I know AMD will be releasing some form of Ray-tracing over the next year or so, which is great. we will end up with both AMD and Nvidia GPUs using half or maybe three quarters of the hardware installed.
example, people with nvidia cards have X tensor cores sitting there in many cases not being used. the cuda cores are all busy bees while the tensor cores are doing nothing.
as we know tensor cores are very quick and love supersampling, I guess AMD will end up with something similar.
Would it be possible to use the under utilised tensor cores to supersample or handle the HMD quality when using VR, we need to tap into the processing power and use our GPUs to the full. Apparently Nvidia can deploy this with driver updates, but the games developer needs to engage with them allowing them to generate the required information to deploy with the driver.
this would allow people with these cores to boost their image quality without taking a massive performance hit. I think.
this would be so cool, please please pretty please allow us to use all of our GPUs potential.
example, people with nvidia cards have X tensor cores sitting there in many cases not being used. the cuda cores are all busy bees while the tensor cores are doing nothing.
as we know tensor cores are very quick and love supersampling, I guess AMD will end up with something similar.
Would it be possible to use the under utilised tensor cores to supersample or handle the HMD quality when using VR, we need to tap into the processing power and use our GPUs to the full. Apparently Nvidia can deploy this with driver updates, but the games developer needs to engage with them allowing them to generate the required information to deploy with the driver.
this would allow people with these cores to boost their image quality without taking a massive performance hit. I think.
this would be so cool, please please pretty please allow us to use all of our GPUs potential.
