By using this site, you agree to our Privacy Policy and our Terms of Use. Close

I think some people still don't get how this works.

Ok, a GPU renders pixels. If it costs a GPU 10% of its power to draw 2M pixels. It would cost the same GPU 40% of its power to draw 8M pixels. There is NO WAY AROUND THIS.

To anti-alias, certain parts of an image is selected for pixel additions. It could be just the edges of objects and stuff like that. This means that the GPU is still rendering more pixels, though these are pixels of a lower quality. This anti-aliasing pass can use anything between 2% to 8% of the GPUs power. Depending on the type of aliasing used.

If You have a lot of GPU muscle to spare, then super sampling (SSAA) means that you can render an 8M pixel image and and shrink it to fit on a 2M pixel screen. It would be near impossible to see any aliasing artifacts on such an image.

What Nvidia is offering, is a hardware solution in their GPUs that will take the image that the GPU renders and upscale the image (albeit very very well) to 4x its resolution (assuming you are starting from 1080p). This is what any upscale chip does. they may just not be using as many reference points per pixel.

Contrary to what and how they may want to make it sound, your GPU is just magically now able to render 4x the amount of total pixels for your benefit. Its just a GPU cost free method of AA. It would be equivalent to having something like FXAA but at no cost to the GPU which could let you use that gained power for something else.