Increasing biophysical detail in multi physical, multiscale cardiac model will demand higher levels of parallelism in multi-core approaches to obtain fast simulation times. As an example of such a highly parallel multi-core approaches, we develop a completely distributed bidomain cardiac model implemented on the IBM Blue Gene/L architecture. A tissue block of size 50 × 50 × 100 cubic elements based on ten Tusscher et al. (2004) cell model is distributed on 512 computational nodes. The extracellular potential is calculated by the Gauss-Seidel (GS) iterative method that typically requires high levels of inter-processor communication. Specifically, the GS method requires knowledge of all cellular potentials at each of it iterative step. In the absence of shared memory, the values are communicated with substantial overhead. We attempted to reduce communication overhead by computing the extracellular potential only every 5th time step for the integration of the cell models. We also investigated the effects of reducing inter-processor communication to every 5th, 10th, 50th iteration or no communication within the GS iteration. While technically incorrect, these approximation had little impact on numerical convergence or accuracy for the simulations tested. The results suggest some heuristic approaches may further reduce the inter-processor communication to improve the execution time of large-scale simulations.