Embedded MRAM, RRAM… enablers of a smarter world

From the Internet-of-Things (IoT) to wearable devices, smartphones and tablets, we observe an ever-increasing demand for smarter and more secure connected devices. These billions of connected objects are “expected to change our lives”; we want self-driving cars, health monitors, computer glasses, intelligent factories, etc. However, this trend will impact the power budget of the whole digital supply chain and won’t be sustainable in the long term. Why? Because the two most common methods of data management have weaknesses. One method is to increase the amount of transferred-data by the object to the cloud fully in charge of storage and computing (more intelligence, more security means more data and more processing). The other is to have an object able to (partially) store and process data, sending less but more meaning-full data to the cloud (edge or fog computing).

The first way is suitable for some scalar sensors but very limited for more complex applications due to the wireless data-transfer’s power consumption and its impact on the battery lifetime, creating a frustrating user experience. Moreover, the increasing amount of data will challenge the cloud’s reliability (data losses…) as well as the infrastructures costs and power budgets.

The second method aims to give more capabilities to the object, like, for instance, the smartphone or the smartwatch. The main advantage is to lower the pressure on the cloud by offering local storage and pre-processing, enabling also low latency decision making mandatory for some critical applications (health monitoring, industry 4.0). Nevertheless, the object’s smartness (more processing capabilities, more storage, more flexibilities…) will also impact the power consumption (decreasing the battery life) and its cost (complexity, maintenance…).

You could also increase the battery size; however, this is more expensive, and inapplicable for most applications due to technical (size, weight) or cost constraints.

The second approach is the only way able to give birth to this intelligent connected world but it is limited by the current technology attempting to square the circle and, unfortunately, unable to solve all those contradictions with those levels of constraints.

eVaderis offers chip designers a solution to this dilemma by providing low power, high-performance, non-volatile memory compilers, logic libraries and subsystems, paving the way for new chip design paradigms.