To mark the anniversary of the first CPU based on the X86 architecture, Intel has released a limited edition CPU in the Core i7-8086K. Although the new 8086K looks nothing at all like it’s namesake the Intel 8086, or iAPX 86 as it was also known, it’s a milestone none the less marking 40 years of X86 based architecture. The 8086 CPU paved the way for modern computing and it’s easy to see why Intel would commemorate such an occasion with a special edition CPU.
Below you’ll see a picture of the original Intel 8086 CPU, as you can see modern CPUs look much different than this one. Those 20 pins are gone and now replaced with 1151 contact points and a nice shiny heat spreader to protect that piece of silicon.
We’ll be comparing the 8086K to the 8700K which it’s based on and see what if any performance differences there may be between the two. As you can see from the picture above they appear to be the exactly the same on the outside, let’s take a look at how they perform and see if that 5.0 GHz boost gives it an advantage over its sibling.
From the outside, the 8086K doesn’t really look any different than the 8700K but all the changes are on the inside. The base speed has been raised to 4.0 GHz from 3.7 GHz while still maintaining its 95 W TDP. Keep in mind that Intel’s TDP rating is taken from the base CPU speed and doesn’t include any turbo function. Looking at this they have added an extra 300 MHz to their base speed of the 8700K and still remain within the same TDP. Does this indicate a slightly higher binned CPU?
Turbo boost, this is where my expectations and reality started to diverge. Initially when the 8086K hit the news, I expected a 300 MHz increase across the board but unfortunately, that’s not the case. Intel has raised the base clock and top boost bin by 300 MHz resulting in a nice increase for the base speed at 4.0 GHz and a single core boost up to 5.0 GHz but the other turbo speeds remain unchanged from the 8700K this CPU is based on. See the chart below, it breaks down the turbo by cores:
|Intel Core i7-8086K Vs 8700K Turbo Boost|
|i7-8086K||5.0 GHz||4.6 GHz||4.5 GHz||4.4 GHz||4.4 GHz||4.3 GHz||4.0 GHz||$424.99|
|i7-8700K||4.7 GHz||4.6 GHz||4.5 GHz||4.4 GHz||4.4 GHz||4.3 GHz||3.7 GHz||$349.99|
Test Setup and Results
For comparison testing of the i7-8086K, the i7-8700K was used. Both CPUs were run at BIOS defaults with XMP enabled as well as MCE (multi-core enhancement) enabled, this should give a good basis for comparison of the “out of the box” experience with the 8086K. To cool the CPUs, the EKWB Predator 360 was used which is a very good 360 mm AIO allowing for maximum ambient overclocking.
I have also added results of an all-core 4.8 GHz overclock which is about the maximum that could be obtained out of both CPUs using Prime 95 for stability. Without de-lidding, the temperature peaks were pushing into the 90 °C+ range which is what limited the overclock even though the voltage was still only 1.3 V and well within the CPU’s tolerance. Yes, the anniversary edition 8086K still uses paste as TIM under the IHS. The 8086K was de-lidded later on to reach maximum potential so the paste is verified.
|CPU||Intel Core i7 8086k/ Intel Core i7-8700K|
|CPU Cooler||EK Predator 360 XLC|
|Motherboard||ASUS ROG Maximus X Apex|
|RAM||2 X 8 GB Corsair Dominator Platinum RGB DDR4 3600 MHz Cl16|
|Graphics Card||ASUS ROG GTX 1080Ti Strix OC|
|Storage||Samsung 850 Pro M.2 256 GB|
|Power Supply||Superflower Leadex 1 kW|
|Operating System||Windows 10 x64 Redstone 4|
- AIDA64 Engineer CPU, FPU, and Memory Tests
- Cinebench R11.5 and R15
- HWBot x265 1080p Benchmark
- SuperPi 1M/32M
- WPrime 32M/1024M
- Intel XTU
All CPU tests were run at their default settings unless otherwise noted.
All game tests were run at 1920×1080 and 2560×1440 with all CPUs at default settings unless otherwise noted. Please see our testing procedures for details on in-game settings.
- 3DMark Fire Strike Extreme
- Middle Earth: Shadow of Mordor
- Metro Last Light
- Ashes of the Singularity
- Rise of the Tomb Raider
AIDA64 CPU, FPU, and Memory Tests
One thing to mention right at the outset of this section is the CPUs scored nearly identical throughout all tests. The results will be posted but in all honesty, there isn’t much to say about them. The 8086K really is just an 8700K with a bit higher turbo boost on one core, the higher base speed is really irrelevant since both CPUs boost all cores to 4.3 GHz on default settings.
One thing to note here is cache speed, we have seen in the past that different motherboards, most notably from different suppliers will set the cache speed on the same CPU at different speeds depending on the board. Typically that can vary anywhere from 4000-4400 MHz for an 8700K when left at AUTO in the BIOS. On the Maximus X APEX at default, the 8700K had a cache speed of 4.0 GHz, the 8086K on the other hand consistently ran the cache at 4.5GHz.
