The only reason it is not a standardised thing right now is because MS have not implemented some core functions to make PT possible on all cards at a hardware level yet, so Nvidia did what they have always done with stuff like this, do it themselves and make it available to devs to utilise. The core function being SER, but AMD has also said not a peep about supporting stuff like OMM and Cooperative vectors API either, so I suspect they will remain silent until late 2026 and UDNA leaks start to come out.
AMD are once again at least 2 gens behind, so this would make sense.
PT is not Nvidia tech, any card supporting SER and supporting functions of PT can do it on hardware, AMD is just late to the party. By the time UDNA comes out MS will have rolled out SER into DX12 along with the rest of the tech to allow any GPU with hardware to run the tech to be able to PT on hardware.
AMD are once again at least 2 gens behind, so this would make sense.
PT is not Nvidia tech, any card supporting SER and supporting functions of PT can do it on hardware, AMD is just late to the party. By the time UDNA comes out MS will have rolled out SER into DX12 along with the rest of the tech to allow any GPU with hardware to run the tech to be able to PT on hardware.