A sparse particle CFD solver for reacting, turbulent multiphase flows with multiple mapping conditioning (MMC) is investigated in terms of its scalability and performance in an HPC environment. Previous versions of the solver, called mmcFoam, were unable to transfer the significant potential speed-up of the sparse particle method to large cases with hundreds of cores. This work investigates the solver’s bottlenecks, and the main issues hindering the scaling are identified and solved. In particular, the selection of particle pairs for the diffusive mixing prevented the scaling of the code due to a required all-to-all MPI communication. A review of the particle pairing model and algorithm led to the development of a new particle pairing method based on a sub-set of processors with a significantly reduced number of required MPI communications. Further, load balancing and particle management have been improved. The overall scaling of the new mmcFoam solver reaches ideal near-linear scaling. Therefore, this work ports the advantages of a sparse particle method for solving turbulent, reacting flows to large-scale HPC environments.