In the AIDA Cache and Memory benchmark shots below the cache speed is indicated as north bridge clock. Both these shots were taken with the CPUs at default, the 8700K reports a cache speed of 3400 MHz but was actually running at 4000 MHz, the 8086k reports 4500 MHz which is the actual speed at which it ran.
This trend continued throughout the testing, even at 4.8 GHz the 8086K retained the 4500 MHz cache speed, meanwhile, the 8700K automatically bumped its cache speed to 4300 MHz.
Even though the 8086K had a slight advantage throughout all the tests it really wasn’t apparent in the results as you’ll see below.
|AIDA64 Cache and Memory Benchmark – Raw Data|
|Intel Core i7 8086K Defaults||51512||52429||47610||45.3|
|Intel Core i7 8700K Defaults||50507||52207||46809||46.8|
|AIDA64 CPU Tests – Raw Data|
|Intel Core i7 8086K Defaults||72304||26395||576.4||29390||7395|
|Intel Core i7 8700K Defaults||72323||26351||575||29391||7395|
The results above are very consistent with each other and well within any margin of error to be called equal.
|AIDA64 FPU Tests – Raw Data|
|Intel Core i7 8086K Defaults||6714||54875||29534||7775|
|Intel Core i7 8700K Defaults||6677||54876||29534||7775|
Here again, the results are so close, even identical in the last two tests.
Real World Tests
Next, we will move on to something a bit more tangible/productivity-based with compression, rendering, and encoding benchmarks.
|Cinebench R11.5/R15, POVRay, x265 (HWBot), 7Zip – Raw Data|
|Intel Core i7 8086K Defaults||15.23||1413||3016.17||46.96||40628|
|Intel Core i7 8700K Defaults||15.3||1417||2991.08||47.26||40632|
|Intel Core i7 8086K 4.8 GHz||16.98||1561||3343.42||51.75||43991|
|Intel Core i7 8700K 4.8 GHz||16.92||1552||3347.31||52.89||44301|
Here again whether at default or overclocked the CPUs score the same.
Pi, and Prime Based Tests
Moving on from all the multi-threaded goodness above, we get to some Pi and Prime number based tests. SuperPi, WPrime, and Intel XTU which is based on Prime95.
|SuperPi and wPrime Benchmarks – Raw Data|
|Motherboard||Spi 1M||SPi 32M||WPrime 32M||WPrime 1024M||Intel XTU|
|Intel Core i7 8086K Defaults||7.436||410.748||3.468||99.43||2381|
|Intel Core i7 8700K Defaults||7.865||429.363||3.467||99.57||2368|
|Intel Core i7 8086K 4.8 GHz||7.67||420.854||3.139||89.151||2609|
|Intel Core i7 8700K 4.8 GHz||7.686||421.096||3.171||89.165||2590|
Well, no real surprises here, the one outlier is SuperPi, being a single threaded application it’s one instance that can take advantage of the i7-8086K’s single-core 5.0 GHz boost when running defaults.
As far as the gaming tests go, these were done at resolutions of 1920 x 1080p and 2560 x 1440p according to our Graphics Testing Procedure which was linked earlier. The CPUs are all at default settings with the 3600 MHz Corsair Dominators and again one set of results with an all-core overclock to 4.8 GHz on both CPUs for comparison.
Up first we have our 1920 x 1080p results as you can see it made relatively no difference running the CPU at defaults compared to 4.8 GHz. Even the faster cache on the 8086K made little to no difference. The only test it was noticeable with was Ashes of the Singularity which is quite CPU intensive.
Looking at the 2560 x 1440p benchmarks we see a similar story Metro Last Light and Shadow of Mordor both saw some modest gains when the CPUs were at 4.8 GHz but for the most part, the results are quite similar across the board.
Moving on to our one synthetic gaming benchmark 3DMark’s Firestrike Extreme we’ll see a breakdown of different aspects of the benchmark. Aside from an overall score these benchmarks also break out the graphics scores and physics scores separately.
Again like the majority of the results throughout testing, there’s very little difference if any between the 8086K and 8700K results. Whether the CPUs are at default or overclocked they still score very close to each other.
Power Consumption and Temperatures.
For the next set of results, the CPUs were tested at defaults once again. Using AIDA64’s Stability and FPU stability tests along with Prime95 small FFT test the CPUs were pushed to test power consumption and maximum temperatures.
Here’s where the 8086K and 8700K really seem to differ from each other. The 8086K just appears to be a hotter, more power hungry CPU. Even at idle, under zero load, the CPU is already pulling an extra 10 W over the 8700K and that just grows as the tests get harder. Moving from a 25 W difference in AIDA64 Stability up to a full 45 W in Prime95.
The 8086K follows the same trend here as it did above with power consumption running consistently hotter than the 8700k.
