In the simulation of the metal injection moulding (MIM), the behaviours of feedstock are much di?erent from which of the polymer injection. It is a mixture of the metallic powder in high concentration and some plastic binder. The advance in simulation of the mould ?lling with such high viscous feedstock is featured by the development of a fully explicit vectorial algorithm. On the basis of previous explicit software realized by the authors, the new algorithm avoids the global solution for pressure ?elds and the use of MINI elements to improve its e?ciency. Except for the operations at element level, neither global solution nor the construction of global matrix is required in simulation. A special strategy is used to regulate the incompressibility condition in ?lled domain at each time step. In case of the MIM problems, this method provides a fast way to simulate the ?lling processes. The computational cost is about linearly proportional to the degree of freedom number. Moreover, this vectorial algorithm can be easily parallelized for high performance computation with multi-clusters. The comparison of numerical results with previous simulations on 3D cases proves the validity and e?ciency of new algorithm. In the simulation of the metal injection molding (MIM), the behaviors of feedstock are much di? Erent from which of the polymer injection. It is a mixture of the metallic powder in high concentration and some plastic binder. The advance in simulation of the mould? lling with such high viscous feedstock is featured by the development of a fully explicit vectorial algorithm. On the basis of previous explicit software realized by the authors, the new algorithm avoids the global solution for pressure? elds and the use of MINI elements to improve its e? ciency. Except for the operations at element level, neither global solution nor the construction of global matrix is ​​required in simulation. A special strategy is used to regulate the incompressibility condition in? lled domain at each time step. the MIM problems, this method provides a fast way to simulate the? lling processes. The computational cost is about linearly proportional to t he degree of freedom number. Moreover, this vectorial algorithm can be easily parallelized for high performance computation with multi-clusters. The comparison of numerical results with previous simulations on 3D cases proves the validity and e? ciency of new algorithm.