Next: Interactive ray tracing
Up: Running Tachyon
Previous: Command line image format
  Contents
  Index
Tips for running MPI versions
Tachyon support the use of MPI for distributed memory rendering
of complex scenes. Most commercial supercomputers and cluster
vendors provide their own custom-tuned implementations of MPI
which perform optimally on their hardware. Homegrown clusters
typically use either the LAM or MPICH implementation of MPI.
While Tachyon will work with any comformant implementation of MPI,
some implementations perform much better than others. In the
author's experience, the LAM implementation of MPI gives the
best performance when used with Tachyon.
johns@megapixel.com