The initial overclock of 4.8 GHz was tested with Prime 95 Blend test for a minimum of 60 minutes but both CPUs were touching 90 °C which is close to the upper temperature limit of 100 °C where the CPU will start to throttle its speed. With a constant core voltage of 1.3 V, both CPUs hit roughly the same temperatures not like we saw at default settings.
To go any further in exploring the full potential of the i7-8086K it was going to need to be de-lidded. With the help of a Rockit88 de-lid tool, the job was done within minutes. To re-TIM the CPU some Thermal Grizzly Kryonaut paste was used simply because liquid metal isn’t the best choice if the CPU may be used with extreme cooling. Although Kryonaut may not be as efficient as the liquid metal it still did a fine job in helping keep those temperature spikes to a minimum. The average temperature was dropped by 6 °C and the maximum went from 90 °C down to 74 °C opening up a bit more room for overclocking.
To continue the journey toward higher clocks was going to require more voltage. By moving up to 1.35V a stable overclock was reached at 5.0 GHz using Prime95 which relies heavily on AVX based CPU instructions. There’s a reason most Intel motherboards come with the ability to set and AVX offset. Unlike older instruction sets AVX really hammers a CPU with the added work that these instructions can handle at the same speed which can limit the CPUs capabilities. 5.1 GHz was almost on the table but cooling ran a bit short, possibly with liquid metal it would have been reachable but not with the current set-up.
This broke the overclocking session into two parts AVX limits on one hand and non-AVX on the other. So following the same principals as Silicon Lottery would after the AVX portion was nailed down the test was switched to RealBench which uses many different types of CPU instructions. This took a bit more time to iron out but in the end, an overclock of 5.2 GHz was reached with a -2 offset set in BIOS for AVX. The voltage was reaching maximum limits touching on 1.44 V occasionally to maintain stability at 5.2 GHz. This was definitely the end of the line for stability testing. This CPU would also run Realbench at 5.1 GHz with the same 1.35 V setting for Prime95 which would keep the maximum voltage lower.
Overall not a bad result. If you put any faith in Silicon Lottery’s statistics this CPU would be in the top 60% of the 8086K CPUs they had tested so a bit above average. Comparing this to the 8700K results and it’s a different story where only the top 17% of 8700Ks tested were able to hold this speed.
To give an idea of what an AVX offset overclock does in the real world, under many situations the CPU will run full speed at 5.2 GHz but in the event, it encounters AVX instructions the CPU speed drops to 5.0 GHz. Below you’ll see an example of both situations. In the first picture of Firestrike Extreme look to the highlighted area of the graph in the lower section the CPU speed is reported as 5.2 GHz throughout the benchmark. In the second picture of HWBot’s X265 benchmark which uses AVX instructions, the benchmark reports the full 5.2 GHz speed but notice the CPUz window below it which reports the actual speed of 5.0 GHz. Proving this overclock with offset is working as it is intended.
Typically we would include a “Pushing the Limits” section here but the i7-8086K was already at it’s maximum as far as voltage and temperatures are concerned. This could be used as a 24/7 overclock but personally, I would back down 100 MHz with a -1 offset allowing for the use of substantially less voltage for 24/7 operation. By doing so the resulting voltage would be 1.35 V vs. 1.45 V losing 100 MHz in non-AVX loads is very insignificant in the big picture and this is much easier on the CPU.
The below collection of benchmarks was performed at 5.2 GHz with a -2 offset just to show what the performance would be if a user chose to keep the system running with this overclock. As you can see there was a significant improvement over the 4.8 GHz overclock that was attainable before de-lidding the CPU. De-lidding is a choice and be aware that this voids any type of warranty Intel may offer. Regardless of the assumed quality of Intel TIM, it’s good enough to run the CPU as intended.
40 years of modern computing is something to be proud of and releasing a CPU to mark this milestone is a great idea. The i7-8086K was a very capable CPU but so is the i7-8700K which is its foundation. During normal operation though, it’s difficult to tell the difference between the two as was evident in all the benchmarks that were performed earlier in this review. Both CPUs scored within a few points of each other and on the rare occasion that the single core 5.0 GHz boost could be leveraged the 8086K came out ahead.
According to statistics from Silicon Lottery the 8086K Overclocked higher than the vast majority of 8700Ks that they tested. If this sample was sold by them already de-lidded and pre-binned it would capture a nice premium over the retail price. It appears though that a user would stand a fairly good chance of getting a 5.2 GHz CPU by purchasing an 8086K, a 60% chance according to Silicon Lottery.
That brings us to value, available at Newegg.com for $424.99 you’ll be paying a $75.00 premium to own one of these limited edition CPUs. What do you get for that premium though, a collector’s edition CPU which stands a better chance of reaching higher clocks than your typical 8700K? Is it worth it? That depends on one’s perspective.
Personally, I would spring the extra cash and take a chance even if, at the end of the day, all I had was a small piece of computer history.
-Shawn Jennings (Johan45)