More cores for the people: Coffee Lake-S with Intel Core i7-8700K and Core i7 8700 in review

For example, you currently only launch two CPUs per class, each with two more cores. If the Core i7 has 6 cores and SMT, which allows 12 threads, the Core i5 can still have 6 cores without SMT. The Core i3 is... The Z370 chipset At first glance, not much has changed about the chipset itself, except that Intel has now reopened up to AMD and has the top three in the nomenclature. PCIe connectivity remains unchanged at 16 Lane... In the AI test of Civilization IV, all CPUs position themselves in a good way as one might expect from the technical data. Even with the Time Spy CPU test, the world of the advertised hierarchies is (almost) still in order. Positive falls here ever... For this benchmark, the Core i7-8700K is exceptionally just ahead of the Core i7-8700, albeit only at average frame rates. The differences are so small that you could almost push everything into the tolerance range. The... Even if a GPU limit is not to be denied here, the CPUs still scale up to a good 140 FPS. However, the overclocked Core i7 of both genrations will no longer bring any extra power, as you can see well on the Min-FPS. ... In Civilization, the overclocked Ryzen 7 1800X can still keep up quite well. Otherwise, the Core i7 are the dominant CPUs in this selection. However, the Core i7-7800X is the inglorious exception, as it still sorts itself behind the... Clock precedes core count and the Core i7-8700 is faster than the K-Pedant. Which brings us to the paradox of our foreword to the tests, where a CPU with nominally lower turbo clock suddenly performs better than the actually more expensive Mo... Project Cars offers a very similar image, in which the non-K model performs a tick faster. The AMD CPUs, on the other hand, have no chance and are clearly behind. Individual results of all maps Other game - same image, except that this time even the older Core i7-7700K can push itself in front of the Core i7-8700K. However, if you run at overclocked to 4.9 GHz with the same clock, the Core i7-8700K is faster again. This shows... This benchmark is effectively in the balance between the two new Core i7, while the core i7-7800X, the core i7-7700K (unclocked) and the AMD CPUs have to be significantly behind. At the same clock, the speed limit is suddenly 4.9 GHz ... In this game, the focus is on the Frame Times and the Min-FPS as a logical consequence. Otherwise, the Core i7-8700 is again a tick faster. The rest repeats itself as well as with the other benchmarks. Single... Important preliminary remark on CPU selection and overclocking Contrary to our usual handling, not benchmarking the productive areas with overclocked CPUs, as it is also rather unlikely in reality, this time we have both Coff... 2D Performance In order to better understand some of the later results, we are putting forward a good old acquaintance. With our GDI/GDI+ benchmark, we are first testing two different output methods for 2D objects, how to... CPU Performance: Workstation Of course, in the production area, not only the 3D graphics performance is important in the production area for the just-set applications, because many things are calculated by the CPU in parallel within these applications (Simu... For these test series, we use benchmarks from the rather versatile SPECwpc suite for workstations, which rely on very different mathematical calculations, which on the one hand can be perfectly parallelized and where it is at the same time a... Overclocking First of all, we are exploring the limits. To make sure that we don't run in thermal limits, because Intel still relies on the tiresome thermal paste between Die and IHS, we initially use the chiller again and cool the... Summary First of all, we would like to thank AMD and their Ryzen family. Without this wake-up call or food for thought, Intel would probably have continued to delight us with now boring four-cores in the consumer sector. But that's how the train-zwan brings...

For example, you currently only launch two CPUs per class, each with two more cores. If the Core i7 has 6 cores and SMT, which allows 12 threads, the Core i5 can still have 6 cores without SMT. The Core i3 now has a real four cores and SMT is no longer necessary for what is certain to hurt. Here, even the most die-hard Intel enthusiast can thank AMD, without whose Ryzen CPUs Intel would hardly have been able to offer at least six-nuclei in the consumer segment.

Today we will compare the two Core i7 and in two further follow-up articles the two Core i5 and Core i3. We want to compare each of the three CPU classes with selected competitor CPUs and their respective older generations. Added value for upgrade rs and retrofits and reason to upgrade? That is exactly what we want to find out.

But the new core diversity also has a downside in the form of a new chipset and a lack of compatibility with the Kaby Lake boards. Even if the number of pins has remained the same, 20 pins are new or to ensure a more stable power supply. This becomes necessary when one thinks of the now available 6-core and the colported 8-core CPUs, which can/will follow later. But more on that.

Coffee Lake-S

Just last year, Intel announced a new process architecture that streamlines the process optimization (PAO) sequence to deliver a new process in every third generation – a profound departure from the glorious era of Intel's dual Tick-Tock Strategy However, the latest 14nm++ process is already the fourth generation of 14nm manufacturing, which was originally introduced with Broadwell in 2014. This in turn proves that you are more likely to step on the spot and THAT PAO seems to have fallen by the wayside again.

