DIKU Bits: Why Computation Needs to be Reversible, and What This Implies
Torben Mogensen, Associate Professor in the APL section
While Moore's Law postulates that the number of transistors per square mm doubles roughly every two years. Dennard scaling postulates that, given the same voltage, frequency and activity level, power use of a CMOS chip is solely a function of its area. As long as both hold, we get twice the compute power for the same power use every two years or so. But Moore's law shows sign of weakening, and Dennard scaling broke down a decade ago. So we can not rely on transistor shrinking to reduce the power use of computers. This can even be proven theoretically: The 2nd law of thermodynamics implies Landauer's Principle: That the theoretical minimum power use of a logic gate is proportional to the number of bits of information that is lost in an operation. Conversely, if no information is lost during a computation, there is no lower limit to power use. Hence, reversible computation is required if we want more computation per kWh in the future.
We will look at reversible computation and what it implies in terms of chip design and programming languages. We will also, briefly, look at how reversible computation is related to quantum computation.
The talk will be conducted in English
DIKU Bits: A new lecture series at DIKU
In this new lecture series you can get a closer look at the research conducted at DIKU, be motivated you to follow your interest in computer science and be inspired when choosing a subject for your bachelors project. The lectures are addressed to DIKU bachelor students - however, everyone is welcome. Find more information about DIKU Bits in Danish / in English.