A former lead on AMD’s FidelityFX Super Resolution (FSR) has suggested that one major reason FSR 4 feels uneven right now is simple: many of the people behind it may no longer be at AMD.
Over the past few months, discussion around AMD’s upscaling and frame-generation roadmap has intensified, especially as gamers try to understand why FSR 4 support is limited on older Radeon GPUs. While recent updates like the latest Redstone release and FSR 4.1 have been viewed as meaningful progress, they also sparked fresh frustration because FSR 4 still isn’t broadly available across previous GPU generations.
The controversy escalated when an FSR 4.1 leak surfaced that reportedly included specific DLL files enabling FSR 4 functionality on older GPUs with INT8 support. AMD moved quickly to remove the leaked files. Many users assumed the incident would push AMD toward an official, fully supported rollout of FSR 4 on older hardware—yet that broader support still hasn’t materialized.
At the same time, FSR 4’s real-world adoption has been slow. Only a small number of games are taking advantage of its more advanced features, and high-profile additions like Ray Regeneration have only appeared in a limited set of major titles, with examples mentioned such as Call of Duty: Black Ops 7 and Crimson Desert. Even among the 100+ games said to support FSR 4, not all of them have native integration; in many cases, users must rely on overrides through AMD’s Adrenalin Software to access enhancements. Those enhancements are also described as being restricted to RDNA 4 hardware, including Radeon RX 9000-series graphics cards.
According to statements shared on Discord by Colin Riley (known there as “Domipheus”), internal staffing changes may be a key factor behind the current state of FSR 4. Riley previously held lead roles on FSR 2, FSR 3, and FSR 4, spending nearly nine years at AMD before moving to a software company called JECO. While he noted he didn’t personally work on Ray Regeneration, he said he knows the person who did—and that individual has reportedly left AMD for NVIDIA.
Riley also claimed that the director who originally started GPUOpen left AMD for Intel, and that another director who managed him around the time of the FSR 4 release also departed for NVIDIA. He further mentioned that several “great engineers” made similar moves. People switching companies is common in the industry, but what gives his comments extra weight is his reflection on how strong morale and retention used to be on the FSR side for years—“until it didn’t,” as he put it.
The broader implication is that it may not be only FSR that’s impacted. If experienced engineers from FidelityFX and GPUOpen have moved on, AMD may be dealing with a talent gap that affects how quickly features get refined, how widely they get deployed, and how clearly limitations are communicated to the public.
That last point—communication—has become a central theme in community criticism. If FSR 4 can’t be enabled on older GPUs due to technical constraints, performance targets, validation requirements, or product strategy, many users argue AMD would benefit from saying so directly. Silence can sometimes reduce short-term backlash, but in a competitive GPU market, clear messaging tends to build longer-term trust.
Competitors have faced similar criticism in different ways. NVIDIA is still called out for limiting certain new features to newer GPUs, even while offering its latest upscaling support on older RTX generations. Intel, meanwhile, has dealt with driver and optimization complaints, but it has steadily improved since its early Arc days, and developers increasingly recognize its ongoing support efforts.
AMD has a massive Radeon user base and is reportedly already working on a future evolution of the technology referred to as FSR Diamond. The company is also heavily invested in consoles and future GPU architectures. Still, with millions of PC gamers relying on Radeon hardware, many in the community see this as a moment for AMD to re-focus on end-user support: accelerate game adoption, reduce confusing “override-only” scenarios, and be more transparent about what’s coming—and what isn’t—for older GPUs.
