IBM’s Resistive Computing Has Potential To ‘Massively Accelerate’ AI

Researchers at IBM aim to perfect analog computing technology that would allow processors to more closely mimic the function of the human brain, Extreme Tech reports.

AI-640x353

The technology, known as resistive computing, would allow computer units to retain their history – thereby allowing them to learn during the training process.

Each small processing unit (RPU) mimics a synapse in the brain, and receives analog inputs in the form of voltages. The system can then draw on past “experience” to decide which result to pass along to the next set of compute elements.

If the technology can be fully realised it has the potential to massively accelerate AI and the progression to machine superintelligence. IBM claim that a fully resistive computing based AI system could achieve performance improvements of up to 30,000 times that of current GPU based architectures.

Using chips densely packed with these RPU tiles, the researchers claim that, once built, a resistive-computing-based AI system can achieve performance improvements of up to 30,000 times compared with current architectures, all with a power efficiency of 84,000 GigaOps per-second per-watt. If this becomes a reality, we could be on our way to realizing Isaac Asimov’s fantasy vision of the robotic Positronic brain.

More at Extreme Tech

Leave a Reply