The first generation of gravitational wave (GW) detectors have produced some interesting astrophysical results. These include a bound on the density parameter of relic GWs that is tighter than that imposed by Big Bang nucleosynthesis, an upper limit on the fraction of the Crab pulsar's spin-down energy loss emitted in GWs, etc. Second generation detectors will start taking data in a few years from now and are expected to make the first direct detection of gravitational waves. They will also likely contribute to multi-messenger astronomy by coordinating with electromagnetic (EM) observatories the hunt for GW counterparts of EM sources, and vice versa. Some of these pursuits will require fast, computationally intensive, GW searches: A compact binary coalescence (CBC) involving at least one neutron star has been conjectured to be the progenitor of a short duration gamma-ray burst. If the conjecture is true, a fast search with a network of GW detectors for the signal from the inspiral phase of such a binary can alert EM observatories of the source location and the approaching time of merger, which is roughly when the GRB is emitted. In this talk I will discuss a fast method for coherently searching for GW signals from CBCs in the data of a network of detectors. This method extends the work of Cannon et al. in applying singular value decomposition to reduce the number of filters required in the search. I will also discuss what astrophysics one can learn from the joint GW and EM observations of short duration GRBs.