The new neuromorphic computing platform from IISc represents a significant advancement in AI hardware by enabling processing across 16,500 conductance states, unlike traditional systems.
Sreetosh Goswami highlighted that this innovation aims to resolve longstanding challenges in neuromorphic computing and offers improvements in energy efficiency, speed, and performance for AI.
The IISc team's neuromorphic platform targets energy consumption and computational inefficiency, making it ideal for training large language models with significantly reduced resource requirements.
Goswami expressed confidence that their neuromorphic chip will be available in two to three years, alongside plans to establish a startup to bring it to market.
Collection
[
|
...
]