40 years ago, Intel released the 8086, arguably the most influential CPU since the dawn of modern x86 computing. This tiny CPU (by today’s standards) housed 29,000 transistors and would go on to shape the future not only for Intel, but for millions of people worldwide, literally defining desktop computing.
This year, Intel decided to celebrate four decades since the 8086’s introduction with the announcement of a limited-edition Core i7 8086K, and I for one couldn’t have been more excited about it. Not because I have fond memories of using the 8086 (I wasn’t even around when the 8086 was released), but simply because Intel decided to celebrate this occasion by giving a nod to overclockers – and at the same time, mark an important milestone for them with their first 5GHz CPU. They’ve gone from a company which at some point discouraged overclocking, to one that now actively supports it by designing their products around this small, but important demographic.
As you may know by now, the Core i7 8086K (8086K from herein) is similar (if not identical) to the Core i7 8700K. I wouldn’t say it’s the exact same CPU as the 8700K as I’ve no way of knowing that factually, but I do suspect based on the configuration and drop-in compatibility with boards, that it is (at the very least) a CPU that shares much in common with the 8700K.
The main highlight here is that when using single-thread applications, the 8086K is capable of a blistering 5GHz on that particular loaded core. That’s a milestone in itself, as it’s Intel’s highest-clocked CPU ever, even though it’s not the first 5GHz CPU. That honour was claimed by AMD several years ago with an unbelievably underwhelming CPU (unlike when they released the first 500MHz CPU, which was revolutionary at the time). That said, the 8086K, unlike AMD’s first 5GHz CPU, actually delivers the goods. In fact, it’s easily the fastest gaming CPU money can buy, as it simply walks all over anything else that’s available at present, regardless of core or thread count.
That shouldn’t be surprising, however, given that the Core i7 8700K was already the fastest gaming CPU on the market, so a higher-clocked model would obviously extend this lead over all other CPUs. So as much as that may be the highlight of this CPU, I’m more interested in what it’s capable of delivering under extreme cooling – and in a gaming context when overclocked. After all, it isn’t an 8086, but the 8086K – that K making all the difference in what one should expect from such a CPU.
Sensible speculation would suggest that these 8086K CPUs are specially screened 8700K dies, which perhaps have some minor enhancements in terms of the TIM (Thermal Interface Material) used, or feature some other minor changes. At this stage we don’t know any of those details, and I’m unsure if Intel will ever release them. What I do know is that, unlike with the Core i7 4790K from 2013, where there were some physical changes to the CPU and the confirmed use of a superior TIM, the 8086K seems to have had none of this treatment. Yes, the 8086K made its debut at the largest global PC trade show, but it was a rather hush release, with the most notable aspect confined to the sweepstakes (which have since ended). In celebration of 40 years since the 8086, Intel was giving away exactly 8086 of these K SKU CPUs, of which I hear there’ll only ever be 50,000 produced worldwide. That makes Intel’s giveaway the biggest ever in terms of dollar value for anything PC related. If you go by the estimated retail price, the 8086K CPU sweepstakes represented over $3.4 million in value. That’s some serious change, and a serious commitment from Intel.
Given that the Coffee Lake CPUs are generally good overclockers already, at the very least fairing no worse than the previous-generation Kaby Lake (seventh-generation Core i7) parts despite their increased complexity, the 8086K needed to offer something special. In a way, it is rather special, even though it’s not in the most obvious way. For the extreme overclockers, it’s the same as the 8700K for the most part – but for the more casual overclockers and gamers, it’s a milestone of sorts. That’s not to say it doesn’t have the minerals to reach the stratospheric frequencies that the other Coffee Lake CPUs can with liquid nitrogen. I was able to validate 7.2GHz on the 8086K, just 44Mhz shy of the reigning 8086K record (set by the one and only Der8auer), and I managed to run a number of benchmarks at 6,800MHz, so these 8086K CPUs are anything but slow. However, as you may or may not know, regular 8700K CPUs do often hit 6.7GHz and can go all the way to 6.9GHz (yes, they’re rare, but they do exist), with a recent 3DMark05 record broken by an 8700K chugging along at an incredible 7.1GHz (do keep in mind that this isn’t a multi-threaded benchmark though, so the load is limited to a few cores).
As exciting and as simple as it was to reach the 6.8GHz clock (courtesy of the MAXIMUS X APEX, which is still the finest Z370 board to date), I was and still am intrigued by the options the 8086K gives you in terms of everyday overclocking.
The Sandy Bridge CPUs (second-generation Core CPUs like the 2600K) set a very high bar many years ago, with nearly every single 2600K capable of 5GHz and beyond, using nothing but AIO and air coolers. Since then, we’ve been seeking this 5GHz mark with every generation, with little to no success. The truth is, even today, not every 8700K or 8600K can do 5GHz on a single core, let alone all cores. The 8086K increases the odds of this by some margin, if only because at least one of the cores is guaranteed to hit this magical number.