Because Intel's new consumer CPUs werenot derived from server products, just like AMD's current Ryzen CPUs, Intel can retain the cache design that has already proven its good suitability for games among the older generations, so that you can up to 12 MB of L3 cache. The graphic shows very clearly how the die is divided and where the thick cache is located:

Anyone who already knows all this is not fooled, because Coffe Lake-S is by no means a new architecture, but is actually still based on Skylake-S and is thus a simple remake of a previous refresh. So it's ticking a lot at Intel and the customer has to keep waiting for the next really real gong. Let's take a quick look at the specifications, in which an old acquaintance also appears in the form of the integrated graphics unit used.

The Intel UHD Graphics 630

More than 60% of the mainstream market does not use a discrete graphics card, which of course gives Intel an advantage in the broader entry-level market. The well-known iGPU as GT2 is still based on generation 9.5, but gets the ultra surcharge: UHD Graphics 630 instead of HD Graphics 630, which should indicate that ultra HD codes have also been implemented for video playback.

Otherwise, this continuation of the familiar also means that you still support the same legacy VP8 and AVC codecs, HEVC 10-bit decode/encode, VP9 8/10-bit decoding, VP9-8-bit encoding (no support for VP9-10-bit encoding), HDR, and Wide Color Gamut support. Configurations of 6+2 (i5 and i7) and 4+2 (i3) for the graphic are still standard.

Apart from HDCP 2.2 support for DCP1.2a, nothing really new has been added to the hardware compared to Intel's predecessors. We still get the same sublices with 8 EUs each (a total of 24 EUs in a GT2 configuration), but Intel has at least increased the maximum clock speed by a staggering 50 MHz. So sensations look different.

The "new" socket 1151

The firmware of the new Z370 boards actively prevents the start of older Kaby Lake CPUs for protection, and a firmware brake for Coffe Lake-S has already been installed in the latest BIOS versions for the Z270 motherboards to prevent possible damage. According to a motherboard manufacturer, in both cases, a real boot operation cannot be carried out because the CPUs are not recognized.

Unofficial scheme of possible pin assignment

If you look at the pin assignment, you can see that the motherboard layout also had to be changed a bit in order to supply the now added pins (predominantly above the middle). According to the manufacturer, however, not only reserve pins were reassigned, but also the assignments as Normal Open (NO) and Normal Closed (NC), so that the measure seems quite plausible in the end. In addition, the power supply of the integrated graphics unit has also been improved.

The specifications

The TDP now gives Intel 95 watts each for the two Core i7 and the Core i5-8600K, and 91 watts for the higher clocked Core i3-8350K. The non-overclockable models without K in the name are consistently indicated as 65 watts TDP, which is quite courageous, as we will see. But in the end you have the good old base clock, which you can simply lower in reverse compared to the previous models in order to get the TDP classes right again.

It remains to be seen what the motherboard manufacturers will make of it and whether all-core turbo clock rates will be made possible again this time, which are significantly higher than this and thus will break the TDP classes. We will come back to this later in the power consumption measurements.

Model	Base clock	Single-core turbo	All-core turbo*	Cores / Threads	TDP-class destroyer	L3 cache	Memory	RCP (per 1000)
i7-8700K	3.7 GHz	4.7 GHz	4.3 GHz	6/12	95 watts	12 MiB	DDR4-2666	359 USD
i7-8700	3.2 GHz	4.6 GHz	4.3 GHz	6/12	65 watts	12 MiB	DDR4-2666	303 USD
i5-8600K	3.6 GHz	4.3 GHz	4.1 GHz	6/6	95 watts	9 MiB	DDR4-2666	257 USD
i5-8400	2.8 GHz	4.0 GHz	3.8 GHz	6/6	65 watts	9 MiB	DDR4-2666	182 USD
i3-8350K	4.0 GHz	–	–	4/4	91 watts	8 MiB	DDR4-2400	168 USD
i3-8100	3.6 GHz	–	–	4/4	65 watts	6 MiB	DDR4-2400	117 USD

An important preliminary remark about the tests

Anyone who wonders about certain results in the following tests, e.g. is faster in gaming the Core i7-8700 than the Core i7-8700K and is back in front of the core and full load of the Core i7-8700K, although both have an all-core turbo of 4.3 GHz, so rest assured: Intel has been able to rely on these reproducible Outliers also do not give a conclusive answer.

But it's certainly not the motherboard or the processors we're using, because we reproduced these tests with motherboards from other manufacturers and other CPUs and got the same deltas. The motherboard manufacturer itself has also been able to recreate this rather abstruse behavior in its own tests, but has not yet received an answer from Intel. So we see the current results in the end with some belly-mas.

Pages: