AMD winds up Zen 6 cores for faster clock speeds

After teasing its 256-core Epycs for over a year, AMD has revealed details of its Zen 6 cores, which promise a 27% increase in instructions per clock (IPC) across various AI and server workloads. The new CPUs are expected to have boosted clock speeds ranging from 4.1 to 5.3 GHz, with even higher frequencies anticipated for consumer chips in the future. However, AMD is not planning to release Ryzen processors immediately.

The Zen 6 core comes in two variants: a frequency-optimized version (Zen 6) and a density-optimized version (Zen 6c). This division is reminiscent of AMD's Bergamo Epycs launch in 2023, which featured 128 compact Zen 4c cores on just eight chiplets. The goal was to achieve higher performance while maintaining a smaller die size compared to Intel's Efficient (E-cores) cores.

Both Zen 6 and Zen 6c share the same instruction set architecture (ISA), with clock rate being the primary distinction between the two. Despite this, AMD claims a 20 to 31% IPC improvement over Zen 5 for cloud and AI workloads. While this performance boost is expected to translate to client-side applications, many PC enthusiasts were disappointed by the relatively modest 5% gains seen with the Ryzen 9000-series processors.

According to AMD's Mike Clark, the front end of Zen 6 is the primary driver of performance improvements. The dual front end design works well for simple workloads but can struggle with more complex, dynamic loads. To address this, Zen 6 features a more sophisticated front-end, enabling better performance in workloads with varying demands. Additionally, AMD implemented a distributed scheduler in the out-of-order pipeline, which helps manage the core's backend execution units when running at slower clock speeds.