The slide mentions that "GorgonPoint" combines up to 12 CPU cores based on "Zen5" or "Zen5c", an iGPU based on the RDNA3.5 graphics architecture and an NPU based on XDNA2 - so the IP remains mostly the same. The performance of CPU, iGPU and NPU has been upgraded compared to the current "StrixPoint" chip.
AMD attached first-party performance data to the specs slide. Multi-threaded CPU performance numbers are slightly higher, which seems to indicate that the 4x "Zen5" + 8x "Zen5c" CPU core configuration seems to be carried over; but single-threaded performance is significantly improved. We're not sure what causes it, but there are two theories. The more obvious one is a significant increase in clock speeds, thanks to updated power management; but the second, more radical theory is that AMD updated the "Zen5" P-core to have full 512-bit FP capabilities, similar to the "Zen5" cores in "FireRange" and "StrixHalo" processors.
If you recall, the FPU of the "Zen5" core in "StrixPoint" was limited to dual-pump 256-bit paths to execute AVX512 instructions, a design choice that was probably due to power considerations. NPU throughput has been modestly increased to deliver over 55AITOPS, AMD has achieved full NPU performance across all Ryzen AISKU tiers based on this chip.
All in all, the relationship between "GorgonPoint" and "StrixPoint" is the same as the relationship between "HawkPoint" and "PhoenixPoint".