SGV 3.0 - a fast detector simulation

The need for fast simulation programs is emphasised, both in terms of the need for "rapid response" to new results - in particular from the LHC - and new theoretical ideas, and in terms of how to cope with multi-billion simulated event samples. The latter would arise both from the need to be able to simulate significantly more events than expected in the real data, also for high cross-section processes, and the need to scan multi-parameter theories. The {\it Simulation \`a Grande Vitesse}, SGV, is presented, and is shown to be able to address these issues. The tracking performance of SGV is shown to reproduce very closely that of the full simulation and reconstruction of the ILD concept. Preliminary results on how to also closely emulate the calorimetric performance from full simulation is presented. The procedure is parametric, with no the need to simulate the detailed shower development, and promises to be many orders of magnitude faster than such approaches. Contrary to what is often the case with fast simulation programs, the procedure gives a somewhat {\it pessimistic} result, compared to the full simulation and reconstruction.