How to fix computing AI energy problem: Run everything backwards

How to fix computing AI energy problem: Run everything backwards


Imagine taking a hammer in your laptop. You separate it and fly to plastic, battery and circuit board firing. This will be a task of vandalism, a shocking ruin of money and resources, so much that it seems absurd. But the truth is that, every time we use a computer, we are working with a machine, which is even more useless at the fundamental level.

All this goes back to a decision taken decades ago about the deep functioning of computer logic and how these machines remove data, a process that essentially produces large amounts of waste heat. For a long time, we have pitched with useless computers. But with the rise of artificial intelligence, which has brought the power demands of computing to new heights, it seems that inconsistent decisions may be about cutting us. We may need to re -design computing from scratches.

Thankfully, we know what really to do. It involves a trick that is unlikely to a touch: the processor to do everything twice, once forward, then in reverse. “Reversionary computing can be much more energy efficient than traditional computing, and it is probably the way we should have originally manufactured,” Hannah Arle In the UK-based reversible computing company Vaire Computing.

Increased energy efficiency is the result of a thermodynamic trick we have known since the 1970s, but had never been used …

Article revised on 16 January 2025

We corrected the work of the gate in computing.

(Tagstotransite) Environment (T) Computing (T) AI