As much as processor manufacturers are focusing their efforts on improving energy efficiency , both through micro-code and by reducing manufacturing lithography, in the end in recent years it seems that we are experiencing a “BOOM” in terms of number of cores in processors, or in other words, we are seeing an increase in performance based on brute force. Is this what the future holds for processors? More and more cores in each new generation, or can we expect improvements in another way?
For many, many years we have been more or less stagnant in terms of the number of cores in processors, where 2 and 4-core CPUs have dominated the market for several decades. When the 6 and 8-core processors arrived, the evolution seemed normal to us, since in addition to reducing the lithography to integrate a higher density of transistors, it was necessary to be able to parallelize a greater number of simultaneous tasks using more cores and threads. However, in the last generations of AMD and Intel CPUs we are seeing a core count that has increased ostensibly, up to the incredible AMD Ryzen Threadripper 3990X with 64 cores and 128 threads, a processor that seems more suitable for servers than for the domestic sphere.
Is the future of processors to increase the number of cores?
There are many examples that we can give you regarding the improvements based on increasing the number of cores that Intel and AMD are introducing in their new generations of processors, and that we are talking about processors oriented to the consumer market and not to the professional, so we cannot help but wonder if this is a mere ” numbers war ” as it has already happened in the past, or if users really need our CPU to be able to perform so many tasks in parallel .
The fact that the processor has more cores is not something that will be a disadvantage in any case, except for some side effects such as a higher TDP in many of these processors with a large number of cores. This means that the processor can parallelize more simultaneous tasks, and that applications can take advantage of it (remember that Windows is an operating system that really gets along with parallel tasks), and if Intel and AMD use it in a particular war of numbers but they do it while maintaining prices (that is, they offer processors with more cores but at the same price as the previous generation) users can only welcome them.
That does not mean that, as we have already discussed previously, in the end the physical space that the processor die has is limited and until the manufacturing processes give another twist to the industry, in the end the number of cores that has a processor is physically limited. By this we mean that, while it is likely that we have not yet reached a ceiling (and we remember those 64 cores of the AMD 3990X), there will come a time when we will touch it and the number of cores of the scope CPUs domestic to stagnate again.
This means that, from our point of view, the future is not that processors will have more and more cores (which will surely have them, at least for a few more years), because the time will come when manufacturers have to really work hard or find alternative solutions to improve efficiency so that the performance of the processors can continue to grow in terms of user performance.
What improvements can we expect in future processors?
We have already mentioned that increasing the number of cores to be able to parallelize a greater number of simultaneous tasks is undoubtedly one of the improvements that we can expect for the next generations of processors, but of course it will not be the only one. As you well know, AMD and Intel are working very hard to also improve lithography and reduce the size of the transistors, which translates into a greater number of transistors occupying the same space, something that greatly improves efficiency in terms of performance by watt consumed.
We can also expect improvements in the amount and speed of the cache memory, something that they are emphasizing (especially AMD) in recent times and that also improves the overall performance of the processors. Likewise, improving the bandwidth of the input and output interfaces is a pending task that sooner or later will have to tackle, since both PCIe and USB are gradually growing in terms of the bandwidth they need and are already close to the limit. that processors can support.
Finally, another of the improvements that sooner or later will have to address (and that could potentially change a lot the form factor of processors as we know them) is thermal performance. Reducing the lithography to put a greater number of transistors in the same space necessarily implies an increase in the density of heat produced, and although we have not yet reached the limit in this area, we are not far from the point where the heatsinks we currently use are not are able to dissipate heat fast enough, so the method used will necessarily have to be changed.
