Intel Coffee Lake

Killian Keller

11 October 2017


The semiconductor industry is one of the fastest growing industries in the world. It creates technological innovation that is unmatched and we as consumers kind of enjoy it. We may still shake our heads when a new device comes out with higher performance and an even higher price tag than the previous one, but in the end, we still buy it, because we love the feeling of a more powerful tool in our hands than the previous one.

One of the key parts in enhancing the performance is the processor. Companies like intel invest billions in the development of new and faster processors.There are several possibilities how to enhance the performance of a processor. On the lowest level, there are the transistors. Tiny switches which can be switched on and off using an electrical potential, they build the foundation of our modern processors. The numbers of transistors on processors exceed billions. To squeeze more and more into a tiny chip, the size of the transistors have to be continually reduced. Intel’s latest Coffee Lake processor architecture uses 14 nm lithography. However, Intel plans to implement 10 nm technology in their next generation (Cannon Lake). The enhancement of the size of size of the transistors is called an enhancement of the fabrication process.

The next possibility to increase performance is changing the architecture, either by adding more cores to the processor, which will then increase the performance if the program can be executed by multiple cores at the same time, or by shortening the longest path the data has to take, such that the clock speed of the processor can be increased (more calculation steps per second). There can also be complete changes in the architecture by changing the computation units included in the processor. These enhancements are called an improvement of the architecture.
There are of course many other possibilities to increase performance, but these exceed the scope of this text.

If one looks at a website of computer shop, one might think: “Wow, there are many processors, how do I chose one?”. Well, this question is hard to answer, as the choice of the processor always depends on the application. Someone who has to run simulations on his laptop would rather choose a processor with high processing power and with many cores (e.g. i7), whereas someone with who only edits text and enjoys multimedia on their device may choose a device with less processing power and less cores, which in turn uses less power (e.g. i5 or i3).

The classification of Intel chips is rather simple, even if it does not seem so. Let’s take an example, the Core i7 8700K. Core is the brand name for Intel’s mainstream CPUs. Next, there is the performance class, this would be the i7. It indicates which kind of application this chip is designed to do. The higher the number after the “i”, the higher the performance. Intel also differentiate between i3, i5, and i7 by using more cores and virtual threads between each series. An i7 for example could have six cores and 12 threads, whereas an i5 may also have six cores, but can only handle 6 threads. The 4-digit number after the iX has multiple meanings. The first digit indicates the generation, in our case the 8th Generation of Intel Core chips. The next digit indicates the performance of the single cores, the higher the number, the faster one core works. The last two digits are a more elaborate coding of the specialities of the processor, but more about that later.

Finally, we have the letter (or the lack of one) at the end of the number. This indicates the special properties of the processor. There are four labels for desktop CPUs and five more labels for laptop versions:

  •  T: T-Series chips are optimized in terms of their TDP, which indicates how much power is dissipated in heat. (desktop)
  •  K: K-Series chips are multiplier-unlocked CPUs which can be overclocked when cooled appropriately. (desktop)
  •  E: E-Series chips are embedded chips, i.e. are soldered directly to the motherboard. (desktop)
  •  P: P-Series chips have a less powerful integrated GPU. (desktop)
  •  U: U-Series chips are again classified in 3 categories (laptops):
    • The last two digits of the classification end in 0: It’s a 15W TDP processor with moderately powerful GPU
    • The last two digits end in 50 or 60: It’s a 15W TDP processor with a more powerful GPU
    • The last digit ends in a 7: It’s a 28W TDP processor, the GPU and the processor can be clocked higher.
  •  H: H-Series chips are rare. They indicate a higher TDP which will generate more heat but they are clocked higher. (laptops)
  •  HK: HK-Series chips are even more rare, they are found in only one processor. The suffix indicates an overclockable mobile processor (Not a very good idea in my opinion, as cooling may become a problem) (laptops)
  •  HQ: HQ-Series chips are again distinguished by their last two digits (laptops):
    • The Last two digits end in 50 or 70: The processor is shipped with the most powerful GPU and with additional eDRAM, which will allow a higher performance even with the same amount of TDP.
    • The last two digits end with anything else: The processor is shipped with a mid-range integrated CPU, which will result in less graphics performance, but the extra room not occupied by the CPU will allow you to run your CPU for a longer time without noticing any spikes in thermal energy dissipated.
  •  Y: Y-Series chips are extremely low power chips, they have a TDP of only 4.5W. (laptops)

What has changed?

The 8th generation of Intel processors includes three codenames: Kaby Lake refreshed, Coffee Lake and Cannon Lake. The Cannon Lake processors will have a change in fabrication process, scaling it down to 10 nm. The Kaby Lake refreshed and the Coffee Lake on the other hand come with less dramatic changes. They do not have a change in the fabrication process, or in the microarchitecture. They both have just optimizations of the microarchitecture. The optimization of the microarchitecture includes more cores in the CPU, which increases the performance by 15 to 25%.

Furthermore, there will be an optimization of the integrated Audio DSP which will enhance audio quality and performance. For most users, this will not be a serious change, as the speakers and headphones still limit the quality.
There will also be an increase in the lowest level cache of the processor, generally increasing performance and speed.

The USB interface will also be brought up-to-date with support for the new USB 3.1 Gen 2 standard, which will allow transfer speeds up to 10 Gbit/s.
Furthermore, there are improvements on the GPU, support for HDMI 1.4 has been implemented in this generation of integrated graphics.

Lastly, we have an optimization of the compatibility of the RAM. Most processors in the Coffee Lake series will be compatible with the DDR4-2666 technology. However, in Kaby Lake designs, there were only two high-end processors with this compatibility.
Unfortunately, not much more has changed. We can expect some improvement in performance, but not in energy consumption or in the fabrication process.

Intel recently launched the Core i9 processors. As these processors are tailored for desktop workstations, they will probably not emerge in the Coffee Lake series, as the main focus has been laid on mobile processors.

Conclusion

While we do not have much changes in the Coffee Lake series, we can anticipate changes for the coming generations of Intel chips. The succeeding series, Cannon Lake, will have an improvement in the fabrication process, and if Intel follows its schedule, we can even expect further scaling down of the process in the years to come. Furthermore, Intel will probably also further improve its architecture, implementing new elements in their Core in the generation to come.

And now as consumers the most important aspect of the Intel processors: Which one to buy?
For engineers, scientists and people who need Computer Aided Design or simulations programs, a Core i7 of the HQ variant might be the best choice. These processors include hyperthreading, which allows for multiple threads to run on the same core. Combined with the large amount of cores inside the processor, you will have a mobile powerhouse, which is able to run simulations and other calculation-intensive programs.

For people with less need of power-intensive programs, for example a person who uses their computer for office application or multimedia, a Core i7 or Core i5 of the U or Y variant will be easily fast enough. The i7 CPUs are generally a bit more future-proof as their performance will also be good enough for future programs, but the i5 CPUs can also handle anything from Office applications to movie watching and even some lighter photo editing.