Everyone has an opinion. I think AMD will continue to develop graphics RDNA IP and technologies because it’s in their interest, just as AI/ML in datacenter CDNA IP is in their interest. There has to be a viable alternative to Nvidia. AMD graphics IP numbers in the millions of devices if consoles are included.
I think AMD knows that the community will not support them if FSR is forked to use AI and limits compatibility to RDNA2 and newer; I can already see the backlash. Yet, in every DLSS vs FSR thread, everyone keeps harping on about how AMD needs to improve FSR’s upscaler (likely Nvidia DLSS users with Nvidia hardware). Well, to do that, you have to break its wide compatibility, and then the options become limited for those without the necessary hardware. So, for these people, I think AMD can improve the upscaler for newer hardware and keep compatibility for older hardware (even if older HW can’t use the new algorithms). It just takes time, and I’m sure AMD is using Xilinx hardware to develop better algorithms for FSR. They just have to not use algorithms patented by Nvidia or anyone else. That’s a tough proposition.
Everyone has an opinion. I think AMD will continue to develop graphics RDNA IP and technologies because it’s in their interest, just as AI/ML in datacenter CDNA IP is in their interest. There has to be a viable alternative to Nvidia. AMD graphics IP numbers in the millions of devices if consoles are included.
I think AMD knows that the community will not support them if FSR is forked to use AI and limits compatibility to RDNA2 and newer; I can already see the backlash. Yet, in every DLSS vs FSR thread, everyone keeps harping on about how AMD needs to improve FSR’s upscaler (likely Nvidia DLSS users with Nvidia hardware). Well, to do that, you have to break its wide compatibility, and then the options become limited for those without the necessary hardware. So, for these people, I think AMD can improve the upscaler for newer hardware and keep compatibility for older hardware (even if older HW can’t use the new algorithms). It just takes time, and I’m sure AMD is using Xilinx hardware to develop better algorithms for FSR. They just have to not use algorithms patented by Nvidia or anyone else. That’s a tough proposition.