Thursday, June 4th 2026
AMD: No Definitive Decision on FSR 4.1 Support for RDNA 3.5 APUs
A report from Hardware Luxx suggested that AMD does not currently plan to bring its upcoming FSR 4.1 upscaling technology to RDNA 3.5-based integrated graphics. If accurate this would leave Ryzen AI 300 and Ryzen AI 400 series APUs without support for the latest iteration of FSR. Back in May AMD previously confirmed that FSR 4.1 which utilizes INT8 processing, will arrive for discrete RDNA 3 (Radeon RX 7000) GPUs in July, while RDNA 2 (Radeon RX 6000) desktop graphics are scheduled to receive support by early 2027. This exclusion affects silicon families that are using the RDNA 3.5 graphics architecture. More precisely Strix Point, Strix Halo, Krackan Point and upcoming refreshes like Gorgon Point and Gorgon Halo. These APU configurations integrate high-performance iGPUs such as the Radeon 890M and the top-tier Radeon 8060S, both capable of running the required INT8 operations.
However, AMD's Frank Azor has since pushed back on reports that RDNA 3.5 has been ruled out. Responding on X, Azor stated that "no such decision as being reported and implied here has been made," while adding that AMD is not ready to discuss future product plans at this time. While this does not confirm FSR 4.1 support for RDNA 3.5 graphics, it suggests that AMD has not made a final decision regarding the architecture's inclusion in the technology roadmap.
Sources:
Hardware Luxx, VideoCardz, Frank Azor @ X
However, AMD's Frank Azor has since pushed back on reports that RDNA 3.5 has been ruled out. Responding on X, Azor stated that "no such decision as being reported and implied here has been made," while adding that AMD is not ready to discuss future product plans at this time. While this does not confirm FSR 4.1 support for RDNA 3.5 graphics, it suggests that AMD has not made a final decision regarding the architecture's inclusion in the technology roadmap.


85 Comments on AMD: No Definitive Decision on FSR 4.1 Support for RDNA 3.5 APUs
RDNA 3.5 It is a Product that is included in the current (new) Ryzen AI 400 Chips.
It should have full FSR 4.1 support from the start!
What are these People even thinking?
But AMD should've learned from the whole RDNA2&3 fiasco to either add support in the first place, or don't say anything at all.
AMD and decent support does not fit. Regardless waht the fanboy claim. I had the first time a graphic card in 38 years of computer use which I had to return because the AMD Radeon 9070 graphic card is not able to properly initialise on AMD X670E and show a POST screen. You know you need POST to enter UEFI, to select your build in boot menu etc. DEAD on Arrival. The list of what does not work on amd gets longer the more I bother with amd cpus and gpus.
I do wonder why techpowerup is quite often the last on those news?
AMD should fire their hole developer team and coders and hire fresh.
How could I do floating point math wiht ucontroller which can only handle integers in 8bit? All that nonsense which people think AMD is right because they are just consumers or have no embedded system coding knowledge in assembler and C. Excuses over Excuses from AMD.
the linux kernel just works again from before radeon 7800xt to the radeon 9070 and now with the ryzen 7700x graphics without changing a single configuration. the only thing i had to change was the graphic cards in past few days and where i plug in the DP cable. uefi of course need to be told today that i want to actiavte and use the ryzen 7700x gpu again. EXCUSES over EXCUSES from amd, driver wise it is the same. Lact even shows some values for my ryzen 7700x cpu graphics. The only bad thing is, i can not reserve less than 512MIB VRAM for the cpu graphics
note: linux kernel and gnu userspace was shown as an example that it is the same software stack. same amdgpu. usually you need another target if it is not compatible. even naev.org plays the same with 7800xt / 9070 non xt and now only ryzen 7700x graphics.
Nvidia is forcing fire hazard power connectors such as 12VHPWR and 12V-2x6 down our throats instead of using tried-and-true power connectors such as EPS12V and PCIe Graphics 8-pin power connectors, forcing bad vibe-coded drivers down our throats after having the best drivers in the industry for decades, and partnering with ethically dubious companies such as Palantir. Property insurance agents need to start asking people if a graphics card with a 12V-2x6 or 12VHPWR connector is installed in an active computer in the homes being insured, and if so, double or triple the property insurance premiums to account for the increased fire risk.
AMD's bone-headed decision to not guarantee FSR 4.x to RDNA 2 and up APUs boggles my mind. I can sort of understand not guaranteeing it for laptops because laptop OEMs often gatekeep drivers to avoid having a bad driver release mess a laptop up and drive up tech support costs from having to support bad drivers, but for desktop APUs that are sold into the do it yourself market, come on! We need FSR 4.1 for desktop APUs at least.
Intel's refusal to bring enough discrete GPU supply to the market and the unwillingness to compete anywhere in the upper mid-range gaming market is also crazy.
I love how a light conversation with McAfee kicked off rumours to swirl around.
During their FSR4 development they would have worked on an INT8 version which uses the NPU of RDNA3.5 for up-scaling. This way the performance increase is even larger than on discrete GPUs.
If AMD is not in agreement within their own groups inside and does bad communication, or rather and also most likely, constantly being vague with impopular choices and maybe go back after big backslash. In either case that's something for AMD to solve, not for people to take important and high AMD representatives statements as "ifs".
David McAfee, "VP & GM of Ryzen CPU and Radeon graphics". It's not just any or random representative and neither someone that wouldn't be aware of whats going on internally.
If RDNA 3, heck even RDNA 2 is supported, then I don't see why RDNA 3.5 couldn't be.
They will lose those too if they don't haul ass on this feature set deficit.
I'll tell you one thing, my next GPU is green until they fix their shit, and they're gonna have to be trying real hard to get me back if they wait too long. Tick tock... time's almost up in my book. Every day counts, as an owner of an RDNA3.5 GPU, even just going on principle.
There's nothing wrong in one person saying that they were not aware of any plans and another person replying that final decisions have not been made.
They are clearly discussing how to address this feature on iGPUs, as there are additional challenges and constraints on performance in low bandwidth conditions, especially typical two-channel APUs.
If it can be done effectively across the stack, I am sure they will do it.Imagine you actually knew what they were testing in labs. You would know details, but instead you only have a shallow speculation. That doesn't sound competent from someone accusing others of lacking competency.
We know that running the upscaler on INT8 has its quirks in many games and on much wider memory bus. Those anomalies are likely more pronounced on iGPU with narrow memory bus.
Intel please make a GPU for the upper midrange at least next time.