Releasing FSR4 INT8 for older Radeon cards would improve AMD’s software support reputation and retain customers in a price‑sensitive market, directly influencing future GPU purchase decisions.
The video calls out AMD for refusing to backport its latest upscaling technology, FSR4, to older Radeon GPUs. While the official version relies on FP8 hardware found only in the RDNA4‑based RX 9000 series, an accidental source‑code release in August 2025 revealed an INT8 implementation that runs on RDNA2 and RDNA3 silicon, which dates back to the popular RX 6000 line.
Testing by the creator and community members showed the INT8 build delivers visual quality close to the FP8 version, with a modest performance hit that still outperforms FSR 3.1 on legacy cards. A simple DLL swap and a few driver tweaks enable the feature on RDNA3 and, with extra effort, on RDNA2 GPUs. Despite this proof of concept, AMD’s official response has been a terse “No updates to share at this time,” while the company pushed an optional AI bundle in its latest driver instead.
The presenter contrasts AMD’s stance with Nvidia’s approach, noting that DLSS 4.5—though optimized for newer RTX 40 GPUs—has been back‑ported to RTX 20‑series cards, reinforcing Nvidia’s reputation for long‑term feature support. He also highlights rising GPU prices and recent driver‑support controversies, arguing that withholding a functional feature erodes goodwill among existing Radeon owners.
If AMD were to release the INT8 version officially, it would provide an easy PR win, extend the life of thousands of GPUs, and narrow the perceived gap with Nvidia’s software ecosystem. In a market where price pressures and supply constraints are acute, delivering FSR4 to legacy hardware could bolster brand loyalty and help reverse recent sales slumps.
Comments
Want to join the conversation?
Loading comments...