Nowadays there is a high speed improvements in processors frequencies and processors amount on a single map. So these opportunities have to be used in such fields as modeling and simulation, prediction models and simulations. One of these fields is strictly connected to the papers subject (heat conduction). Heat conduction calculation in 3D space is quite a problem for 3D space calculations because of time spent on calculation for usual approaches is quite large number. It is possible to separate heat conduction calculation full iteration into several portions of calculations. These portions of calculation could contain separated calculation blocks. It is possible to implement using ADI (Alternating Direction Implicit) principles in dividing heat conduction calculation full iteration into 3 parts. Each of these parts ignores one of coordinate axes directions, but allow to calculate only three diagonal matrix using Thomas algorithm. It means that additional effort on difference scheme construction has the payback of calculation time reducing because of separated calculable blocks. One another boost of calculation speed is dynamic time step implementation by taking into a count prediction matrix of next iteration heat transfer calculations. This approach has no strict impact on time step calculations for each next iteration, but it can be bordered between minimal and maximal possible time step defined values. Provided solutions allows to manage algorithm calculation time by applying as many computers as many times it is needed to reduce the calculation time.