Forward-looking: When AMD releases its Zen 6 processors, the company won't simply iterate on Zen 5 – it will introduce a fundamentally new CPU core architecture, shedding decades of legacy baggage embedded in the x86 platform. One of the most significant changes in this redesign is AMD's adoption of Intel's Flexible Return and Event Delivery (FRED) instructions, a modern replacement for a system mechanism that has been part of PCs since the early 1980s.

Every time a computer receives a network packet, completes a disk write, or a user moves the mouse, the processor handles an event known as an interrupt – a low-level signal that requires switching between user code and system code. For more than four decades, this process has been managed by the Interrupt Descriptor Table (IDT), first introduced with Intel's 80286 processor. Despite countless CPU generations, the fundamental way interrupts are handled has changed very little.

FRED aims to modernize this outdated foundation. Instead of forcing programmers to juggle multiple steps and ring-level transitions for each interrupt, FRED processes them in a single, atomic operation. This results in a cleaner, more reliable handoff between user and kernel code, reducing the risk of race conditions and inconsistent states when multiple events occur simultaneously.

AMD's adoption of FRED is part of a broader collaboration between AMD and Intel to align their work on the x86 architecture. Both companies are members of the x86 Ecosystem Advisory Group, a partnership announced in 2023 to ensure a unified instruction set across both vendors.

A year into that alliance, AMD confirmed that Zen 6 would incorporate FRED, following Intel's upcoming Nova Lake and Panther Lake platforms. While no current CPUs support the feature, its simultaneous rollout across both ecosystems marks an unusually coordinated shift in the historically fragmented x86 landscape.

// Related Stories

This transition was not guaranteed. AMD had developed its own mechanism, Supervisor Entry Extensions (SEE), as an incremental workaround that preserved compatibility with older software. Industry consensus, however, favored Intel's cleaner, full-scale replacement.

Even Linux creator Linus Torvalds endorsed Intel's approach in a forum post, calling it a more complete solution that eliminates long-standing architectural cruft.

From a technical perspective, FRED simplifies CPU privilege handling by reducing it to just two levels – ring 0 for kernel code and ring 3 for user code. This streamlining not only removes layers of complexity for developers but also reduces CPU cycles spent on event handling. The practical result is improved performance and lower latency for workloads that generate a high volume of interrupts, such as network-intensive processes, high-refresh-rate gaming, and audio processing pipelines.

The performance gains may be even more pronounced in virtualization environments, where event handling often passes through multiple layers of software. Fewer transitions between privilege levels translate directly into reduced overhead and faster context switching – long-standing challenges in x86-based hypervisors.

Software support for FRED has already started appearing in major platforms. The Linux kernel has included preliminary FRED support since version 6.9, and future Windows releases are expected to implement it as well. Because only operating systems and low-level drivers interact directly with these instructions, end users are unlikely to notice any visible difference – but developers and system maintainers will.

Image credit: Fritzchens Fritz