Hmmm it looks like who ever is covering this IBM story is re posting the article with new images and tidbits of new information. Looks like a pretty serious PR campaign saying, hey look what we think we can do, or will be rolling out in 3 years from now.
It is interesting to be sure.
One thing that I saw in the previous article was that this archatecture requires a custom programing languige and is geared more towards sensing an invironment, computing as the brain does. That to me presents a bit of a hurdle for integration. Even if the hardware is ready for market in 3 years, software will be way, way behind.
Do we know if POET requires bigtime software updates? or do we think that software will simply be able to utilize POET hardware as is? With some sort of POET driver I wold imagine.
This is new in this article. 176,000 times more energy efficient? nice!
Ultimately, the main purpose of the SyNAPSE project is to take existing systems that simulate the functionality of the brain in software — such as deep neural networks — and run them on hardware that was specifically designed for the task. As you may already know, dedicated hardware tends to orders of magnitude more efficient than simulating/emulating the hardware in software on a general-purpose CPU. This is why IBM is touting some utterly incredible efficiency figures for TrueNorth. For neural networks with high spike rates and a large number of active synapses, TrueNorth can deliver 400 billion synaptic operations per second (SOPS) per watt. When running the exact same neural network, a general-purpose CPU is 176,000 times less energy efficient, while a state-of-the-art multiprocessor neuromorphic system (48 chips, each with 18 cores) is 769 times less efficient. While it’s not directly comparable, the world’s most efficient supercomputer only manages around 4.5 billion FLOPS per watt.
