News Nvidia Reportedly Cancels RTX 4090 Ti, Plans 512-bit Bus Next-Gen Flagship

[blppt]

Real RT will be ludicrously compute-intensive and by extension power-hungry no matter who makes the hardware. If you don't want the hefty compute and power cost, stick to approximated RT.

Even without using RT, the current crop of GPUs use a ton of energy.

 

[InvalidError]

Even without using RT, the current crop of GPUs use a ton of energy.

And there is no miracle solution to that either, all of the math to calculate where triangles land on the screen, the position and direction of each texture that is going to get mapped on each face, all of the shading, lighting and other processing applied on top, etc. all have to happen and will be fundamentally the same no matter who makes it. Given the same fab process, architecture and driver maturity, GPU chip vendors would be mostly interchangeable.