Abstract: A method of explicit lockstep for functional safety includes spawning, by a main thread, a first safe thread core and a second safe thread core. The method also includes initializing and mapping a first data register associated with the first safe thread core to each safe variable of a set of predetermined safe variables. The method further includes initializing and mapping a second data register associated with the second safe thread core to each safe variable of the set of predetermined safe variables. The method also includes comparing, by a hardware comparator, a first safe variable value in the first data register to a second safe variable value in the second register. The method further includes issuing an error completion to the first safe thread core and the second safe thread core when the hardware comparator detects a mismatch between the first data register and the second data register.

Abstract: A method of performing fault detection during computations relating to a neural network comprising a first neural network layer and a second neural network layer in a data processing system, the method comprising: scheduling computations onto data processing resources for the execution of the first neural network layer and the second neural network layer, wherein the scheduling includes: for a given one of the first neural network layer and the second neural network layer, scheduling a respective given one of a first computation and a second computation as a non-duplicated computation, in which the given computation is at least initially scheduled to be performed only once during the execution of the given neural network layer; and for the other of the first and second neural network layers, scheduling the respective other of the first and second computations as a duplicated computation.