AMD's forthcoming RDNA 5 GPU architecture promises a significant leap in performance, with an anticipated 5% to 10% boost in Instructions Per Clock (IPC) compared to its predecessor, RDNA 4. This enhancement is attributed to optimized compute units that maintain the same power consumption. As per MLID, RDNA 5 will also witness a 5% to 10% improvement in rasterization performance, with the potential for even greater gains in ray tracing capabilities. Notably, these advancements are achieved without increasing clock frequencies, suggesting that RDNA 5 offers enhanced performance while maintaining power efficiency. Furthermore, RDNA 5 supports configurations exceeding 64 compute units, marking an improvement over RDNA 4.
