PC hardware evolves practically every day and the arrival of artificial intelligence has meant that we are increasingly seeing additional components that we can add to a board to increase the capabilities of this technology. Now we know one of the latest developments that an American startup has carried out that has a format that can be used a lot, the M.2since although it is very similar to an SSD that is connected in this way, it is really a accelerator of AI capable of providing up to 24 TOPS of power with a single module.
There are a large number of PC components that may seem like one thing but are actually another completely different, many people who do not know very well what expansion cards are like, for example, can confuse some that are quite similar (although the ports of output are different). But what is clear is that this is probably new accelerator of AI is the one that can cause the most confusion, since its appearance is extremely similar to that of an M.2 SSD, so the company has had to make sure to make it very clear what type of piece it is, so that people do not have doubts.
It has a really small size, but it does a lot for the AI needs of a PC
Many times we may think that in order to add a series of functions related to artificial intelligence to our computer we need a graphics card or a processor that incorporates an NPU to take care of it, but the company MemryX has shown that it is not necessary to have any of these two to add this type of functions to a PC. Although it is clear that the price is not cheap at all, since this small module that can be connected to a board through an M.2 PCIe Gen 3 slot costs nothing more and nothing less than about €150 in exchange.
But in terms of performance, what it is capable of offering is really great, the fact that a single accelerator is capable of providing 24 TOPS allows us to greatly improve the capacity that some computer models designed for artificial intelligence may have. If, for example, it is integrated into a system like the new ones that NVIDIA has launched with a compact size but that are capable of offering between 67 and 157 TOPS, it could greatly increase the overall performance of the AI applications that run on them. , making work and energy efficiency much easier.
As indicated, the module MemryX MX3 M.2 It is really easy to integrate thanks to the open source drivers and developer tools that the company offers, making it not only capable of greatly improving the AI power that a computer can have but also facilitating its implementation. But the most surprising thing is the fact that even though it provides the 24 TOPS that we have mentioned, it only consumes between 6 and 8 watts without needing to depend on active cooling to reduce the temperatures it reaches, which obviously are not too high either.
