Artificial intelligence is one of the disciplines that moves the most data when running on the hardware of our PCs, it involves huge bandwidths, which are closely related to the power consumption of data transfer . In an age when the flow of information is more expensive than the processing of information, this has led to the highest bandwidth in the world.
However, in the future, we may need to opt for different technologies than usual in order to reduce the power consumption per transmitted bit. Not only advanced intercom systems based on advanced packaging based on 2.5DIC and 3DIC, as well as the use of photonic interfaces. It is precisely the use of photonics that promises the most advances in terms of artificial intelligence, especially in a future where integrated circuits will not be in a single chip, but in several intercommunicating with each other.
Lightmatter, an AI processor more efficient than an NVIDIA GPU
There is no doubt that NVIDIA GPUs are currently the most powerful type of processor for artificial intelligence. It is precisely this discipline that has dictated the advancements of the latest NVIDIA architectures. However, a startup named Lightmatter has developed a processor that can be considered a generational leap from NVIDIA GPUs. The reason? This company’s chip, called Lightmatter, uses photonics for information communication.
The big problem concerning the development of future processors for artificial intelligence is that within a same processor where the data is sent at a very local level, the consumption is very low, but if we move away then the consumption increases. This problem does not arise with intercommunications based on photons instead of electrons, since the energy consumption in optical interfaces does not increase with the distance the data is located. And it is in photonics that the people of Lightmatter bet to create their processors.
NVIDIA will not be left out
As GPUs and other chips get more complex and move to the chip paradigm, the use of different types of interfaces to mitigate the increase in power consumption becomes a reality. Something the folks at Lightmatter want to solve before anyone else. However, NVIDIA and other companies are not left out and have been developing communication interfaces based on the transmission of data by light for years. Which we’ll see first in the server market, but for now the folks at Lightmatter have moved ahead of NVIDIA by developing a GPU that communicates using optical interfaces.
In any case, Nvidia is not left out, since we know that they have developed a version of their NVLink interface which transmits information using light. What they presented a few months ago. We’ll see how it all plays out, if at all. Will we see it on our gaming PCs? Who knows, but we will have to wait a few more years to see it.