Massively parallel implementation of the multireference Brillouin-Wigner CCSD method

► We show scalability up to 1000 cores of our MR BWCCSD code for molecules with hundreds of orbitals. ► We show further speed-up by task scheduling based on regrouping data independent procedures. ► Future work will enhance scalability using processor groups for each reference-related set of equations. ► These changes should enable routine calculations for systems described by 700-1000 orbitals.