We already know that the AMD Ryzen 7 8700G APU enjoys memory that has been overclocked; preliminary assessments showed that adjusting the “iGPU engine clock and the processor’s memory frequency” could yield significant graphical improvements. While gamers can enjoy a boosted integrated graphics solution that is comparable in performance 1080p stakes to a discrete Radeon RX 6500 XT GPU, AI enthusiasts are eager to experiment with the “Hawk Point” pat’s Radeon 780M IGP and Neural Processing Unit (NPU)—the first generation Ryzen XDNA inference engine can unleash up to 16 AI TOPs. Just before the official Ryzen 8000G processor introduction earlier today, one person, chi11eddog, shared their results on social media. The first set of findings focused on the Radeon 780M feature; NPU-centric data might be available later.

They conducted brief testing using the recently launched Ryzen 7 8700G desktop processor from AMD, along with a 16 GB DDR5-4800 memory stick and an MSI B650 Gaming Plus WiFi motherboard. Further up in the tables, the MSI-only “Memory Try It” function was used; this helped achieve and measure a number of “higher system RAM frequency” settings. This is chi11eddog’s brief analysis of the benchmark findings: “7600 MT/s is 15% faster than 4800 MT/s in UL Procyon AI Inference Benchmark and 4% faster in GIMP with Stable Diffusion.” chi11eddog’s inference chart indicates that the processor can generate 210 Float32 TOPs in its default memory state.The Radeon 780M iGPU in the flagship APU performs 7% better than the baseline at 6000 MT/s, and 11% better at 7200 MT/s. This suggests that bandwidth is heavily dependent on this GPU. The integrated RDNA 3 graphics solution was also put to the test by their GIMP w/ Stable Diffusion benchmarks, and it was found to be fairly bandwidth hungry.