Video compression architectures: dedicated, programmable, or hybrid?

A panel-session summary on video compression architectures is presented. Early approaches to compression were highly polarized: dedicated and programmable. The dedicated architectures were optimized for performance and cost, but had little adaptability to changing environment or developing standards. The fully-programmable solutions could be programmed to do almost anything and could be introduced quickly, but were large and expensive. Recent devices implement the core computational parts of the algorithm in dedicated hardware while providing programmable hardware units to implement the adaptive parts of the algorithm.<>