During testing, it turns out that you can truly run 5GHz on all cores (including AVX workloads) with some minor tuning. As a general rule (or common sensibility), I don’t operate any CPU above 1.35v using off-the-shelf AIO coolers (for power and cooling reasons), but with the 8086K, it turns out that you can run 5GHz on all cores with HT enabled, a 4.6GHz ring bus frequency, and a relatively high DRAM clock frequency of 4,133MHz at 1.325v (load voltage with LLC on the APEX is around 1.32v), which is a little below what I would consider the cut-off voltage for everyday use. Mind you, the default VID for this particular sample was 1.344v under load.
Needless to say, running such a configuration gives blistering game performance, and in fact it creeps eerily close to matching the competition’s eight-core parts in multi-threaded productivity apps, clawing away at the four-thread disadvantage through sheer clock frequency, lower DRAM latency and memory bandwidth. (Note that these are aspects of a CPU that can be easily examined by us end users to some degree using commonly available software. I’m by no means saying these are the only differences between these CPUs, but rather that this is how these differences manifest themselves to us.)
What about those who only use the CPU for gaming, for whom that 5GHz all-core overclock won’t be as impressive? To test for these individuals, I disabled hyper-threading and this is when things become really interesting, as the clock frequency climbed another 200MHz (AVX offset was set at 2 for a 5GHz clock when dealing with the relevant workloads) to 5.2GHz, which was then paired with a 5GHz ring bus frequency and the same DDR4 4,133MHz frequency.
As we saw at this year’s Computex, there are some 5GHz kits and boards (or rather, a board) out there which can reach this frequency as well. In light of this, it’s very possible to configure a system of 5GHz clocks all around using such a motherboard (this motherboard uses a different signal layout) and slightly different DRAM modules. In such a configuration the 8086K would be key, if only because at the very least it guarantees that the CPU can reach that magical 5GHz, and all that’s left is to use the appropriate DRAM.
Going back to this HT-disabled configuration of 5.2GHz on the ROG MAXIMUS X APEX, the gaming performance was further increased, resulting in the highest performance I’ve ever recorded where gaming is concerned – further cementing the 8086K as the undisputed leader of the CPU market in terms of gaming performance. It’s not a baseless claim either, as the benchmarks speak for themselves, scoring the 8086K a convincing lead in every game test performed, including a few not represented in the graphs.
As with all things, there are some caveats to this performance which you should know about. All overclocking and tuning was done on a de-lidded CPU. In fact, an interesting thing happened when this particular CPU was de-lidded. Prior to this, 5GHz on all cores under load in Cinebench R15 showed a peak core temperature of 74 °C. It’s definitely useable and well within safe operating temperatures. However, after the HIS was removed and the paste replaced with Thermal Grizzly Kryonaut, the same configuration had a peak temperature of 81 °C.
It’s puzzling indeed. This, to me, could either mean that Intel is literally using a different TIM for the 8086K when compared to the 8700K, or that my application of the thermal paste was appalling. I’m leaning towards the latter, but I’m open to the idea that both could be true.
That said, I then cleaned the CPU again and this time applied Thermal Grizzly’s Conductonaut, which dropped the temperatures significantly, to a peak of 57 °C (this is the highest core temperature reading, as always). All the results from this review (barring the LN2 ones of course) were done with the Conductonaut between the CPU die and IHS, with Kryonaut between the IHS and the CPU block of the AIO (the EKWB Phoenix).
If you do buy the 8086K, I’d recommend you test the overclocking limits first, before deciding to replace the TIM, as you may not need to do this. If you do want to replace the TIM, do a better job than I did, as I seriously doubt that the Kryonaut would result in higher temperatures than the stock TIM. As always, remember that doing any of this will void your warranty, so exercise extreme caution.
Overall, I’m thoroughly happy with the 8086K and am rather pleased with what Intel has delivered. I do wish there was a matching or companion 8085K or some such thing without hyper-threading at a slightly better price, but for what’s being offered, this is a solid, no-nonsense, high-performance CPU. The results I got were achieved with a random CPU sample, and you may be able to do even better with the specific 8086K you get – but no matter the silicon sample you have, there’s a high chance of you hitting 5GHz on all cores.
For those who are interested, you might be surprised to find out that if you stick to the default CPU clocks (i.e. if you don’t tinker with multi-core enhancement or anything else that may deviate from the Intel specification), you can run this CPU at a voltage as low as 1v, coming in well below max TDP, but still getting the best gaming performance money can buy. And that’s not bad at all.