X-ray crystallography resolves the structures of the molecular machines in living cells at an atomic level of detail, but only in states that can be captured as crystals, which are often not functional states. Cryo-electron microscopy enables a more complete view of biomolecular conformational variability, but at lower resolution. The molecular dynamics flexible fitting (MDFF) method combines the atomic detail of crystallographic structures with lower-resolution cryo-electron microscopy to synthesize all-atom models of complex macromolecular aggregates such as the ribosome in multiple functional states. The 2.9 release of NAMD combines GPU acceleration of implicit solvent simulation, optimizations exploiting shared memory within a single machine, and a faster "lite" grid forces implementation to bring MDFF capability from the supercomputer to the desktop. VMD can connect to a running simulation to visually monitor the progress of the simulation or to interactively steer a molecule with either the mouse or a haptic (force-feedback) interface device. The convergence of methodology, software, and hardware advances thus opens what was once the domain of extremely expensive equipment to commodity computers.