The researchers plan to overcome these limitations by testing the connection of multiple NorthPole chips. The development of this chip is part of a move towards an "edge" computer system, where data is physically closer to the AI applications using them, reducing time delays. The research was published in the journal _Science_.
Key takeaways:
- IBM Research has developed a dedicated computer chip, named NorthPole, that can run AI-based image recognition apps 22 times faster than current market chips.
- The NorthPole chip uses a two-dimensional array of memory blocks and interconnected CPUs, allowing the computing cores to communicate directly with far-away blocks as easily as with those that are nearby.
- Despite its speed, the NorthPole chip can only run specialized AI processes and cannot run training processes or large language models like ChatGPT.
- The research team plans to overcome these limitations by testing the connection of multiple NorthPole chips together in the future.