With the start of the new year, new things will also come in the world of hardware, which lately has been locked in a series of cycles where we only see new products in the form of components and devices, which are nothing more than an improved version of what already seen before. Well, in this article about what will be the technological trends of the future We are not going to talk about new processors, graphics cards or any other component.
Every hardware device that we buy is still the final work of the application of a technology that has been given an outlet in the form of a product that the manufacturer considers to have value for the final buyer. That is, it is useful to solve a problem or solve a job. However, these base technologies are often not talked about and in this case we are going to talk about those that will emerge in 2023 and will change the way our PCs are known.
Four technological trends that will mark the future of hardware
The technologies that we bring to many of you will sound strange to you, some of them you will have already seen applied to the different PC components, others, on the other hand, we will begin to see them deployed this year in products other than the PC. Whether in the server market or very high-end mobile phones. In any case, we have considered it important to review these technological trends of the future.
The high manufacturing costs of the chips with the smallest transistors, together with the fact that not all the parts of an integrated circuit need to scale in the same way, have meant that in order to reduce costs, many designs have ceased to be based on a single transistor. chip to disintegrate or separate into several. This technique also makes it possible to create processors that, if they were a single block, would be impossible to build because they exceed the maximum size per chip allowed.
Although AMD has been the one that has supported this solution the most with its desktop Ryzen CPUs and its new RX 7000 graphics cards, in reality it will be the Chinese component manufacturers that, having vetoed access to technologies to manufacture chips with more advanced nodes, will seek said solution to be able to advance technologically and have processors in terms of performance that are competitive.
Talking about 3 nm for PC components is still too early and rather we will have to wait until 2024 to be able to see the first chips under said manufacturing node. However, TSMC factories are already at the top of their lungs making the chips for very high-end phones with chips from Apple and Qualcomm, which is the prelude to what we will see in PCs in the future.
However, as has happened in previous generations, the cost of manufacturing wafers will rise again, causing more expensive chips under the same size. Specifically, the average improvement in cost per transistor will only be 15%. Which will greatly mark the designs that we will see in the future. Not in vain, this will be the reason why we will see a greater deployment of solutions based on chiplets.
Using optics to transmit data
One of the problems of breaking up a chip into several chiplets is communication, due to the fact that having a greater distance between the components also increases their energy consumption. In the case of Intel and TSMC, this has been solved with proprietary technologies whose prices are controlled with an iron fist. The answer to it? Using photonics or the use of optical interfaces to communicate various chips with each other.
The idea is none other than to eliminate the increasing resistance with the length of the semiconductor cables. Many of the designs that you will see based on chiplets from this 2023 will use optical interfaces and will be based on the principles of photonics to communicate with each other without raising consumption to the stratosphere.
AI will take a bigger role in hardware
Today through services for the automatic generation of images by words, or automatic responses such as ChatGPT, since we find that the majority of the public is fascinated with AI. However, its implementation in the field of software does not only require changes in the hardware, especially adding new units never seen before, one of the technological trends that we are seeing implemented in recent years.
If not, AI will also begin to be implemented with respect to hardware, not only for applications. For example, we will see monitored voltage regulation and clock speed systems being managed by an algorithm developed by an AI and integrated into the chip. When organizing the different processes to be executed in a processor and a graphics card, we will see how Deep Learning and Machine Learning will take care of it.