...

Thanks ehume for a detailed answer.

I jumped on Ryzen because it ran cooler and it does give me 120-144 fps in pubg np, but I was still curious about the other side since I have a 165hz monitor now.

However I am not going to do water cooling.. as I don't consider it safe. I'll probably wait till the process has matured to a 60-70C temp and not 88-99C.

 

[Ranulf]

Considering how much has been changed in PUBG since December, those benchmarks are out of date. It's also the worst game to use as a benchmark because it uniformly runs worse than it should for the visuals whether you have a GTX 1050 or a GTX 1080 Ti.

I notice exactly ZERO difference while playing PUBG between my 8700K and my 1800X at 1440p and 4K.

The keyword when using PUBG for benchmarks seems to be unoptimized. Much like Starcraft2 which has been used off an on now for almost 8 years as a cpu benchmark. Flawed, to say the least.

 

[IllogicalGlory]

Did you even read the right post?
Instead of running the 8700k at stock or even underclocking it to ~4Ghz (both of which options would give him a cool and fast(er) CPU), his logic dictated that he better get a slower CPU because it runs cooler...

That was not what was said, he said he has a Ryzen and was wondering if he should get an 8700K to get a few more FPS. Are you seriously arguing that it's a no-brainer to get rid of it and drop $500 on a Z370 setup for an extra 15 FPS?

If you're just calling him an idiot for getting a Ryzen CPU, that's not particularly helpful, plus he may have purchased it before the 8700K came out.

 

[Indus]

That was not what was said, he said he has a Ryzen and was wondering if he should get an 8700K to get a few more FPS. Are you seriously arguing that it's a no-brainer to get rid of it and drop $500 on a Z370 setup for an extra 15 FPS?

If you're just calling him an idiot for getting a Ryzen CPU, that's not particularly helpful, plus he may have purchased it before the 8700K came out.

Thanks.. I did buy the ryzen upon launch.

 

[Indus]

The keyword when using PUBG for benchmarks seems to be unoptimized. Much like Starcraft2 which has been used off an on now for almost 8 years as a cpu benchmark. Flawed, to say the least.

Yeah I can definitely agree with that.. since I see Shroud complaining about 30fps with his 8700k @ 1080p but I get better than that at 1440p with my Ryzen.

 

[TheELF]

That was not what was said, he said he has a Ryzen and was wondering if he should get an 8700K to get a few more FPS. Are you seriously arguing that it's a no-brainer to get rid of it and drop $500 on a Z370 setup for an extra 15 FPS?

If you're just calling him an idiot for getting a Ryzen CPU, that's not particularly helpful, plus he may have purchased it before the 8700K came out.

No, what he said was I got ryzen because intel is too hot,how can you tell from that if he got ryzen at launch as he said right now or if he got it after realizing that intel is "too hot" ?



Indus already stated he'd had the processor since launch but you just had to keep on arguing about it.

AT Moderator ElFenix

 

[Indus]

No, what he said was I got ryzen because intel is too hot,how can you tell from that if he got ryzen at launch as he said right now or if he got it after realizing that intel is "too hot" ?

Look in my mind as far as you're concerned, I will NEVER listen to a fanboy like you.

You have quoted nothing that is factual.

I'll just put your fake propoganda on ignore.




Can't believe we need two warnings about calling others fanboys in one thread.

AT Moderator ElFenix

 

[TheELF]

Look in my mind as far as you're concerned, I will NEVER listen to a fanboy like you.

You have quoted nothing that is factual.

I'll just put your fake propoganda on ignore.

Well, the only things I quoted where posts from this topic so...yeah, maybe you're right about that one.
Also trying to pretend that only a 8700k at well over 5Ghz is any alternative to ryzen makes you seem terribly biased and not me.

 

[Indus]

I'm biased towards price vs performance. If something is $180 vs $195 and that 195 fetches me 5-8% more performance, I'll buy it. But I'm not spending $800 ($350 cpu, $100 cooler, $150 motherboard, $200 memory) for the same amount.