Interactive computer graphics still depends on rasterization hardware; ray tracing cannot yet replace it

Despite ray tracing’s superior image quality, current interactive graphics (games, real-time applications) cannot operate without rasterization because dedicated rasterization hardware delivers far higher throughput than current ray tracing hardware at interactive frame rates. Rasterization effectively provides primary visibility ‘for free’ — its GPU hardware was designed for exactly this workload. Removing rasterization and replacing it with ray tracing would reduce frame rates drastically. The GPU’s rasterization circuitry is thus the foundation on which real-time rendering stands: ray tracing is added on top of it, not substituted for it. This may change as GPU architectures shift toward ray-tracing-optimised designs (the lecturer’s group researches ray-tracing-only GPUs), but hybrid rasterization+ray tracing is the practical ceiling today.

Examples

A modern game at 60 fps uses rasterization for its primary-visibility frames per second. Adding 1 spp ray-traced shadows costs perhaps a few ms per frame. Replacing rasterization entirely would currently require far more compute for the same throughput.

Assessment

Explain why interactive graphics still requires rasterization hardware even though ray tracing produces better images. What specific hardware design difference makes rasterization faster than ray tracing for primary visibility today?