In an engaging talk, Fujimura-san (D2S) reviewed the trajectory of MDP over the years. He pointed out the inexorable trend from rule-based, context-independent methods to simulation-based, context-sensitive processing. Ohara-san (Nippon Control Systems) reviewed the trials and tribulations faced by MDP software providers in actually delivering software following this trend.
Abboud (Intel) made the claim that multi-beam will relieve the pressure on MDP throughput. At the same time, it is evident that such relief (should it appear) will quickly be swallowed up by yet more demanding corrections. In the panel discussion, TSMC warned that MDP is near to becoming the greatest cost in the mask production pipeline, assuming multi-beam writers enter production on time. For the VSB writers to date, the speaker reported the worst case MDP time for TSMC as 500k CPU hours.
Fujimura, et al., Trends in mask data preparation, doi:10.1117/12.2073368
Ohara, MDP challenges from a software provider’s perspective, doi:10.1117/12.2070163
Abboud, et al., Mask data processing in the era of multibeam writers, doi:10.1117/12.2072135