Is overclocking over?

Soldato
Joined
19 Nov 2015
Posts
4,905
Location
Glasgow Area
I notice that it would appear that Kaby lake is simply just overclocked skylake.
Reducing the overclocking headroom on kaby lake itself.

Pascal cards are overclocked out of the box, judging by the lower overclock % achievable compared to maxwell.

Is this a trend? You used to be able to get huge overclocks from CPUs and GPUs. Is consumer overclocking done for?
 
I notice that it would appear that Kaby lake is simply just overclocked skylake.
Reducing the overclocking headroom on kaby lake itself.

Pascal cards are overclocked out of the box, judging by the lower overclock % achievable compared to maxwell.

Is this a trend? You used to be able to get huge overclocks from CPUs and GPUs. Is consumer overclocking done for?

Someone more knowledgeable may correct me, but I'm guessing a lot has to do with the maturity of processes and the experience of the engineers and the industry as a whole. Standardised clock speeds were based on the low best, what the majority of chips can expect to achieve to improve yield and reduce wastage. Peromance is probably considerably more consistent these days, meaning that production runs yield chips that can achieve closer to their theoretical max before being overclocked. Boost technologies also fill in this middle ground, essentially allowing these devices to over lock themselves.

I'm on a 4790K and can get another 500Mhz out of my chip with decent temps and no concerns, that's a 12.5% increase, so I consider that worthwhile. No experience with Skylake or Kabylake though.

Intel has no incentive to innovate when they occupy so much of this space, so let's hope Zen shakes things up.
 
Last edited:
I've always thought of boost speed as the actual rated speed of the chip, when it's not at boost speeds its essentially underclocking itself, which is fine for efficiency but boost is not overclocked, its just a marketing term.

If a chip is running within manufacturers specified spec, it is by definition not overclocked, overclocked is when you run it faster than the manufacturer specifications.

As wanton said, improvements in manufacturing process and design mean a much more consistent higher yield end product, so you dont get chip's with buckets of untapped potential sneaking through the net so much.
 
Last edited:
The glory days of overclocking are long gone. Back then we could get up to double a cpu's clockspeed which I did with a E2160 and was mighty chuffed about. I also had 3.8Ghz out of a 2.4Ghz Q6600 and 2.4Ghz E6600, 3.4Ghz out of a 1.8Ghz E4300, 4.5Ghz out of a 3.16Ghz E8500 and 4.13Ghz out of a 2.83Ghz Q9550. Haswell was a poor clocker, Skylake even more so and Kabylake seems to be no better. Even if all 7700k's hit 5Ghz (which they won't) that's only 800mhz over the base clock and only 500mhz over the turbo clock and that's just pathetic. Then they have the nerve to charge extra for a unlocked platform with very little overclocking left in it!!

The GPU side is even worse. I haven't run a GPU overclocked since my awesome AMD 5850 which could do 1Ghz on the core and 1.2Ghz on the memory. Since then overclocking headroom has been so small that I haven't even bothered. My current 1070 will do 2152Mhz on the core and 9914Mhz on the memory but it makes so little difference I just run it at the 2012/8208mhz it does out of the box.
 
Last edited:
I was initially chuffed when I overclocked my i5 6600k to 4.5ghz so easily.

Then I did the maths, that's a 15 percent oc. Im not going to try and squeeze another 200mhz out of it because I probably won't notice the difference for the effort involved.

Coming from an oc Q6600 at 3.3 ghz, almost 40 percent oc it doesn't sound quite so exciting.
 
The glory days of overclocking are long gone. Back then we could get up to double a cpu's clockspeed which I did with a E2160 and was mighty chuffed about. I also had 3.8Ghz out of a 2.4Ghz Q6600 and 2.4Ghz E6600, 3.4Ghz out of a 1.8Ghz E4300, 4.5Ghz out of a 3.16Ghz E8500 and 4.13Ghz out of a 2.83Ghz Q9550. Haswell was a poor clocker, Skylake even more so and Kabylake seems to be no better. Even if all 7700k's hit 5Ghz (which they won't) that's only 800mhz over the base clock and only 500mhz over the turbo clock and that's just pathetic. Then they have the nerve to charge extra for a unlocked platform with very little overclocking left in it!!

The GPU side is even worse. I haven't run a GPU overclocked since my awesome AMD 5850 which could do 1Ghz on the core and 1.2Ghz on the memory. Since then overclocking headroom has been so small that I haven't even bothered. My current 1070 will do 2152Mhz on the core and 9914Mhz on the memory but it makes so little difference I just run it at the 2012/8208mhz it does out of the box.


+1

IMO the purpose of overclocking was to take a cheaper variant of the microprocessor and make it perform close to or exceed the more expensive one. Overcoming the locked down aspects of the CPU by modding, pins or sockets or just by applying more voltage and better (if noisier) cooling.

Today you have 'permissible overclocking' where the manufacturer allows a certain amount of overclocking if your pocket is deep enough to afford the K variant, and decent ancillaries. An interesting development recently was the bios mod to allow certain processors to be overclocked without the manufacturers permission this was not widely pursued but it appealed to the anarchic overclocker in me.

It is a lot easier to overclock with the cooling and voltages possible now, but hardware is far more competent today compared to software improvements. Unlike 10 or 20 years ago when software was driving the need for faster better hardware.
 
