IBM hitches the power of a doe for its new launch AI Chip

By | August 20, 2015

To achieve something great, one needs lot of patience and effort. After immense effort IBM finally launches AI Chip, a tiny chip but with the power and energy of a Human Brain and the TrueNorth project have touched its glory. From last year, IBM have been putting a lot of effort to transform this technology and to bring it down to that level where it says “the computational equivalent of a mouse’s brain”.

According to Wired, “Around 48 millions of artificial nerve cells is used to make circuit board for the chip, and these chips inside the TrueNorth design are built to act like neurons”. The above mentioned nerve cells are the same amount of nerve cells that a doe consists of, therefore the comparison.

The name Neuromorphic chips are titled because of the fact that they mimic organic brain cells and IBM is making lot of effort to steer the improvement of “deep learning” algorithms that are steered by the cells via a boot camp at its silicon valley R&D Lab. The initial verdict is heartening, the Government and Academic researchers are taking part in the boot camp and have started to create software that can be used as a combination with the TrueNorth design and steer the Internet’s AI services  of today and tomorrow. This could be anything from instant translation on Skype to Facebook’s face recognition.

Though there are many companies who produces chips, but IBM sets its own standard and it is way better than the current companies chips. Firstly because it can easily manage these algorithms in much smaller area, secondly it consumes less electrical power supply and thirdly, it also allows us to fix the chip into other phones or devices like wearables and medical aids.

Highlight: Less Power Consumption

In figures, the IBM architecture uses around 70 watts of energy to power 5.4 billion transistors. To put that into our frame of reference, modern processors use 35 to 140 watts to power just 1.4 billion transistors.

TrueNorth has to construct its own path towards reaching for commercial purpose. A plan has been discussed to create an identical architecture that has the computing power of the human brain. This is under process and it is predicted to be unveiled in the year 2020.