News Categories
UCL Scientists Claim to Have Created Crash-Proof Computer
By Michael Low & Wong Chung Wee - on 19 Feb 2013, 10:36am

UCL Scientists Claim to Have Created Crash-Proof Computer

A team of computer scientists at the University of London (UCL) claims to have invented a self-healing and crash-proof computer. Their computer is based on systemic computing where each set of computer instructions are executed independently, making the whole system less prone to our familiar BSOD.

Once in a blue moon, you may witness a red screen of death

Our current computers carry out their instructions procedurally, usually one after the other in a fixed order. Workflows of such are fine until a process fails for some reason, resulting in an operating system lockup which Windows users have affectionately termed BSOD (Blue Screen of Death).

The computer developed at UCL is different as the instructions and data are essentially mirrored across several different digital entities that are called 'systems'. These digital systems work simultaneously although independent of each other; however, they do share a section of the host's memory for context-sensitive data like global variables and parameters.

A graphical representation of a suggested programming methodology for a systemic computer <br>Image source: Department of Computer Science, UCL

In the event a digital system crashes, the computer is able to recover the failed system by rebuilding its data to start fresh again. The digital systems are said to execute in a random order using a "pseudorandom number generator " that acts as a task scheduler designed to mimic nature's randomness. According to one of the scientists, he said that the pool of systems interact in parallel, and randomly, and the result of a computation simply emerges from their interactions.

The publication did wryly note the far-fetched claims of being entirely crash-proof but the researchers are determined to push forward as they attempt to teach their systemic computer to rewrite its own code in response to environmental changes using machine learning. For more details, the team's research paper can be read here.

Source: New Scientist, University of London