Every few years we have new generations of processors from Intel and AMD appearing on the market, competing to see which one achieves the highest performance and, therefore, the interest of each of the buyers. However, we are not going to talk about route maps or key names that appear in a PowerPoint in the middle of a conference to talk about three ideas that we will see in the processors of the future. If not more of a series of common trends which we are going to review.
The world of PC CPUs is evolutionarily quite boring, from time to time we see new processors appear with a greater number of cores, these being faster per MHz and having more cache memory. Apart from this, the evolution has been maintained in that sense, which is not a small thing due to the titanic effort involved in being able to bring these beasts to the market. Whether we are talking about chips today of billions of transistors.
What improvements will we see in the processors of the future?
We are not going to talk about extremely complicated concepts of computer architecture, but about a series of trends that we can all understand and that in a few years you will see appear on the roadmaps and specification sheets of Intel and AMD. Most of these changes are intended to achieve better system performance, but deviate from the traditional path followed so far. Why? Because they are much more efficient solutions, both in cost and consumption.
Even more heterogeneous nuclei
With the Intel Core 12 we have seen the introduction of E-Cores, making there two types of core in each processor, both binary compatible and sharing access to the same programs. The idea of launching smaller cores to perform background tasks is something that comes from mobile processors and will not be the idea we will see copied later.
Many processors from Intel and AMD have a Boost mode that allows one of the cores to achieve a higher clock speed than the rest for certain tasks. Well, what we will see in the processors of the future will be the implementation of what we could call superkernels, with a higher processing capacity than normal cores. Let’s not forget that programs have parts that work in parallel and parts that work in series and it is important to speed up both points.
The idea is that instead of performing a Boost in the clock speed that moves consumption to the stratosphere to place a core at maximum speed, one of these cores with higher capacity is used to solve those specific moments in the execution of the tasks. programs.
Using accelerators for common tasks
An accelerator is an integrated circuit that performs a task or series of tasks in much less time than the central processor of the system, in addition to freeing it from performing said task. Well, things like:
- Compress or decompress a .ZIP file or other similar format
- Convert a file from one format to another
- Moving data between memories or within the same data well will be transferred to specialized units in the future.
This is something we’re already seeing in server processors being integrated into versions of Windows and Linux for those systems. As with everything, over time these will find their way into desktop and laptop CPUs where programs will take advantage of them.
Configurable peripheral connections
One of the things that we will see will be the use of FPGA chips inside the processor with the aim of being able to configure the different peripheral interfaces as desired. For example, instead of having a fixed configuration of USB ports of each type on a chipset, your motherboard manufacturer will be able to configure the processor’s integrated chipset according to your system configuration. In this way, they will not have to wait for a next generation to integrate certain connection ports, and can add and remove them depending on the target market for which they intend to sell the system.
