Support For AMD GFX11.7 "RDNA 4m" Pending For RADV & RadeonSI Drivers

Key Takeaways

• Mesa patches add ~3,000 lines for GFX11.7 support
• Six patches under review target RADV and RadeonSI drivers
• GFX11.7 aligns RDNA3 ISA with upcoming RDNA4m APU
• Support slated for Mesa 26.1 release cycle
• No product details revealed; AMD's next APU remains secret

Pulse Analysis

AMD’s upcoming RDNA 4m graphics architecture, identified internally as GFX 11.7 (GFX1170), has been on the radar of the open‑source graphics community since February when LLVM shader compiler patches surfaced. The IP sits between the current RDNA 3 family and the next‑gen RDNA 4, blending ISA tweaks that hint at a future APU product line rather than a discrete GPU. While AMD has not officially announced the hardware, the presence of GFX1171 and GFX1172 targets suggests a family of chips that will extend the AI‑focused Ryzen AI series.

The Mesa project, which supplies the open‑source RADV Vulkan driver and RadeonSI OpenGL implementation, just submitted a six‑patch series to enable GFX 11.7 support. Roughly three thousand new lines of code are required, a larger footprint than typical driver updates, reflecting the complexity of integrating the new ISA and address library. These patches are currently under review and are slated for inclusion in the upcoming Mesa 26.1 release. Once merged, Linux users with compatible AMD hardware will gain early access to Vulkan and OpenGL features tailored to the RDNA 4m silicon.

For developers, the driver enablement means they can begin testing and optimizing applications for the next AMD APU generation before silicon ships, potentially accelerating software readiness. The broader Linux graphics ecosystem also benefits: upstream support reduces reliance on proprietary drivers and aligns with AMD’s open‑source strategy. Market‑wise, the move signals AMD’s confidence in its roadmap and may pressure competitors to accelerate their own open‑source graphics initiatives. As the hardware approaches launch, performance benchmarks and power‑efficiency claims will shape how the RDNA 4m platform competes in the AI‑centric computing segment.