Hidden profiles in AMD’s Adrenalin 26.6.2 build reveal an 8X scaling ratio for FSR Multi Frame Generation, though the feature remains non-functional. The upcoming RDNA 5 architecture will likely handle the actual rendering path when FSR Diamond launches in 2027.

AMD Drivers Hide 8X FSR Multi Frame Generation in Plain Sight

Hidden driver stubs point to a major leap in upscaling, though the feature isn't working just yet.

AMD’s latest Adrenalin 26.6.2 WHQL build has quietly slipped into circulation, and it is carrying a hidden toggle for FSR Multi Frame Generation scaling all the way up to 8X. If you dig past the standard interface, the driver profile values are already there, waiting for code that hasn’t been written yet.

Third-party tool RadeonTuner caught the stub first. Developer “Dumbie” exposed two new entries under the experimental FSR panel: MfgOverride to flip the feature on, and MfgRatio cycling through 1X to 8X. They sit behind the “Show Experimental Settings” switch, which means AMD isn’t rolling them out to the general public. Keep in mind that this is purely a driver reservation at this stage.

The Stub Reality

Tests on the RX 9070 XT and 9060 XT ran through Forza Horizon 6, Pragmata, Resident Evil 9, Crimson Desert, and Death Stranding 2. Zero of them triggered the function. The ratio options exist in the registry, but the underlying rendering pipeline remains dormant. Dumbie pointed out that this is a common development pattern. AMD leaves the API surface area open, reserving the profile entries months before the actual feature code is ready. It's essentially parking space.

The architecture timeline tells you what to expect. ADLX 1.5 documentation from April laid out the exact methods for reading and setting custom MFG ratios. Fast forward to GDC, and AMD SVP Jack Huynh confirmed FSR Diamond as the next-generation stack for Microsoft’s Project Helix. That build targets 2027, locks into RDNA 5 hardware, and explicitly calls out a new ML-based multi-frame generation path. RadeonTuner currently restricts the stub to RDNA 4 or newer, which tracks with AMD’s first-gen AI accelerators, but the heavy lifting will likely wait for the next architecture.

Roadmap and Rivals

An 8X MFG ratio sounds aggressive, but it works on a simple multiplication. You get one natively rendered frame fed into the system, and the AI generates seven intermediate frames. The output hits eight frames total. By comparison, NVIDIA’s current DLSS 4 release caps at 6X MFG, though they have already dropped hints about a future 16X path. Intel’s XeSS sits at a static 4X. Current AMD FSR 4 and Redstone only handle 2X single-frame generation. The gap is wide, and AMD is clearly aiming to close it with a fixed high-ratio target rather than dynamic scaling.

Dynamic MFG matters more than the headline number. NVIDIA’s approach adjusts the ratio per-frame to balance latency and visual stability. A hard 8X cap could introduce severe frame pacing issues or crush responsiveness if AMD doesn’t pair it with a major update to Anti-Lag. Generating seven frames from scratch requires motion vector analysis that current temporal upscalers still struggle with under heavy camera movement.

The hardware requirement restriction to RDNA 4+ makes technical sense, but it leaves a question hanging. Will the 8X feature actually ship with RDNA 4, or is the driver stub just a forward-compatible placeholder until FSR Diamond drops next year? The ADLX 1.5 docs hint at a broad API implementation, but AMD has a habit of gating next-gen rendering features behind their latest silicon. Pure driver skeleton. Just a placeholder for now.

For now, the 8X toggle is a developer warning. It signals ambition, and it tells competitors exactly where AMD wants to draw the line in the dust. Quality will still outpace raw ratios, but the architecture is clearly heading in that direction. Head here to monitor the ADLX SDK updates if you want first-look access. The actual implementation won't show up in a consumer driver for another twelve months.