I suspect its because as they improved the safety features to protect the chips they felt they could run the chips closer to their limit at stock speed.
 
I suspect its because as they improved the safety features to protect the chips they felt they could run the chips closer to their limit at stock speed.

That is possibly true, but I suspect many non K variants could approach K frequencies with the right overclocking techniques.
 
That is possibly true, but I suspect many non K variants could approach K frequencies with the right overclocking techniques.

Yeah probably, i dont know how they select the chips, maybe they come from the same place in the wafer and they just lock and under clock them to fill another gap in the market. I suspect the only speed you can consider as intels stock speed is the top of each range.
 
All the guys here are far to back in the day, the reason why everything has slowed down is that it's become harder to make progress so all the manufactures are now just rinsing every last mhz out, with mobile computing becoming more popular it's low power most consumers want.

If you look at most things now it appears to be mobile chips are made then desktop is a afterthought, hopefully mobile will top out soon.
 
Possibly a point..More physical cores and hyperthreads is probably an easier route to more power at this point, but software that can take full advantage of 8 or more cpu cores is still a fairly uncommon beast aside more specialist requirements.

Or maybe more gpu cores, but again it needs software to utilise them.
 
My system is always at stock and only overclocked for FSX, this is how I have always used my systems and even for regular gaming I keep it stock as the difference is not that huge, but FSX does benefit as it is very CPU dependent. If a game I really like does benefit I would keep it overclocked for that, but a lot of the games I like are GPU dependent. So the new CPUs that have out of the box high clocks is a good thing, but as others have said overclocking is not what it was like in the past, the idea of buying a cheap CPU and getting the most out of it. Today is a different story as to overclock on an Intel system requires a K chip and a Z chipset so the idea of cheap overclocking as in the past is slowly becoming over with in that way and a more expensive system from day one.

It was after the Q6600 I stopped keeping my systems always overclocked, I stopped at the Q9650 and would keep it at stock and as above only for FSX overclocked.

All systems before the Q6600 were always left at the best overclock I could get, but then overclocking really did make a nice difference to the overall system speeds, now it's not that huge and sometimes hardly noticeable in real world use, unless you are doing benchmarks or FPS peeping.
 
Last edited:
Possibly a point..More physical cores and hyperthreads is probably an easier route to more power at this point, but software that can take full advantage of 8 or more cpu cores is still a fairly uncommon beast aside more specialist requirements.

Or maybe more gpu cores, but again it needs software to utilise them.

But more cores needs to become mainstream before it is utilized properly. I feel with Zen if it's cheap enough and Intel start putting more cores in most people will have 4 cores minimum with the average having 6 and "enthusiast" having 8, all it needs is for a higher adoption rate of higher cores.
 
I remember getting 4.5GHz out of my E8400, although it degraded a fair bit from all the volts it took to get it benchmark stable, 4.2 was its proper 24/7 stable clock. Back then it took more effort though, tweaking FSB speeds and NB volts. Now you just change a multiplier and tweak your volts and away you go. You just don't go that far anymore. I only really push a chip when I first get it to see what it can do, and then when it's getting a bit tired looking I go hell for leather and throw caution to the wind while I save up for its replacement.
 
I'm on a 4790K and can get another 500Mhz out of my chip with decent temps and no concerns, that's a 12.5% increase, so I consider that worthwhile. No experience with Skylake or Kabylake though.

Intel has no incentive to innovate when they occupy so much of this space, so let's hope Zen shakes things up.

Your sig says you're at 4.4. Are you saying you can do 4.9 but choose not to?
 
Thats rather normal, some people would prefer quieter fans and less power consumption everyday use. Its nice to know that your CPU can overclock in the future when it starts dragging behind a little rather like an early i7 would now without a significant overclock.
 
I think it depends on your budget and expectations.

If your on a small budget then there are gains to be had.
Overclocked non K skylake is a ok deal, but it's not amazing.

These days you pay extra to overclock, it's all about paying extra for K chips or £500 motherboards. If your interested in having the fastest pc and loads of overclocking toys things have never been better.

I had more fun overclocking durons and figuring out my own vmod and making a cheap chip twice as fast.
Now I just buy something very fast and make it a little bit faster.
 
It seems like modern cpu's are no faster than 5 year old ones, my cheap i5 3570k is running at 4.8GHz and there's no option to buy anything significantly faster than that today, other than very expensive socket 2011 cpu's and even then there's only a benefit in highly multi threaded applications like encoding. I want more performance but I can't get it, technology has come to a standstill!
 
I overclocked my E8400 to see real gains in TF2 but haven't seen any tangible benefit at all with my i5 overclocked. I prefer silence these days.
 
It seems like modern cpu's are no faster than 5 year old ones, my cheap i5 3570k is running at 4.8GHz and there's no option to buy anything significantly faster than that today, other than very expensive socket 2011 cpu's and even then there's only a benefit in highly multi threaded applications like encoding. I want more performance but I can't get it, technology has come to a standstill!

What a great CPU this is.. GHZ overclocking and now dirt cheap second hand. I stupidly bought a Z170 DDR4 with an i7 6700k.. I've noticed a significant performance leap when exporting in PS but nothing else.. The oc'ing ratio is a lot further behind the 3570 and performance is negligible.
 
Back
Top Bottom