Optimized FSR for low power GPUs

[SeppNel]

Expected behavior 预期的功能

This is a more performant implementation of fsr with slightly lower quality, great for low power systems.
The developer claims the EASU pass now takes 1.8ms from 4.7ms (measured running in an iPhone12).
Here is the code. You can read more about it here.

Alternative behavior (optional) 近似的功能（可选）

No response

[hooke007]

I'm sure it is dup years? ago. EASU is fast enough for any desktop GPU, unless you want it running on mobilephone.

edit
Original post: #94 (comment)
Quality regression: #23

[Blinue]

EASU is fast enough for any desktop GPU, unless you want it running on mobilephone.

I totally agree, the performance improvement comes at the cost of image quality. But porting FSR_EASU_Lite is not difficult, and it’s a good thing to give users more choices. Although it’s a bit late, let's provide an “optimized” version of FSR, just like Lossless Scaling.