“…In addition, novel and complex modes are tested, such as the Affine/Warped mode in VVC and AV1. Works targeting dedicated hardware designs for the inter prediction explore several techniques to obtain gains in coding efficiency, energy consumption, required area, high throughput and memory bandwidth, such as: multiplierless implementations [11], [12], [13], [37], [39], [40], [38]; approximate computing solutions [37], [13], [42]; reuse of common subexpression [11], [39], [12]; parallelism exploration [11], [37], [39], [40]; and hardware reconfigurability [41].…”