Abstract. We introduce a model of infinitary computation which enhances the infinite time Turing machine model slightly but in a natural way by giving the machines the capability of detecting cardinal stages of computation. The computational strength with respect to ITTMs is determined to be precisely that of the strong halting problem and the nature of the new characteristic ordinals (clockable, writable, etc.) is explored.Keywords: infinite computation, infinite time Turing machine, length of computation Various notions of infinitary computability have now been studied for several decades. The concept that we can somehow utilize infinity to accommodate our computations is at the same time both appealing and dangerous; appealing, since we are often in a position where we could answer some question if only we could look at the output of some algorithm after an infinite amount of steps, and dangerous, since this sort of greediness must inevitably lead to disappointment when we suddenly reach the limits of our model. At that point we must decide whether to push on and strengthen our model in some way or to abandon it in favour of some (apparently) alternative model. But if we do not wish to abandon our original idea, how to strengthen it in such a way that it remains both interesting and intuitive?The behaviour of infinite time Turing machines (ITTMs), first introduced in [1], has by now been extensively explored and various characteristics have been determined. While we are far from reaching full understanding of the model, we nevertheless already feel the urge to generalize further. Perhaps the most direct generalization are the ordinal Turing machines of [2], where both the machine tape and running time are allowed to range into the transfinite. But perhaps this modification seems too strong; with it all constructible sets of ordinals are computable. We would be satisfied with the minimal nontrivial expansion of the ITTM model, i.e. something which computes the appropriate halting problem but no more. Of course, we can easily achieve this goal within the ITTM framework by considering oracle computations, but this somehow doesn't seem satisfactory. We intend to give what we feel is a more natural solution to this problem but which falls short of the 'omnipotence' of ordinal Turing machines.