News stories tagged with #Zen-6
AMD is expanding its RDNA 4m GPU architecture with the addition of new targets, GFX1171 and GFX1172, now integrated into the LLVM compiler. These GPUs, expected to power the upcoming Medusa Point APUs, share the same instruction set as GFX1170 and signal a strategic evolution for future AI and gaming workloads. With support for FSR 4, FP8/BF8, WMMA, and SWMMAC, the architecture marks a significant leap in integrated graphics performance and machine learning capabilities.
AMD Medusa Point with Zen 6 and RDNA 4m: Next-Gen APU Spotted in Geekbench
AMD's upcoming mobile APU series, Ryzen AI 500 (codenamed 'Medusa Point'), has surfaced in the Geekbench database. Based on the new Zen 6 architecture, the chip features ten cores, 32 MB L3 cache, and is manufactured using TSMC's 2-nm process. Early benchmarks reveal impressive efficiency, approaching that of the Strix Point despite significantly lower clock speeds. Additionally, FP16 support for AI workloads has been spotted for the first time.
AMD Unveils Zen-6 with RDNA 4m and Up to 24 Cores for Next-Gen Ryzen APUs
AMD is developing its next Ryzen generation with the Zen-6 architecture, featuring processors with 6 to 24 cores, enabled by a chiplet design with up to two CCDs. Alongside this, the RDNA 4m, a refresh of the RDNA 3 architecture, is being designed for these APUs and will support FSR 4 (FSR Redstone). The architecture introduces enhanced AI and GPU processing capabilities through FP8/BF8 conversion and WMMA instructions, improving performance for AI and GPU workloads. These advancements position AMD to better compete with Intel's Nova Lake.